@@jaber5345because the boosting mechanism in ow 2, sometimes when you get lucky , you can rank up really fast. But the dark truth is , if you rank up fast , so is every other low rank players ,they all have some lucky games and boosted in to high ranks. So in reality , you are still playing in a gold/plat lobby even you ranked up to Gm or high masters.
@@jaber5345 bro what? assuming you get no inters ranking up in league is easier than its ever been, why they added emerald rank you think? lol. both ranking system has nothing in common whatsoever
Potentially, but I think the primary motivation is just to keep games a bit higher quality in general in the quickplay environment. If the losing team always quits then nobody gets to play any games at all.
@@Casual314This might be a cynical opinion, but I personally think people are going to troll or throw games regardless, even (especially?) in quick play, so I think punishing leaving unranked games will increase the amount of people throwing a game they don't like instead of just leaving, which might just be the better outcome for all parties lol. Basically, I think trying to punish leaving to improve quick play match quality has the opposite effect.
@@strangejune A single anecdote doesn't count for too much, but I've already had a match where somebody who would've left on the enemy team started afking and throwing, when questioned on them afking they said they would leave if not for the penalty.
@@strangejune so true tho, when im tank and i get so heals or bad dps i can just wonder off as tank and explore the map while i wait to be defeated and try again lol
i love how you mentioned that the system only evaluates your relative skill within the ladder and not your absolute skill. so many people dont understand this,you can absolutely be improving steadily and even make large skill gains but stay at the same rank if other people improved at the same speed or the ladder doesnt reward your skill gain(eg,meta changes or the skill gain is just inefficient in getting you wins)
Absolutely, in fact if you stay at the same rank for years, it probably means you're improving, but only at the average rate of everyone else so you don't see an increase in your visible rank.
Honestly I miss seeing my elo numbers like in OW1 because win or loose once u get those 5 wins seeing my rank go down is still demoralising. Ive gone from P5 to G3 in my most recent games. Im not the greatest player but still id rather know if im going up or down than making it to 5 wins and then derank. Feels bad man
Agreed, I would much prefer a straightforward number that shows your rating going up and down between games. My guess is that Blizzard simply sees that engagement has increased with the current system, so they're not motivated to change it.
It's designed that way to keep you playing. Explanations that try to tell you that the system isn't perfect but it works and that you just need to get better keep feeding into that time sink. Buy the Battle Pass. Unlock that new shiny Hero. Maybe you'll do better and rank up.
I can hardly play 20 ranked games in any given season these days. I really wish they offered quickplay with a ranked ruleset, like how it was in the alpha/beta. Great video - you really put things into perspective.
To me, the frustrating thing is that rank doesn't have anything to do with skill or knowledge of the game. I want to play with and against people of a similar skill. But everyone is all over the place. I've seen Grandmaster players in my diamond lobby losing to the most basic things (like running alone out of spawn after the enemy captured, dying and staggering) where I think that's gold or silver level behaviour. And yes, it happens in the enemy team too. But that doesn't help, I don't feel challenged when the enemy just feeds or are way below my skill level. I often take a look at profiles and see the wildest things. People who start in plat in S1. Don't play S2. In S3 they're masters after 2 games. Then in S4 they get GM5 without a single win. When they have played
It's inevitable with a system based on statistics unfortunately. Randomness means that it's impossible to guarantee the exact quality of a match unless they somehow got real people to handpick players. Even if an account isn't new, as long as they don't played ranked often enough, the system can't really be sure where to place them, so you can get some pretty high variance in behavior. Even if the system does well on average, the only moments you will notice are the outlier events, in your case it would be the system overestimating the skills of certain players by a large margin.
The second paragraph happened to me once. I met a genji player who flamed me for being a “typical rein main” when we lost this control game 1-2. I called him out for dying 15 times (more than anyone on my team), which he replied with basically him trying to carry the team, hence why he’s dead so much. Then he boast about how he’s a diamond player, and a “lowly platinum” player like me should respect him Then I looked into his profile. Dude was diamond 5 in season 6 as a SUPPORT. Funny thing is, he didn’t even played comp on season 6. He was Gold 3 in season 5, yet somehow he got into diamond 5 the next season without playing. He only consistently played Damage in comps and he’s hardstuck silver. Yet he uses being “diamond” as an excuse lmao.
@@lordpeterturbo5216I’m also currently hardstuck high silver/low gold on DPS, and High Diamond on Support, but I also usually have the least deaths on my team and close to the highest amount of elims, or the highest. I play a bunch of different characters depending on team comps and maps, but the trends for my performance are still there in my K/D ratios for them. It really makes me wonder if I’m in that limbo of being hardstuck due to the mmr system’s confidence, or if there’s something I’m missing for DPS. My positioning and timing are great on support, and I can get solo picks or coordinate kills on DPS, so it’s just like “What gives?”
@@kommandantkillcode Is your winrate consistently much higher than 50%? If not, then you have to think maybe you're missing something. If you jumped into gold lobbies as a high dia support, you'd definitely have a high winrate. Same would apply if you were actually higher than gold/silver skill at DPS.
Everything you said in this video is true. The one issue i have with ranked is not being able to see your teammates ranks. i feel like adding private profiles, removing ranks from borders, and the removal of sr was not to stop toxicity but to allow for quicker matchmaking. I started checking in on my teammates profiles and of course most peoples profiles are private but for the few that are public im seeing a mix of all ranks in every match. Im in masters and theres everything from bronze to gm in every match. This wasnt possible in ow1 because your rank was displayed on your border and visible to everyone that everyone was within the same rank more/less.
Could be possible. They probably keep track of stats that show how player engagement correlates with queue times. When the queue times are say longer than 5 minutes, maybe people just don't play the game as often. Being able to select from a wider skill range means exponentially faster queues, and I would guess casual players prefer that. Not saying it's good or bad, but that's probably how Blizzard sees it. The good things about Elo/MMR is that it already takes into account skill difference to calculate your score, so even if the quality of games are bad, you can still climb.
@@Casual314 I think that that's were most of the frustration with ranked comes from. In ow1 you had less impact in 6v6, but in ow2 it feels like 'among us' with "which players gonna be the bronze". It's climable but the quality of matches just feels so bad sometimes. Although I'm sure there are matches where I'm the "bronze player" of the team and it probably evens out.
OW's rating system was closest to Elo in season 1 of the original six-player OW. Elo's system by the way is not effective at team games, *unless the team is the same group of players* for all games. This is well understood in the mathematics community and there are other systems out there that *do* work for pickup team skill metrics. It was after players complained in S2-3 that it "took too long to rank up" that Blizzard added skill metrics to the rating system, giving some players bonus points if the "played well" during the match. Scott Mercer explained this himself in the forums many times. It was at this point that players began to bounce wildly as much as 2000 points up and down between seasons. It *is* broken. The next change was to hide your adjustment when it's negative. You get to see your SR go up when you win a game, but if you lost, you'd just be humbly presented with your new SR and have no idea how many points you'd just lost unless you'd written it down. Then Blizzard started hiding your SR between seasons and calling it "placement," although nothing was different. Your SR goes up and down after every game and you get to find out what it is after playing so many games. This also lets Blizzard manually shuffle everyone around arbitrarily to make up for failures in the matchmaker. This has also been explained by Scott Mercer himself. All of the changes since have been purely cosmetic. Your skill used to be a decimal number between 0 and 100. Then it got changed to an integer between 0 and 4000. Then they made it 0-5000. Now increments of 100 are named starting at B5, B4, B3... and so on; and the last two digits are used to tell you "you're ranked higher than __% of players". *It's the same exact system* we've always had. And you hit it right on the head when you said it's not transparent. If players knew what to do to rank up they'd find the game more satisfying to play. But they don't and that's just how Blizzard likes it. You can find a hundred "how it actually works" threads in their own forums and each one is filled with dissent.
tbh it feels like if you lose at least 2 games, the chances of you ranking up goes down the drain. i can go 5-2 and drop in rank % and drop tiers completely if i have more losses. I tend to straight up win 4 games in a row and then get hit with a loss streak to ensure that i dont climb.
In QP at least, it often feels like the system gives you forced losses, for example you have a plat tank on your team and the enemy team has a GM tank. When I felt like i was getting a streak of forced losses, I would also get games where my team had 4 players for the entire game. What actually seems to matter is the hidden MMR system, but it would seem like in OW2 they are willing to sacrifice accuracy for match creation time.
Overwatch also has the problem of so many different heroes being not only different to play but also being good at different things that it's quite difficult to accurately guage someone's skill based on that. Other games like Counter-Strike have a much easier time assessing your skill because every player has the exact same options in terms of gameplay as every other player and also there are just certain guns that are a staple of certain stages of the game. If you can handle the AK/M4, the AWP, the Deagle and the SMGs quite well then you are golden for pretty much any stage of the match, be it buy rounds, eco rounds or pistol rounds. Your team is not handicapped because it has two amazing AWPers who have to compete for the same spot. But on Overwatch if you have 2 Genji OTPs as your DPS then you have a problem. One of them will be pissy about not being able to Gengu and has to play a hero they can't perfom at the same rank as they are matched into.
Your perspective on the topic is spot on. It's true that many players feel trapped in "elo hell" because they're consistently matched with players of similar skill. It's not about a few bad matches; it's about recognizing where one's skill truly lies and working to improve from there. Well said!
But people arent. I'm gold, and I get a diamond player in yj games ALL THE TIME. Does that mean I'm diamond? Hell no. It means the diamond player has played the system to be ranked lower. It has happened all the time in OW.
Until you go 72 and 12 in a game. And start to wonder wtf is match making doing. Elo hell does exist. I had to beg my dps teammates to switch off their picks to become useful after going 2 and 2 for the first 5 mins. While the rest of the team is around 15+ elims and 5+ deaths.... Elo hell is playing with a team where you're doing your best and your teammates just wanna have fun on widow or hanzo while the enemy team is rolling you.
@@mmaenjoyer1the problem is that the matchmaking is so inconsistent that often whether you win or lose is entirely up to chance. having played there is genuinely no difference between bronze silver and gold. as a gold tank player I have gone against other tanks that are honestly so bad they should be bronze 5 while at the same time the few times I've played as a dps in bronze I've gotten teammates who are way better than the some of the ones I get in gold. this would be a lot less of a problem if your rank was based on game performance, but it's purely off of win/loss rate. The problem is that in OW2 they changed the matchmaking to be less accurate for the sake of faster queue times but didn't adjust the ranking system to account for the less accurate matchmaking. I feel like they should have never changed it, it was perfectly fine in ow1 wasn't perfect but it at least worked.
This proves you are a better tank than DPS. I need you in my team. I play in low diamonds and most matches are just steamrolls by your team or by enemy team. Sometimes I feel may be it's because of me and I don't belong here :).
If performance-based MMR doesn’t exist, then why can 2 players create brand new accounts and play every single game together (even QP before reaching Comp) on the same role (e.g. 2 Supports or 2 DPS) and still end up at different ranks after completing their placements? The more skilled player will always be placed higher, so there must be some other way that the game is tracking individual performance, even if they claim that performance-based MMR doesn’t exist. And this is something that has been proven many times by many streamers who have done Unranked to GM in duos on the same roles since the launch of OW2. Any idea why this is happening if performance-based MMR is really not a thing?
Quickplay games and probably performance are used to estimate what initial MMR you should be at. This is how unranked to GMs are done so fast. When I was said that performance doesn't affect MMR, I was not talking about this, I was talking about ranked games themselves. Your performance in ranked games do not determine how much MMR you gain. You may ask, if they can estimate performance in Quickplay why don't they use it in Ranked? Because it's not a good estimator, it's only used really to prevent smurfing, and cannot be used to make statistically confident claims about someone's skill level.
Do these players play from the new IP adresses and PCs? The assumption of the new account of the old player for the familiar PC configuration is very natural and can be confirmed by some in-game signatures. It is not that they surely do this but this absolutely could be done.
@@Casual314 Hmm I'm not sure if only QP takes into account individual performance, I tend to believe that Ranked might do so too even though there is no definitive proof. Because I've seen Unranked to GM duo streams where people buy Bronze 5 accounts with near 0% winrates and KDR and play in a duo in the same role, and yet the better player always climbs astronomically faster than the weaker player. In that scenario you would think that they should climb at the same rate if both accounts start at near-equally bad MMRs to begin with. And even if one of the throwaway accounts has a *slightly* higher MMR than the other, the better player shouldn't be climbing 2X faster than the weaker player if they're playing all their games together on the same role. Of course, it may not necessarily be performance-based MMR that is causing this, there could be some other factors that we might not have taken into consideration or even thought of, and perhaps there is some other explanation and hidden calculation that Blizzard simply isn't telling us.
@@AIIMight They've confirmed performance in matches does not affect mmr at all anymore, in competitive games. And even in ow1 it was only for ranks below Diamond. For new accounts now, who havent played comp at all, their QP performance does affect the initial seed though. There it might be "performance based" even to some degree just for the "seeding" unranked (or bronze) to gm duo stream sounds super pointless, though. And also, if talking only about QP - your "qp mmr" is not affected by performance either.
@@AIIMight Blizzard says that there is no performance based MMR but who is to say that is actually true? In fact with the example you gave with streamer duos going bronze 5 to GM with bought accounts and climbing at different rates despite playing every comp game together, I don't think there is any explanation other than performance based MMR existing.
For certain roles- tank especially- MMR should absolutely take into account game stats. If you’re performing well, but your team doesn’t take advantage of your good performance, it shouldn’t be on you. Also, if someone leaves the game, that game should not negatively affect MMR for those who stay.
Leavers actually benefit you in the long run. There are 5 potential leaves on the enemy team, but only 4 on your team, so statistically leavers make it easier to climb. It's understandably frustrating, but the reason they don't cancel games with leavers is to prevent people from asking their bad teammate to leave, asking their friend to leave to help them boost their MMR, or wanting to waste the enemy teams time by leaving at the last second. If they cancelled games with leavers, it would just result in frustration when the enemy team leaves rather than your own.
Absolutely not (stats not the leaver), the losing team is always going to have poor stats. Did the tank die off cooldown because they were bad or because their supports refused to heal them? Your output on tank is currently limited by your supports.
Good explanation. If accurate, it highlights a problem we experience a lot in support roles: you talk about the 'stickiness' of MMR, how as you play lots of games it becomes harder to rank up or down because of the averaging effect (the sort of Bayesian 'prior modifier') that is slow to update. The problem with that model is that when you are in a support or even tank role, where it is hard to carry your team, that exacerbates how slow skill improvements are to update, if at all. For example, let's say that your win probability in a support role is only 30% tied to your performance and 70% tied to your team performance. That mutes the effect you play on outcomes, so that any skill improvement you have takes a lot longer to be recognized (because the 'noise' of your team's performance anchors you at whatever your current MMR is). So now you're doubling slowing updates. Funny enough that problem could be solved, by the way, if they did something that probably hurts the rest of the experience: add MORE variance to match up MMRs, but only as an individual. For example if you are at MMR 2000, they should occasionally stick you in MMR 2800 games, where all 9 other players are similar MMR except for you. Having say 5-10 games like that would reveal whether your MMR is appropriate. If it is, you'll lose a lot more than you'll win
I stucked forever in gold, shortly before I finally hit Plat I got a big losing streak. They sent me from gold one to silver 4, now I stuck in silver 2. I overperform most of the time, but getting bad teammates, trolls and leavers in like every match. I lost 7 matches in a row, two times someone leaves, one troll who run in all by himself like he is a tracer and other times my dps or tank getting 0 kills but 11 deaths.
I can relate, sometimes the games are indeed frustrating. The other day I had a 7 loss streak as well with 3 leavers on my team. I ended up going 10-9 at the end of the day, but those 7 losses felt very frustrating when it was happening. I think maybe you're losses are maybe affecting your performance. Try your best to stay positive, and not care too much about the win, and I'm sure you'll be able to climb back up and beyond.
I appreciate the video and all, great work, but I am BEGGING that in future the screen isn't cutoff around the sides. It's been putting me on edge seeing the HUD cutoff slightly on both sides xD. That is all thank you.
This video is spot-on about the ranking system. I wish everyone knew this. I probably wouldn't have known this if I didn't have a gm tank duo who's been coaching me since I started playing the game. A few of my biggest problems trying to rank up was just not doing well on a certain character. I really like Ana but I used to tunnel-vision and focus mainly on healing tank, so I missed healing my other teammates. I switched to Kiriko and I won more. It ended up fixing my tunnel-visioning problem so I did get better on Ana but if I had just played Kiriko instead of all the Ana games my rank would have been higher (for reference I fluctuated from initial placement of gold to plat, back down in gold even lower than my initial placement, and now I'm diamond). I'm now experiencing this on dps. I love playing Ashe and Widowmaker but my aim is extremely inconsistent. I can carry plat lobbies when it's good and be mid in gold when it's bad. I initially placed plat 3 and went on an epic loss streak to gold. A few days ago I figured out Soldier is my most consistent dps and went on an 8 game win streak. I play a lot in diamond lobbies and unfortunately get placed in masters lobbies sometimes. I know I "belong" in plat at the very least, but my character and lobby choices don't help me out. I enjoy playing high-skill heroes but I'm not at a point where I can always carry on them. I'm also learning the most difficult character in the game (Tracer) rn, so it's a rough experience. Deranking feels bad, but for people who are dedicated to improving (i.e., not only caring about rank), it's actually a symptom of improving. I'm challenging myself by playing in higher lobbies. It's frustrating when I derank but I'm getting better at the game by playing against higher-skilled opponents. Before my duo reached gm, he got carried by his smurf duo from either diamond or masters to gm and got shit on. He went on a huge loss streak to diamond but he had improved immensely by playing against gm players. He climbed back to gm and actually belongs this time. The best advice I can give to people who are unhappy with their rank is to stop caring!! If you keep improving at the game, you will eventually rank up, so who cares what your rank currently is? Treat every game like a learning experience. Every time you die, ask yourself "What could I have done to not die here?" or "How could I have played that differently?". Be excited to play against high ELO players in qp. You'll find it very difficult to get better at the game if you hang out in low ELO queuing with your bronze friends. Also, I don't know if this is a common experience, but I hyperfixate on a specific role. Tank was my first love. I actually started playing this game because of Wrecking Ball (I started playing earlier this year so idk what ow1 looked like). I initially placed gold and just climbed through plat. Finished plat 5 my first season playing the game. Didn't even place on other roles. Climbed to plat 3 the next season and then started disliking the role because of the Orisa meta. I turned my attention to support and then more recently to dps. I came back to tank last season and assumed I was still stuck at that rank, but to my surprise, I was carrying all my plat games. My game knowledge had improved simply by trying to get good at other roles. I easily climbed to diamond. At that point, my support was gold, so I focused on getting better at that role and eventually got stuck. I turned my attention to dps and I just ranked my support to diamond. It's a cycle of improvement. I think I'm stuck and then I rank up on some other role. I think it's a great way to improve. Keeps you from being hardstuck in a singular role.
As a Mathematician my biggest gripe about the system is the lack of transparency to how they adjust the ELO based on the 5 person team. Is it an average, is it counter role, is it a mix of the two...these things really affect the outcomes a ton and it "feels" like quite often one really bad, or really good, player can absolutely make a game unwinnable. And that can then skew the results and over time end up soft locking, a player into more and more of those match-ups. Add in a liberal sprinkling of active smurfs, and the fact that it seems that there's some sort of match-up multiplier for people who play in groups (a multiplier that seems to be very heavy handed) and the game can become really un-fun really fast. I don't really care where I am on the ladder. But I DO care about having games that are fun and competitive and I can easily play 10+ games that are just so lopsided (either way) that it's not even fun. And yeah, I acknowledge I'm probably an outlier who doesn't enjoy stomping my opposition any more than being stomped.
One thing to remember is that within every rank there is a level of competency from the worse 1% to the top 99% with the botton 1% being much worse than the top 99%. This is extremely noticeable in the very low ranks and very high ranks as variation of skill sets such as mechanical skill can vary wildy.
In the lower ranks, players also behave very "randomly" in their performance. They play poorly but not in the same way every game (because they don't really know what they're doing) and so you can get very low quality games in low elo due to the matchmaker not being able to predict accurately the skill of lower elo players.
I have a friend who just climbed from Bronze to Silver after many months of just starting the game. He's an insane Widow shot. Just fast and immaculate. But! Has the game sense of a Magikarp. If the enemy team has a Sombra or good Reaper or anyone who can hide their approach, he's screwed. I've had to aggressively coach him to think about more than his next headshot target.
@ericaltman4087 I had randomly a game last night in comp, and it was bizarre. Like 3 out of 5 on the other team were making such bizarre decisions such as the bastion firing on their own team while sitting near us, their genji trying to fight our tank with just the sword dash and reflect(none of the projectiles or jumping unless by accident it seemed) and their Moira throwing the healing balls at our team. It didn't feel like throwing but rather just putting 3 absolutely brand new players together on a team in game whom haven't played a game at all of overwatch. I have never seen anything like it, even in bronze before.
Losses and wins should count for the same, regardless of rank difference between you and the average rank of the match. This hidden SR should only affect what average rank you play agaisnt. Basically the current system results in players who've had a losing streak have an extremely difficult time to regain their status. It inversely rewards lucky players who've played the game little. More than anything else, all rankings and matchmaking algorithms should be transparent and visible. No hidden "man behind the curtain" bullshit.
Agreed on transparency, but it's just the unfortunately reality of human nature that the more people know about something, the more likely they are to interpret it incorrectly leading to even more confusion, and therefore less engagement. The reason that taking skill difference into account is necessary is because of the natural bell curve distribution. Unless you're right in the middle of the curve, most of your games will be skewed towards either being lower than your skill or higher than your skill depending on which side you're on. In theory, the game is supposed to give you games of similar skill on average, but that becomes less possible towards the ends of the bell curve.
Nicely explained, great video. It doesn't change the fact that their matchmaking system is sadistic. Feels as if the main character was the system itself. Trying to constantly figure out where you are, starts juggling and forgets about the most important thing: player's FUN. I play to learn as I challenge myself through situations that I learn to solve differently. I learn nothing if u drop me in a match with Masters and GM, while I'm a Diamond and u put plats in my team. The game does it and thinks "let's see how u do". Then ofc u lose. And again, and again, and again. And then u end up 5 - 12 and the game says "oohhh...unlucky, u are not good enough, let me put u in a lower level, cause u are not ready". And the same happens if I get free wins. I learn nothing by stomping the other team, and again I have 0 fun. Absolutely no reward. What's the point of having such an overworked system if the game has the most random pairings, resulting in matches that, far from fun, are fucking cruel. I have never seen a game that, no matter how incredibly hard u try, crushes you so bad. As I said, sadistic.
It's the balancing act of trying to decide what to compromise on. In this case it would be queue times vs match quality. Higher quality matches = very high queue times, and most people aren't willing to wait 10 minutes for a 10 minute game.
This is why I don't mind Splstoon or Apex's ranking systems. You win? Go up. In splat? Play a series, gre more points based on how good you won. Let the rest tbe players fight for their spots, but let the way points are awarded be equal and consistent across players. How much you need is based on current rank, but you get no less or more then others. How high you are shows plainly how better you are. It's a more transparent system.
One thing with the Splatoon system, and the Apex system is that they tend to be top heavy I believe. Meaning that more players are at the top than the bottom. This is undesirable if the purpose of the system is to be purely an indicator of skill, rather than simply a progression system.
@@Casual314 Precisely. It really makes you wonder why Blizz feels the need to make ranked another progression system, when the game has Like 3 other progression systems now to keep the players carrot and sticked. They don't need ranked to be anything other than a skill placement. They're telling me they can't fix ranked sooner or go back to a displayed mmr system, when they already have other progression systems to keep players occupied... BP gross as it is, challenges, hero mastery, cosmetics to a lesser extent. Didn't feel the need to mention since it I just wanted to talk about how much I like apex and splat's better and describe it, but yeah, what they have now is purposeless and needlessly opaque. And they're not even fixing it till like... middle of next year or some shit? Very dumb.
A lot of people in OW1 got carried OW2 system makes you carry and play your own games to where you rank your self up through performance skill, and wins
Thank you for the video, now i know exactly what had happened to me after playing ranked for he first time. I got ranked to silver, then to gold and then i had a big loosng streak and went all the way to bronze 1 and felt like crying, but after improving i am now at gold with support (70% wr with ana) and i'm climbing, since i actually got more skilled, can play at a more consistend level and i brought my mmr up through winning games by improving on my playstyle. I'll ge so happy to start playing again once my unrightful ban is lifed (i got banned for telling my bf to shut up for teasing me in all chat when he was specating me bruhh)
Sorry to hear about the ban. Yeah you have to be careful with the automated chat monitoring, since it can't detect sarcasm or jokes. But I'm glad to hear you're motivated to start ranking up. Best of luck.
@@Casual314 thank you so much for the response to my very sleepy written comment, didnt expect that. Yup i realised that, sadly the moderators didnt even care about my explanation when i send a ticket in, but thats not the topic. But maybe if you have time you could make a video on their ban system aswell ^^ Thank you! :)
Some of the things you mentioned are the exact reasons why I don't like the change in the comp matchmaking. Before, you could play against higher or lower ranks, and you could "prove yourself" better against the Elo. But now, I only get matched against players of my exact rank, so the Elo is way slower to readjust.
This is the opposite problem I hear from others, most people complain that the range of skill in matchmaking is "too high". With the system right now, you'll still get into games that are above and below your skill level.
I quit competitive a few seasons ago, there no point in climbing when a system doesn't let you climb, no matter if you're on winning streaks or not. Until they actually fix the rank system as they are planning to at this year's Blizcon, I want nothing to do with the mode.
My Qp games are insanely sweaty with grandmaster ranked players. I don’t play competitive but maybe I should, I could be a lot higher rank than I thought if QP is keeping track.
Potentially yes. Just know that the game can get very frustrating once you're keeping track so try to stay positive as best you can and the experience will be a lot better.
Great Video just a couple of things worth noting specifically for Overwatch 2. Win/Loss Streaks have been mentioned that they will or already modify MMR change it was mentioned for the S9 Comp 3.0 explanation at Blizzcon. Also, every old account going into Overwatch 2 kept the OG Overwatch 1 MMR with decay based on the number of seasons since a consistent competitive experience (mentioned in one of the Comp dev blogs). (This was because in a closed beta they saw a marked uptick in stompy games from Diamond+ players (OW1) in games against players they shouldnt) This also duplicates and extraploates the Old Account vs New Account issues, because new accounts have bonuses to stop Smurfing as well as New Player MMR flexibility. Also another thing Blizzard have tuned the MatchMaker to find people with equivalent-ish MMR in the same role and place them between teams. I.e. your Tank has a Gold Badge but Diamond levelled MMR & Your opponents Tank is Diamond badge and MMR. This means that the teams average is usually less than 3 minor Divisions of MMR (An action of this is to normalise MMR change between rankd cause the teams are eqivalent skill on paper)
i think it's possible to model every type of value in overwatch with the right stats / weights and also not be abusable. i also think this would lead to the best matchmaking / ranking system possible for any team game for multiple reasons. one reason is that it unties your rank from other people (that tie is the source of most frustration and toxicity in competitive games). if you play extremely well in a game you lose, you should not lose rating; the opposite is true if you play badly in a game you win. another reason is that truly good stats would help people learn / improve at the game and focus on their own play. a player should be able to target specific stat improvements to rank up, and if done well, improving a stat would always equal more value (obviously not the stats we have now). if it were up to me, i'd have a person's rank literally be their place on a leaderboard of all players, where your place is determined by all your stats in different ways
In theory yes, but it's unlikely that a human would be the one to figure it out. With extremely high computation power, and some kind of performance tracking AI that has been trained on billions of games, you could theoretically have a system that can accurately correlate how certain players play with their likelihood of winning. You would of course have to track and calculate the game to every last detail, and the game would have to stop getting frequent updates because every update would change the meta making the AI predictions inaccurate. Given that Blizzard already has trouble creating a good POTG system, I wouldn't trust them to create a system that could somehow track and rank every moment of the game accurately.
Something to think about is a lot of those stats don't matter at all. Your ability to manipulate the enemy team into losing is the most important skill after all. An example is how I sometimes play tracer. We could be near the checkpoint but the enemy can touch and reasonable win the battle. But I start diving alone when they leave spawn and trick them into chasing me. I waste so much of their time they forget about the payload and we cap. You'd be surprised that this works even in masters for me. What are my stats while doing this? Very little damage, zero elims, and I may even die. What was the value of this play? Immeasurable, considering my team captured the objective. When you think about the game in this way, climbing becomes easy and you're not worried about the ranking system anymore.
I'd like to see them do the same system paladins (and I think League? Not sure) did, where you get the different tiers (diamond I->V for example) but to get to another tier u have to reach a certain number of point (on paladins it's 100). You see a number go up and down so you know if you're making progress within your tier while also being confident you're IN that tier. You can also introduce a mercy system where if you're at 0 point you still need a few more losses to go down a tier to not discourage people.
It kind of works in both directions though, if it's harder to go down, then it'll also be harder to go up. The league system is built more to forced you to grind which personally I'm not a fan of. I'd be fine with LP, but in some ways it just adds a different kind of confusion.
The League system is horrible. It's more about grinding and keeping you playing than actually rewarding skill and performance. I hope OW never gets a system like that. At least in OW I can reach GM through individual performance.
Great explanation. Good humor throw in as well. Definitely gonna send this one to my nephews... I tried explaining the first point to them, but they didnt believe me.
This was actually a great explanation of the way the statistical side works! It can be hard to discuss these things in a way the general public can understand but I think you did a phenomenal job breaking it down and building up to comprehensive package.
Still looking for explanation of: literally performing the best in the whole lobby with highest elims, damage, healing (yes im a moira main) Yet STILL HARDSTUCK. I’m in Gold and almost every game I’m watching my team spawn in, then one by one running straight into the enemy team 1v5 and dying. Rinse and repeat. Im watching DPS (when they actually hit shots) focusing enemy tank that is being double pocketed by their supports. I was Plat (occasionally touched Diamond - PS5 fwiw) in OW1 and had to quit because of these frustrations. Recently picked up OW2 and I’m experiencing the exact same thing. It’s like, the better you perform in a game, the worse players it puts you with the next game. So playing well becomes a vicious cycle of frustration. Im trying my hardest to focus on my own gameplay and not blame my teammates (that top 500’s keep reminding us to do) but come on….
Smurf detection is ironically responsible for a lot of the matchmaking woes. I've been playing with a friend recently who's new to OW (still working on grouping as five), and he's got an inflated QP winrate because I carry so many games. Since he hit 50 wins, all we've been getting is Master/GM lobbies, and based on other folks' testimony, I'm certain that's the smurf detection pushing his MMR way, way higher than it should be.
The game pretty much only counts wins/losses so if you play with your friend, and you guys win a lot of games together that might be something that happens. Overtime, it will correct itself.
Thank you, this was so enlightening. I feel in your videos you talk to a big crowd of people out there who are seldom addressed by the pro streamers and content creators. OW is my first FPS (away from some Q3 arena and Half Life 2,and that also tells you something about how old I am 😂). I learned a lot since I started playing and while I still mage a humongous amount of mistakes I feel like I shouldn't be b5 anymore (but the who does right?). I'm super hard stuck even with positive win rates for a while now, but considering how seldom I play compared to the average player, I now have a clear understanding of why I can't even get to a higher bronze rank even if I should really deserve it. With tank and dps I at least can play a bit of silver though I think I'm worse than on support. But I might be doomed there 😁.
Yeah part of the ranked experience is the grind, which is why there's no reason to feel bad about your rank if you don't have the time to play hundreds of games every month. It's hard to improve without a lot of time investment. No one's paying you to play Overwatch, so my philosophy is just enjoy the game in your rank. There's nothing "more fun" in the higher ranks anyway. Sure some people on the internet might make fun of people in Bronze 5, but are they having fun playing their own game? Probably not, so who's actually winning here?
Dropped from GM2 to M5. Currently at M3. This gave me the motivation to just play more ranked, cause if I do a good job over the course of like 200 hundred games. I will most likely climb. Thank you :)
Trying to model skill leaves out a bunch of factors that people don't tend to count, marks and call outs, pushing enemies back/off of positions but not eliminating them, occupying enemy attention away from the objective which may result in being killed without an elimination but significant capture time for the team etc
Overwatch is the only game where it makes you feel bad for winning. Like even if they don't show MMR anymore, at least have some movement in between the ranks (e.g. you go silver 4 to 2 or 3 depending on those 5 games). I am new to FPS on the computer and I am new to overwatch, so of course I was bad. And I played bad for a while because mouse and keyboard was not natural. It's taking forever to rank up now, where I have been consistently getting more wins than losses (even if it's 5-4). I've done 15 games with no change. Staying the same for me is worse than having a number go up and down that changes each match, especially when it feels like there's forward progress. It just puts fear that the next games will drop me lower. I've had one time where a bad loss streak didn't drop rank. I've had many win streaks where I've dropped rank. I've had countless 5-0s and 5-2s that kept me the same rank.
They basically try to do that with the percentages that they show when you get your rank update. If you've been playing at the same rank for a while, the percentage will be the best indicator of progress, although it's not perfect.
I really enjoyed listening to your voice- it was really soothing & I appreciate the explanation. Although it won’t stop me from thinking the universe is against me when it comes to choosing my teammates- at least I can look at it from a different lens. I have liked & subscribed! 😎
Still in the progress of trying to figure out how it works (through unkranked to GMs + bronze to GMs) and all I can say is it's VERY stupid. I miss SR more with each passing day :(
I still have a gripe with this system in that I believe incentive of other players (teammates/enemies) and the quick play experience can skew the data very poorly for the user. Quite a few free to play players just don't care about their MMR, they get placed wherever and do not care where they end up. Mostly though, QP is a terrible grounds to base skill upon, you can have top 500's playing against new accounts and the new accounts win against them because no one tries in QP, or they can just be screwing around for fun. Now this new account has a wacked out skill value that places them way too high when they go to comp (I have witnessed this as I have been climbing out of gold [now masters] since I was a literal child and now I can drink). So now these new accounts aren't having fun because they're getting toasted in comp, and then they artificially tank the MMR of every single person they get into a lobby with. This is a massive issue I have viewed countless dozens of times firsthand. Diamond seems to be the worst spot for this. If I had to conservatively guess, one half of all my games climbing out of diamond had at least one brand new account that underperformed either on my team or the enemy team. On top of wacking out the MMR system, it doesn't feel good to win or lose when there's obviously someone being dragged along for the ride. Just as is with a leaver, it is nearly impossible to win when 20% of your team is dead weight. I also understand that statistically over time, these accounts will reach equilibrium between enemy team and my own, but it still does not feel good at all to play like that, and I also have to already work against the MMR system that is statistically confident in my metal rank OW1 days as a teen. Having one on my own team feels worse than beating one on the enemy team, because the statistically confident game likely takes more MMR from a loss than it would give from a win in this scenario.
Yeah I can understand that the way the system works can lead to some very bad games. It could have something to do with the QP estimation they implemented, but I can't be sure about it.
After playing many games i've come to realise that it works through a really frustrating system of "Winner" and "Losers" qeues. Once yo start winning you can get into a 5 winning streak and the system places ypu with good teammates that are climbing against those who are deranking from above. As long as you do not lose the game by yourself because you are no longer able to keep pace, you keep winning. And then it does the opposite. It places you in the "deranking queue" where you will be forced to carry your horrible teammates who are also on a losing streak until you derank so low that you can actually carry the game. Surprisingly, as long as you perform well in these "deranking" games, you dont lose that much SR but it is really tilting and frustrating to see how over and over again you are the best or second best player in your team but carrying that mess of a team is just impossible. When you look it up online people just say" git gud bro" and they are right but 50% of the games are already sealed before they start and thats creates a terrible game experience. Edit: I know it because when I'm on a losing streak I change to my alternative account and suddenly the system is friendly again.
I'm skeptical of the scientific validity of that kind of experiment. Blizzard has already said on their official blog that they don't do that. If I have two coins and I flip one and I go on a "head-streak", so I flip the other one and coincidentally I get tails, does that prove that the coins are rigged? Now, not knowing doesn't prove that it's not rigged, but I'm not sure the evidence is strong enough to draw a conclusion either way.
I’ve had a similar experience. The first 3 games are always very easy, I just need to not throw. I don’t think I’ve ever lost my first game in ranked in a session. Once I hit that 4th game mark, it gets much harder. My theory is that your mmr goes up from those wins and now you’re expected to carry teammates with a lower mmr, even if ranks are relatively the same. So your enemies get stronger and you’ll get some weaker teammates.
@@Casual314I mean... Bringing up "blizzard said" to prove a point is not exactly the move... Blizzard said many things which were not true. Blizzard once said that Ashe NOT getting ult charge while Bob is active was a bug. Fast forward 3 years and they said that it was not a bug, and that she shouldn't get ult charge during the time. So I don't see why they can deceive people in terms of their game balancing decisions, but not in terms of their mmr balancing.
@@129iphone Cynically, I think it might have to do with he way the rank system works, which means they want you play more to see you rank up/down. Yesterday I won 4 games in a row and then lost 4 in a row before finally winning another. I still ranked up, but it felt like a waste of time and doesn't make a ton of sense given I have a low WR%
I've clawed my way from silver to platinum twice, in overwatch 1 and overwatch 2. I've also clawed my way from silver to high gold multiple times. I am not sure what kind of "certainty" the system has but if im good enough to do this multiple times, I think the system has failed to place me correctly. In all honesty a more pure elo system would feel better and be less of a time waster than this "hidden mmr" garbage, statistics and machine interpretation should never see someone with a %60 WR and do anything but push them up. it feels so extreme that nothing short of just hopping into kovaks for an absurd amount of hours until I am grandmaster level aim on a hero with a one shot and able to basically "smurf in my elo" would get me out.
Good video, the main frustration I have with the system is the combination of needing hundreds of games to balance out "everything that could happen" and the fact that the more games you play the more confidence the system has in your ability. I guess id add the fact that there's also zero individual performance metrics. All this culminates in me in a place where I know I am higher skilled that the lobbies I'm in and I AM climbing but its so slow because I had a rough start with leavers/throwers/people who look like they've never played an fps before. (Gold2 initially to Plat2 took me 150 games over 2 seasons) I consistently have a better stats than my team win or lose and I thought the system would recognize that and move me up faster, learning that it does not is kind of annoying. /rant
This is so true, I always try to tell people who complain about their rank all these things and they never understand. This is just intuitive, there are tons of bias and statistics in place. You are statistically not the unluckiest person on the planet
They used to have individual performance back in the day, and I miss it. It incentivised paying more attention to your own performance, and gaining +30 on wins and -10 on losses communicated very clearly "keep it up and you will rank up". That W/L is the main metric means people will be so toxic when they think a teammate is bad, and subsequently throw their own games as a result. This feeling those players get is bad for all players in ranked when in colors the perception of super fun losses that were tight games. IMO the solution should be to focus on sub-objectives like "2-3" on a flashpoint map probably meant it overall was even, and if those games don't feel good to lose, what hope do we have of immature players ever taking responsibility for their actions? The frequent draws didn't feel great on 2cp in OW1, but at least there was this feeling of it always being a possibility preventing teammates from tilting. So I'm not sure it helps that draws are so rare now. I'd rather take a 30 min draw over a 15 min loss for competitive integrity. A single 5-6 payload game should be counted as 5 wins 6 losses just imo.
The bad thing about this statistic approach is that you may need an enormous quantity of games, and some players may leave the game by pure frustration by then, because first of all a game needs to be entertaining and fun. Not being able to win 50% or winning a lot and not ranking up is not gonna work in keeping players involved in an already not as played game.
Yep that's a valid point, and some games take that into consideration, and make their ranked system lean towards ranking up so that players will continuing climbing as long as they keep playing. Then you just reset the system every once and a while so that not everyone is at the top rank. It's more of a design decision between feeling of progression or true competitiveness. Neither is necessarily better inherently, it just depends on what you're going for.
This was actually a really nice video. I like being able to think things over and understand a situation better so I can possibly help it. Especially if it is case or time-sensitive. There's a lot of factors that go into wins/losses and sometimes that can be wholly in or out of our control. You can do amazing and still lose on one small error that was made by anyone else on the team, including yourself. Accountability is important but so is making sure you don't go nuts trying to psycho examine things out of your control. There are going to be bad games. There are going to be good games. The important thing regardless though is that they are just games! Have fun, take it easy, and feel a fire when you get competitive that instead of burning out at the first sign of trouble, gets brighter and bigger for the want to win and commit to the passion even more. If your games ARE genuinely stressing you out and it's unhealthy, step back and try something new! You can always return later when you feel more patient.
Big fan of this mentality. If you're motivated to improve even when you're losing, it can go a long way. I think it's perfectly ok to say that having a bad team sometimes makes it hard to win certain games. The important thing is to take it easy and make sure you're actually enjoying the things you spend time doing.
Ah, this completely makes sense, so I had an account which I connected to a friend's PlayStation account when looking after their house and got told I couldn't reassociate my account to my PlayStation until a year later. On the account which I can't use, I was stuck in (no joke) bronze 5 as support. I create a new account to keep playing and that gets placed in silver 1... Definitely makes sense that Blizzard has something behind the scenes trying to keep accounts in the same rank then.
It doesn't help that it feels the matchmaking is really broken right now. I love OW and do enjoy comp but I feel like it's never consistent and I'll practice and bust ass only to rank down randomly. I get why the system is how it is but it feels like a lot of rage would ease up if Blizzard fixed matchmaking.
It's better to just get a number like in Chess. One problem (aside from transparency) is they have assigned bronze too high. e.g. in chess: 600 is a new player 1000 is a player who can play but isn't good 1400 is an ok player 1700 is a decent club player 2100 boarderline pro 2500 GM 2800 super GM In OW2: Bronze is a new player Bronze is a player who can play but isn't good Bronze is an ok player Silver and Gold are decent players Plat and Diamond are very good Master and GM are incredibly good, not dissimilar to chess I've seen B5 players who clearly understand their role, and can crush unranked lobbies, I've also seen B5 players who looked like it was their first time playing and didn't have a clue. There needs to be more separation in the lower ranks so that players of a lower skill level can still feel like they are competing and getting some progression.
I agree, it's the reason that League of Legends added the Iron rank. I think Blizzard knows that Bronze 5 is too big, but they're hesitant to add an Iron rank because it serves to make players feel even worse if they're sensitive about their rank.
Bro I had the exact problem earlier in the season I Litteraly thought s4 was broken. I am a g1 dpsin console and just switched to pc just to be stuck in s4 with no movement. I went 5-0 and nothing I left no games or anything.
@@Jamilbaker12I've been consistently getting an 80% winrate on DPS. Yet, been stuck on S4 as well. I've seen videos of the top dogs of OW analysing Bronze and Silver lobbies and I'll tell you - there's no way I'm at the same rank as these people.
The individual competitive updates don't matter, it's every single game you've every played that matters. Imagine someone has played 10,000 games at Silver 3 at 50% win rate. Even if they go on a relatively big loss streak, it's probably just some luck.
Me and my fiancé play overwatch together. We play competitive, and I have a higher win rate than him. I also play on my own, he usually only plays with me. I play all different game types, and have a positive win ratio on all of them, along with a higher win ratio to my fiancé. I am currently at Silver 1. That’s currently my top this season. Wanna know where he got to? Fucking platinum. He managed to rank all the way up to platinum while I’m still stuck in silver, despite every single part of my account would tell you that I should be higher. I have more time on different heroes, higher progression level, higher win ratio with more games to make it accurate. All of my statistics are higher than his. We make a joke that it’s because blizzard is sexist, but I honestly think it’s because I main healers. For some reason, I can’t climb as a mercy main. This makes me think that they rank you similar to deathmatch, on your K/D…and don’t really pay much attention to healing or assists. I maybe get a max of 5 kills in a game, because it’s not my job as a mercy to frontline damage. However, it doesn’t matter how many people I save, resurrect, assist…it feels like the matchmaking system doesn’t look at that…or more that it can’t also put that into consideration. It’s why we don’t get healing POTG, because they don’t know how to judge what’s good healer activity against someone who got a triple kill. It doesn’t matter that you kept them alive which allowed them to do it, that’s not what’s measured. Blizzard has said that it’s not the case, but they already have that type of shit for POTG, and I don’t think that they’re being entirely truthful. I love playing healers, it’s my favourite because I like being fast and helping people rather than going in myself. But it feels like I’m gonna have to start choosing between roles that I like, or ranking up.
The ranking system is flawed, but I don't know exactly how to fix it. In season 6, I decided to play both my accounts. My main, which I've consistently sat around D3-4 (3300ish) since ow1, and my alt account, which is only slightly newer than my main account. I played completely through the battlepass on both in S6. In total, on both accounts, I "won" 200+ support games. I played hundreds of games in S6 across both those accounts. It was a long season and I had no other games to grind as I waited for the autumn releases. My main account ended at D4. My alt account ended at P5. I did duo with a few friends on my alt account, but that shouldn't make the SR discrepency so apparent. These are both accounts that have hundreds of hours on them, both played consistently since 2017. You would think the MMR on them would closer? I played a large number of games on both, playing the same heroes. My alt account actually had a negative win-rate overall for S6. Main account was around 55%. It shouldn't be this tin-foil hat thing to figure out your mmr or theorize why it is the way it is. Blizzard can just change it. It's a pain point for every person who plays ranked and turns people off from even using the mode. The queues in ranked are already long enough without people forfeiting the mode all together.
I don't really care enough about rank. Shiny metal badges are nice, i just want to have some fun with a little bit of pressure. What gets me is when you have a terrible match and someone blames it all on you 'cause stats. When I'm pretty sure they've experienced both terrible and fantastic matches, but never think its them. It will never change, it's humanity. But it does wear me out sometimes. I helped to win a game as Sombra in a cheesy way in a match that was probably Silver - i captured the first point and moved the payload to the end in times where the enemy team was unaware. The first point i had 0 kills and 569 damage. Regardless of what anyone thinks, that was value for the team and had nothing to do with damage, or kills, or deaths. You may have 1,000,000,000,000,000,000 kills but if you forgot the payload and weren't with your team it just doesn't matter.
Yeah it's just something that you have to learn to deal with. Try to remember that most people are quite unreasonable. If you remember that, then just realize that when they say something silly, it's supposed to happen, because they're unreasonable. Don't try to reason with unreasonable people because then you would be doing something silly as well. This is the mindset I use so I don't get frustrated by silly people.
Coin flip you say? The system approaches 50%? That would be accurate if they stopped matching stacks with solo players... Also, the longer you stay in a rank harder it is to leave, so one can not improve and feel the improvement as the system will tell them they are the same. Yep, amazing system that promotes "smurf" / "boosted" accounts.
I think the system should overlook past data, this will allow faster movement through the ranks. Also it should look for the impact you created within the confines of your specific role. A tank cannot be expected in the current meta state to just create the diff needed to win the game or to get 4k every team fight. So even if they lose the game there should be a way to see if they created the right impact but the team was not able to capitalise on it or just the tank didn't get the right amount of help/heals to combat effectively. Same applies to DPS/Support roles as well, if their tank doesn't create space and hold the lines they can get mowed down very easily. Perhaps if we can measure these, we can give people better ranks.
I’m happy with the competitive changes that are coming into season 9, I’ve been playing Overwatch since beta and was never really into competitive like that but more into QP. I always placed gold after my placements but recently hit platinum. Hopefully in season 9 I can hit diamond
From my brief research on the Splatoon ranked system, it seems that the system is skewed towards ranking up, probably to keep players motivated to play. Basically as long as you keep playing, you'll rank up eventually, just at different speeds depending on your skill. This is a valid design choice, but it does lead to the distribution eventually becoming "top heavy" meaning that most players will be at the top ranks. This also means that rank is in some ways more of an indicator or time played, rather than purely skill, which is probably why Blizzard wouldn't go in a similar direction.
MMR said I'm mid masters like 6 months ago. Ranked system said I'm plat/diamond. My account is like 500+ hours aged at that point, and I slingshot backwards even though my performance was fine.
I believe that one of the biggest parts of the video that you didn't address very well is just how bad the displayed rank is. I understand that the point of your video was to explain the innerworkings of the ranked system, and to explain all of the hidden mechanics, but when you talked about the displayed ranked system you so quickly glossed over just how bad the displayed rank system is. Speaking in general terms, even if the hidden mmr of a ranked system was PERFECT, meaning it could accurately predict the exact rank you should be in after every game, but the displayed rank system was terrible (akin to what we have in the game right now), then that ranked system would be a terrible system regardless of the back end being perfect. That's because the player would have no clue what they were doing wrong or right, they would have constantly have feelings of not being at the correct rank, and they would feel very frustrated at the system even though it's technically doing it's job perfectly. All that is to say that the inner workings of the ranked system, feelings of players, feedback given to the player, transparency of your rank, and displayed rank all matter. They're all connected and you can't just pick one piece of the puzzle out of the whole system to analyze it separately when considering the efficacy of the ranked system. It's the same fallacy that Blizzard falls into when making balance decisions: they primarily look at statistics like win rates and pick rates in a vacuum without considering how it feels to play a role, how it feels to play a certain character, whether or not mechanics are fundamentally good for the game, how it feels to play against a certain character, why it feels terrible to play against certain characters, etc. Therefore, I believe that most of the grievances that people have with the ranked system mostly lie in the displayed rank system. You mentioned the positives of the new displayed rank system those being less anxiety about ranked and the feeling of going from 2000 to 2500 feeling like more of an achievement, which I completely agree with. However, the negatives of the new displayed rank system are horrible. The benefits mostly arise from moving from a single number SR system to a badge system, gold 1, gold 2, plat 5, etc. But if you also gave the players a number to reference exactly how close they are to ranking up to the next division, then the transparency, feelings of the players, and feedback to the player would all improve dramatically. You could keep the benefits of the badge system while keeping all of the benefits of the old SR system without the ranked anxiety. The displayed rank system matters way more than you give it credit for in the video. Like I said, I understand that you're trying to explain the innerworkings of MMR, but you have to do a better job of acknowledging the problem. Your video frames ranked system frustrations as simply not understanding statistics and how MMR systems work. However, the problem is way way bigger than people not understanding probability and you should really state that better otherwise people will take away the wrong conclusion from an otherwise great video.
Thanks for watching the video, and adding your thoughts. I understand what you're saying. You are correct that I wanted to mostly stay on topic, which is why I didn't address the perception of the ranked experience as much, especially because how the visible ranked system works could change a lot in the future. I also prefer to keep the descriptive and normative discussions separate, I don't want to add too many of my opinions to a video about something objective, because then it can be seen as dishonest. Which is why all I had to say about the visible ranked system is that it's based on the hidden MMR. Why it's designed that way, and what would be better is a different topic that in my mind would deserve a separate video. My interpretation of your criticism is that I should have taken the opportunity to say that the visible ranked system is bad, and here's how it could be better. But I don't think anything I said would have been particularly surprising to anybody in the online OW community. Since we're here though, I'll just say that my opinion is that the visible system should be as detailed as possible, letting you see the number, how it's updating, what you can gain/lose, and make it visible in your profile. In my experience, most people don't understand how probability and statistics works, which leads to a lot of questions in the form of "why doesn't the ranked system just (insert terrible idea)." They now add a percentage to your ranked update which kind of tells you how close you are to the next rank, but my hope was that if one understands how the system fundamentally works, then it's obvious why the visible system is designed the way it is, regardless of what it looks like. It seems that the no.1 problem with the badge system is that players wonder why they can win games but not see progress, but the same problem existed even with the SR system. Players would wonder why they were losing more SR on losses than they would gain on wins, and then claim that the system was broken just like they do now. But if you understand how it works in the background, then you can probably answer both of those questions quite easily.
12:30 The game does take your stats into account though, even if not explicitly for the purpose of ranking you up or down. If you're one-clipping left and right in a bot lobby, gradually you will be matched against people with higher mechanical skill irrespective of your rank or the age of your account. The only way to disprove this is to have Blizzard display everyone's rank again.
If you're one clipping people consistently you're probably hard smurfing winning most of your games, which means your MMR is going to climb faster and faster which is why you get placed with higher ranks very quickly. Blizzard is not purposefully placing people will similar mechanics together.
I wish the game cared more about playstyle and stats instead of just win loss ratios, I can go 50-6 in a game and lose and the game doesnt care how well i played
Yep I can absolutely relate, and I'd be lying if I say I didn't find moments like those frustrating as well. I just try to remember that some games I get carried when I'm performing poorly as well, and it all balances out in the end.
shit is so frustrating. its where im at rn i can do so good but it just doesn't matter if we lose like all the effort was for nothing and there's 0 metrics to aid me for my performance besides a scoreboard
At the end of the day the W-L based system is a measure of how good you are at carrying games. Number of kills/amount of damage are related but not the same as carrying games. I agree it is very frustrating to lose or derank when you feel you deserve the opposite, but the best course of action to climb is to simply improve even more. If you widen the margin by which you are the best in the lobby, it will become easier to overcome "bad luck" in matchmaking, and as you continually improve, you will continually climb.
The problem with that is, it would be super hard to make the "playstyle" rating fair across all heroes comparatively. With some heroes, like widowmaker you could say its straigthforward, just look at number of picks etc. But then you have so many "weird" heroes and matches. Purely W/L based is the 'fairest' pretty much without a huge development and testing effort, and also try to find all scenarios where it couldnt be abused. iirc the performance based system was abused at times in ow1, as people found out which stats mattered, they began "stat padding" those, irrespective of other things. mercys going for "bad rezzes" for example just to pad their rezzes per 10 min stat, and so on.
Never heard of the term "Doomed account" until now but i had a feeling sometimes like that existed and i just proved it. Just made a new account last season and i threw(in a stack) down to bronze and sat there and marinated in it for a long time. Recently began climbing the account and i could win 19/20 games and i only climbed 1 rank tier at a time. Despite how well i performed. I could have 30, 40, 50 kills with less than 10 deaths and still only go upn1 tier. The account is nearing diamond and i finally got to skip a tier from plat 3 to plat 1. Meanwhile my main account has been hardstuck low master for years. The reason i threw was to prove to myself that i can actually climb to my true rank and i belong there. If im gonna be a hardstuck master player thats fine butni needed to prove it to myself. Next gonna make a new account and tryhard from the start to see if i can get higher thannmy main
I'm convinced it's my teams a lot of the time. My last games either have over extended tanks, Moria players who think they have high healing means they're doing good and players with overall bad positioning. I've gone games were I only die once and my team dies around 10
Well for whatever reason my rank updates are incredibly linear. I get 1/4 of a tier per win or loss +/- 2%ish per game. So for example, if I go 5-0 I will go up one tier + 25% of a tier basically. It might actually be one tier + 28% or something but it's entirely predictable.
Yes the system on average behaves like that, which is what causes people to expect that it will always behave like that. Once you're no longer climbing steadily, you won't get as large of an increase in your ranked updates.
Hi Casual! Interesting video - thank you! Coming from game development myself, could you argue why do you trust Blizzard has implemented the elo-system ”as is” / without modifications (e.g. coming from business guidance, user experience, etc.) ?
15:30 "somehow the universe itself is working against you, putting all the worst players in existence on your team" I know for a fact its doing this. Every game I'm in there is this one guy named Jing who is throwing my game.
Another huge issue is that unlike Apex or Destiny, players are not incentivied to learn more (game sense) or improve their gameplay. If you compared the game sense and knowledge of a bronze Apex player and a bronze OW player and youll see that players simply dont take the time to learn how the game should be played. Thats why there isnt much of a difference in metal ranks. And sadly for those who are trying to learn and practice playing the game properly they still have to feed into bad bahaviors to get a win because the rest of their teams are not playing the heros right. Unless you have amazing mechanical skills to essentially carry your teams you will struggle greatly to climb regardless of how the comp system judges you because a lose is still a lose at the end of the day.
so recently i used my old blizzard to play ow2 to see what rank i get after 50 wins, i usually buy accounts that are done to bypass the 50 games. i played all my 50 matches one tricking genji. my first few comp games were GM2 and GM3, im a genji player, GM5 lasted me 25 games untill i went 5-14 and went down to M3. i was so excited hitting GM, this makes alot of sense @8:05 ok so if i lose 50 matches in GM5 - M4, with my MMR being in GM2-3 elo, wonder what rank ill get after this season ends.
I think they should go back to just showing the number. Maybe it's just because I'm a chess player and am used to seeing it, but seeing it after every match just helped communicate what was going on so much better. In the case of OW I think it's got a number of things going against its ranked system. It feels like a lot of the games are so much more imbalanced. I've played everything from Smite to Rocket league ranked and OW has the most number of uncompetitive matches out of any game ive played. I do think OW could fix a lot of them with a ban system for ranked. Also it's win 5 or lose X amount of games is just. Especially when you win 5 and lose 1 and go no where. Like I understand how that would Jalen in an Elo system. I'm a 1700 in chess if I lose to 1400 but then beat 5 1500's I'm not going to raise my Elo. However, when they are telling me all of these games are in my rank it visually makes no sense.
It's the weird problem of trying to attract casual players into a competitive environment. Casual players don't like seeing their rank from a psychological standpoint because ignorance is bliss. Truly competitive players don't have this problem because they're in it specifically for the stakes so they want to see them, me included. Chess doesn't have this problem because it's not owned by a company that wants to expand into the casual market. Casual gamers probably aren't playing chess anyway.
My hottest take: The overwatch ranking system is nearly perfect for representing your skill as long as you are Masters or below, and you play enough games. I will leave the GM rank debates to GM players who experience that, which is almost like a different game
I just want to know what the game actually takes into account because I've played roughly 50 matches at this point and am still stuck in Gold 2. I get average 14k healing, 20 or more assists, sometimes (if I'm Zen or Brig) I also get roughly 15 elims and I'm usually good at positioning and dodging attacks. It's just really frustrating at this point.
That exaplains a lot. Most of my comp games I played 'just like that', 'for fun'. I didn't think about being better. I just played. In one day I made a decision I wan't to be better. I fixed many bad habits, got more skill, better aim... and what? I hard stuck on same level for many many games. Even if I was playing better than before. Eh. It seems it's time to create new account and check what is my real rank currently.
Are you sure that an account that is stuck in a rank for an extended period of time would have to win such an exorbitant amount of games to begin ranking up? I am confident a player that plays 500 games at silver 3-4 can still climb to silver 1-2 if they win something like ten out of their next twelve games. It would not be a good system if it does not account for a player suddenly improving. It would also mean that there would be some accounts out there than even a top 500 player could not boost out of their rank in a reasonable period of time.
I was exaggerating a bit when I said 150 games, but the general idea is correct. A top 500, although it wouldn't be "difficult" it would take them much longer to boost a stuck account compared to a new account.
the thing i dont understand is im a silver 3-4 support but when i play with my friends they are a low plat on dps we end up getting a lot of gold games and i can usually keep up with the rest of the supports in the game but when i solo que it seems like im playing against dimond players and i cant do anything to provide value to my team. when im queing with my friends i normally average 10-14k heals with usually about 9 to 4 kd but when i solo que i average about 8-9k heals with a kdr of 1 to 1
The mmr versus division difference is very jarring. I could be considered gm 3 by mmr but masters 3 by the divisions and this is clear when you go 5-5 and still go up 2 divisions
Yeah this usually only happens on newer accounts, or accounts that are being hard boosted by a smurf though. The bigger the difference in MMR and visible the rank, the faster it changes your visible rank.
Overwatch is the only game where it feels like the rank system is actively trying to keep you from ranking up.
try rank up in league m8, ow ranking up is infinitely easier
100% id be going 5-0 on 20% already and I won’t move at all 😮😢
OW2 worst ranked in history of mankind.
@@jaber5345because the boosting mechanism in ow 2, sometimes when you get lucky , you can rank up really fast. But the dark truth is , if you rank up fast , so is every other low rank players ,they all have some lucky games and boosted in to high ranks. So in reality , you are still playing in a gold/plat lobby even you ranked up to Gm or high masters.
@@jaber5345 bro what? assuming you get no inters ranking up in league is easier than its ever been, why they added emerald rank you think? lol. both ranking system has nothing in common whatsoever
This leads me to believe the new penalty for qp leavers is to encourage full games for better accuracy to profile mmr on players.
Potentially, but I think the primary motivation is just to keep games a bit higher quality in general in the quickplay environment. If the losing team always quits then nobody gets to play any games at all.
@@Casual314This might be a cynical opinion, but I personally think people are going to troll or throw games regardless, even (especially?) in quick play, so I think punishing leaving unranked games will increase the amount of people throwing a game they don't like instead of just leaving, which might just be the better outcome for all parties lol.
Basically, I think trying to punish leaving to improve quick play match quality has the opposite effect.
@@strangejune A single anecdote doesn't count for too much, but I've already had a match where somebody who would've left on the enemy team started afking and throwing, when questioned on them afking they said they would leave if not for the penalty.
@@strangejuneyeah i was about to just put my controller down the other night but played it out because there was a minute left
@@strangejune so true tho, when im tank and i get so heals or bad dps i can just wonder off as tank and explore the map while i wait to be defeated and try again lol
i love how you mentioned that the system only evaluates your relative skill within the ladder and not your absolute skill.
so many people dont understand this,you can absolutely be improving steadily and even make large skill gains but stay at the same rank if other people improved at the same speed or the ladder doesnt reward your skill gain(eg,meta changes or the skill gain is just inefficient in getting you wins)
Absolutely, in fact if you stay at the same rank for years, it probably means you're improving, but only at the average rate of everyone else so you don't see an increase in your visible rank.
Honestly I miss seeing my elo numbers like in OW1 because win or loose once u get those 5 wins seeing my rank go down is still demoralising. Ive gone from P5 to G3 in my most recent games. Im not the greatest player but still id rather know if im going up or down than making it to 5 wins and then derank. Feels bad man
Agreed, I would much prefer a straightforward number that shows your rating going up and down between games. My guess is that Blizzard simply sees that engagement has increased with the current system, so they're not motivated to change it.
It's designed that way to keep you playing. Explanations that try to tell you that the system isn't perfect but it works and that you just need to get better keep feeding into that time sink. Buy the Battle Pass. Unlock that new shiny Hero. Maybe you'll do better and rank up.
J
I can hardly play 20 ranked games in any given season these days. I really wish they offered quickplay with a ranked ruleset, like how it was in the alpha/beta. Great video - you really put things into perspective.
Agreed, I really wish they had the same ruleset for Escort/Hybrid
To me, the frustrating thing is that rank doesn't have anything to do with skill or knowledge of the game. I want to play with and against people of a similar skill. But everyone is all over the place. I've seen Grandmaster players in my diamond lobby losing to the most basic things (like running alone out of spawn after the enemy captured, dying and staggering) where I think that's gold or silver level behaviour. And yes, it happens in the enemy team too. But that doesn't help, I don't feel challenged when the enemy just feeds or are way below my skill level.
I often take a look at profiles and see the wildest things. People who start in plat in S1. Don't play S2. In S3 they're masters after 2 games. Then in S4 they get GM5 without a single win. When they have played
It's inevitable with a system based on statistics unfortunately. Randomness means that it's impossible to guarantee the exact quality of a match unless they somehow got real people to handpick players. Even if an account isn't new, as long as they don't played ranked often enough, the system can't really be sure where to place them, so you can get some pretty high variance in behavior. Even if the system does well on average, the only moments you will notice are the outlier events, in your case it would be the system overestimating the skills of certain players by a large margin.
I was in a diamond match where an ana shot the bottom of a car with her entire clip trying to heal us😭
The second paragraph happened to me once. I met a genji player who flamed me for being a “typical rein main” when we lost this control game 1-2.
I called him out for dying 15 times (more than anyone on my team), which he replied with basically him trying to carry the team, hence why he’s dead so much.
Then he boast about how he’s a diamond player, and a “lowly platinum” player like me should respect him
Then I looked into his profile. Dude was diamond 5 in season 6 as a SUPPORT. Funny thing is, he didn’t even played comp on season 6. He was Gold 3 in season 5, yet somehow he got into diamond 5 the next season without playing. He only consistently played Damage in comps and he’s hardstuck silver. Yet he uses being “diamond” as an excuse lmao.
@@lordpeterturbo5216I’m also currently hardstuck high silver/low gold on DPS, and High Diamond on Support, but I also usually have the least deaths on my team and close to the highest amount of elims, or the highest. I play a bunch of different characters depending on team comps and maps, but the trends for my performance are still there in my K/D ratios for them.
It really makes me wonder if I’m in that limbo of being hardstuck due to the mmr system’s confidence, or if there’s something I’m missing for DPS. My positioning and timing are great on support, and I can get solo picks or coordinate kills on DPS, so it’s just like “What gives?”
@@kommandantkillcode Is your winrate consistently much higher than 50%? If not, then you have to think maybe you're missing something. If you jumped into gold lobbies as a high dia support, you'd definitely have a high winrate. Same would apply if you were actually higher than gold/silver skill at DPS.
The way you explained the probability of all variables in a match and 50-50 win rate was done excellently .
Everything you said in this video is true. The one issue i have with ranked is not being able to see your teammates ranks. i feel like adding private profiles, removing ranks from borders, and the removal of sr was not to stop toxicity but to allow for quicker matchmaking.
I started checking in on my teammates profiles and of course most peoples profiles are private but for the few that are public im seeing a mix of all ranks in every match. Im in masters and theres everything from bronze to gm in every match. This wasnt possible in ow1 because your rank was displayed on your border and visible to everyone that everyone was within the same rank more/less.
Could be possible. They probably keep track of stats that show how player engagement correlates with queue times. When the queue times are say longer than 5 minutes, maybe people just don't play the game as often. Being able to select from a wider skill range means exponentially faster queues, and I would guess casual players prefer that. Not saying it's good or bad, but that's probably how Blizzard sees it. The good things about Elo/MMR is that it already takes into account skill difference to calculate your score, so even if the quality of games are bad, you can still climb.
@@Casual314 I think that that's were most of the frustration with ranked comes from. In ow1 you had less impact in 6v6, but in ow2 it feels like 'among us' with "which players gonna be the bronze". It's climable but the quality of matches just feels so bad sometimes. Although I'm sure there are matches where I'm the "bronze player" of the team and it probably evens out.
OW's rating system was closest to Elo in season 1 of the original six-player OW. Elo's system by the way is not effective at team games, *unless the team is the same group of players* for all games. This is well understood in the mathematics community and there are other systems out there that *do* work for pickup team skill metrics.
It was after players complained in S2-3 that it "took too long to rank up" that Blizzard added skill metrics to the rating system, giving some players bonus points if the "played well" during the match. Scott Mercer explained this himself in the forums many times. It was at this point that players began to bounce wildly as much as 2000 points up and down between seasons. It *is* broken.
The next change was to hide your adjustment when it's negative. You get to see your SR go up when you win a game, but if you lost, you'd just be humbly presented with your new SR and have no idea how many points you'd just lost unless you'd written it down.
Then Blizzard started hiding your SR between seasons and calling it "placement," although nothing was different. Your SR goes up and down after every game and you get to find out what it is after playing so many games. This also lets Blizzard manually shuffle everyone around arbitrarily to make up for failures in the matchmaker. This has also been explained by Scott Mercer himself.
All of the changes since have been purely cosmetic. Your skill used to be a decimal number between 0 and 100. Then it got changed to an integer between 0 and 4000. Then they made it 0-5000. Now increments of 100 are named starting at B5, B4, B3... and so on; and the last two digits are used to tell you "you're ranked higher than __% of players".
*It's the same exact system* we've always had. And you hit it right on the head when you said it's not transparent.
If players knew what to do to rank up they'd find the game more satisfying to play. But they don't and that's just how Blizzard likes it. You can find a hundred "how it actually works" threads in their own forums and each one is filled with dissent.
Why do they do this...
tbh it feels like if you lose at least 2 games, the chances of you ranking up goes down the drain. i can go 5-2 and drop in rank % and drop tiers completely if i have more losses. I tend to straight up win 4 games in a row and then get hit with a loss streak to ensure that i dont climb.
In QP at least, it often feels like the system gives you forced losses, for example you have a plat tank on your team and the enemy team has a GM tank. When I felt like i was getting a streak of forced losses, I would also get games where my team had 4 players for the entire game. What actually seems to matter is the hidden MMR system, but it would seem like in OW2 they are willing to sacrifice accuracy for match creation time.
Overwatch also has the problem of so many different heroes being not only different to play but also being good at different things that it's quite difficult to accurately guage someone's skill based on that. Other games like Counter-Strike have a much easier time assessing your skill because every player has the exact same options in terms of gameplay as every other player and also there are just certain guns that are a staple of certain stages of the game. If you can handle the AK/M4, the AWP, the Deagle and the SMGs quite well then you are golden for pretty much any stage of the match, be it buy rounds, eco rounds or pistol rounds. Your team is not handicapped because it has two amazing AWPers who have to compete for the same spot. But on Overwatch if you have 2 Genji OTPs as your DPS then you have a problem. One of them will be pissy about not being able to Gengu and has to play a hero they can't perfom at the same rank as they are matched into.
Your perspective on the topic is spot on. It's true that many players feel trapped in "elo hell" because they're consistently matched with players of similar skill. It's not about a few bad matches; it's about recognizing where one's skill truly lies and working to improve from there. Well said!
But people arent. I'm gold, and I get a diamond player in yj games ALL THE TIME. Does that mean I'm diamond? Hell no. It means the diamond player has played the system to be ranked lower. It has happened all the time in OW.
Until you go 72 and 12 in a game. And start to wonder wtf is match making doing.
Elo hell does exist.
I had to beg my dps teammates to switch off their picks to become useful after going 2 and 2 for the first 5 mins. While the rest of the team is around 15+ elims and 5+ deaths....
Elo hell is playing with a team where you're doing your best and your teammates just wanna have fun on widow or hanzo while the enemy team is rolling you.
@@starshottai2519bro if ur truly not supposed to be in your rank just keep playing and you will go up
@@mmaenjoyer1the problem is that the matchmaking is so inconsistent that often whether you win or lose is entirely up to chance. having played there is genuinely no difference between bronze silver and gold. as a gold tank player I have gone against other tanks that are honestly so bad they should be bronze 5 while at the same time the few times I've played as a dps in bronze I've gotten teammates who are way better than the some of the ones I get in gold. this would be a lot less of a problem if your rank was based on game performance, but it's purely off of win/loss rate. The problem is that in OW2 they changed the matchmaking to be less accurate for the sake of faster queue times but didn't adjust the ranking system to account for the less accurate matchmaking. I feel like they should have never changed it, it was perfectly fine in ow1 wasn't perfect but it at least worked.
This proves you are a better tank than DPS. I need you in my team. I play in low diamonds and most matches are just steamrolls by your team or by enemy team. Sometimes I feel may be it's because of me and I don't belong here :).
If performance-based MMR doesn’t exist, then why can 2 players create brand new accounts and play every single game together (even QP before reaching Comp) on the same role (e.g. 2 Supports or 2 DPS) and still end up at different ranks after completing their placements? The more skilled player will always be placed higher, so there must be some other way that the game is tracking individual performance, even if they claim that performance-based MMR doesn’t exist. And this is something that has been proven many times by many streamers who have done Unranked to GM in duos on the same roles since the launch of OW2. Any idea why this is happening if performance-based MMR is really not a thing?
Quickplay games and probably performance are used to estimate what initial MMR you should be at. This is how unranked to GMs are done so fast. When I was said that performance doesn't affect MMR, I was not talking about this, I was talking about ranked games themselves. Your performance in ranked games do not determine how much MMR you gain. You may ask, if they can estimate performance in Quickplay why don't they use it in Ranked? Because it's not a good estimator, it's only used really to prevent smurfing, and cannot be used to make statistically confident claims about someone's skill level.
Do these players play from the new IP adresses and PCs? The assumption of the new account of the old player for the familiar PC configuration is very natural and can be confirmed by some in-game signatures. It is not that they surely do this but this absolutely could be done.
@@Casual314 Hmm I'm not sure if only QP takes into account individual performance, I tend to believe that Ranked might do so too even though there is no definitive proof. Because I've seen Unranked to GM duo streams where people buy Bronze 5 accounts with near 0% winrates and KDR and play in a duo in the same role, and yet the better player always climbs astronomically faster than the weaker player. In that scenario you would think that they should climb at the same rate if both accounts start at near-equally bad MMRs to begin with. And even if one of the throwaway accounts has a *slightly* higher MMR than the other, the better player shouldn't be climbing 2X faster than the weaker player if they're playing all their games together on the same role. Of course, it may not necessarily be performance-based MMR that is causing this, there could be some other factors that we might not have taken into consideration or even thought of, and perhaps there is some other explanation and hidden calculation that Blizzard simply isn't telling us.
@@AIIMight They've confirmed performance in matches does not affect mmr at all anymore, in competitive games. And even in ow1 it was only for ranks below Diamond.
For new accounts now, who havent played comp at all, their QP performance does affect the initial seed though. There it might be "performance based" even to some degree just for the "seeding"
unranked (or bronze) to gm duo stream sounds super pointless, though.
And also, if talking only about QP - your "qp mmr" is not affected by performance either.
@@AIIMight Blizzard says that there is no performance based MMR but who is to say that is actually true? In fact with the example you gave with streamer duos going bronze 5 to GM with bought accounts and climbing at different rates despite playing every comp game together, I don't think there is any explanation other than performance based MMR existing.
For certain roles- tank especially- MMR should absolutely take into account game stats. If you’re performing well, but your team doesn’t take advantage of your good performance, it shouldn’t be on you.
Also, if someone leaves the game, that game should not negatively affect MMR for those who stay.
Leavers actually benefit you in the long run. There are 5 potential leaves on the enemy team, but only 4 on your team, so statistically leavers make it easier to climb. It's understandably frustrating, but the reason they don't cancel games with leavers is to prevent people from asking their bad teammate to leave, asking their friend to leave to help them boost their MMR, or wanting to waste the enemy teams time by leaving at the last second. If they cancelled games with leavers, it would just result in frustration when the enemy team leaves rather than your own.
Absolutely not (stats not the leaver), the losing team is always going to have poor stats. Did the tank die off cooldown because they were bad or because their supports refused to heal them? Your output on tank is currently limited by your supports.
@Casual314 This is so true, leavers actually work in YOUR favour, assuming you're not one of them, which you shouldn't.
Good explanation. If accurate, it highlights a problem we experience a lot in support roles: you talk about the 'stickiness' of MMR, how as you play lots of games it becomes harder to rank up or down because of the averaging effect (the sort of Bayesian 'prior modifier') that is slow to update. The problem with that model is that when you are in a support or even tank role, where it is hard to carry your team, that exacerbates how slow skill improvements are to update, if at all. For example, let's say that your win probability in a support role is only 30% tied to your performance and 70% tied to your team performance. That mutes the effect you play on outcomes, so that any skill improvement you have takes a lot longer to be recognized (because the 'noise' of your team's performance anchors you at whatever your current MMR is). So now you're doubling slowing updates. Funny enough that problem could be solved, by the way, if they did something that probably hurts the rest of the experience: add MORE variance to match up MMRs, but only as an individual. For example if you are at MMR 2000, they should occasionally stick you in MMR 2800 games, where all 9 other players are similar MMR except for you. Having say 5-10 games like that would reveal whether your MMR is appropriate. If it is, you'll lose a lot more than you'll win
I stucked forever in gold, shortly before I finally hit Plat I got a big losing streak. They sent me from gold one to silver 4, now I stuck in silver 2. I overperform most of the time, but getting bad teammates, trolls and leavers in like every match. I lost 7 matches in a row, two times someone leaves, one troll who run in all by himself like he is a tracer and other times my dps or tank getting 0 kills but 11 deaths.
I can relate, sometimes the games are indeed frustrating. The other day I had a 7 loss streak as well with 3 leavers on my team. I ended up going 10-9 at the end of the day, but those 7 losses felt very frustrating when it was happening. I think maybe you're losses are maybe affecting your performance. Try your best to stay positive, and not care too much about the win, and I'm sure you'll be able to climb back up and beyond.
Here's my advice, delete your account...and make a new one
I appreciate the video and all, great work, but I am BEGGING that in future the screen isn't cutoff around the sides. It's been putting me on edge seeing the HUD cutoff slightly on both sides xD. That is all thank you.
This video is spot-on about the ranking system. I wish everyone knew this. I probably wouldn't have known this if I didn't have a gm tank duo who's been coaching me since I started playing the game. A few of my biggest problems trying to rank up was just not doing well on a certain character. I really like Ana but I used to tunnel-vision and focus mainly on healing tank, so I missed healing my other teammates. I switched to Kiriko and I won more. It ended up fixing my tunnel-visioning problem so I did get better on Ana but if I had just played Kiriko instead of all the Ana games my rank would have been higher (for reference I fluctuated from initial placement of gold to plat, back down in gold even lower than my initial placement, and now I'm diamond).
I'm now experiencing this on dps. I love playing Ashe and Widowmaker but my aim is extremely inconsistent. I can carry plat lobbies when it's good and be mid in gold when it's bad. I initially placed plat 3 and went on an epic loss streak to gold. A few days ago I figured out Soldier is my most consistent dps and went on an 8 game win streak. I play a lot in diamond lobbies and unfortunately get placed in masters lobbies sometimes. I know I "belong" in plat at the very least, but my character and lobby choices don't help me out. I enjoy playing high-skill heroes but I'm not at a point where I can always carry on them. I'm also learning the most difficult character in the game (Tracer) rn, so it's a rough experience.
Deranking feels bad, but for people who are dedicated to improving (i.e., not only caring about rank), it's actually a symptom of improving. I'm challenging myself by playing in higher lobbies. It's frustrating when I derank but I'm getting better at the game by playing against higher-skilled opponents. Before my duo reached gm, he got carried by his smurf duo from either diamond or masters to gm and got shit on. He went on a huge loss streak to diamond but he had improved immensely by playing against gm players. He climbed back to gm and actually belongs this time.
The best advice I can give to people who are unhappy with their rank is to stop caring!! If you keep improving at the game, you will eventually rank up, so who cares what your rank currently is? Treat every game like a learning experience. Every time you die, ask yourself "What could I have done to not die here?" or "How could I have played that differently?". Be excited to play against high ELO players in qp. You'll find it very difficult to get better at the game if you hang out in low ELO queuing with your bronze friends.
Also, I don't know if this is a common experience, but I hyperfixate on a specific role. Tank was my first love. I actually started playing this game because of Wrecking Ball (I started playing earlier this year so idk what ow1 looked like). I initially placed gold and just climbed through plat. Finished plat 5 my first season playing the game. Didn't even place on other roles. Climbed to plat 3 the next season and then started disliking the role because of the Orisa meta. I turned my attention to support and then more recently to dps. I came back to tank last season and assumed I was still stuck at that rank, but to my surprise, I was carrying all my plat games. My game knowledge had improved simply by trying to get good at other roles. I easily climbed to diamond. At that point, my support was gold, so I focused on getting better at that role and eventually got stuck. I turned my attention to dps and I just ranked my support to diamond. It's a cycle of improvement. I think I'm stuck and then I rank up on some other role. I think it's a great way to improve. Keeps you from being hardstuck in a singular role.
As a Mathematician my biggest gripe about the system is the lack of transparency to how they adjust the ELO based on the 5 person team. Is it an average, is it counter role, is it a mix of the two...these things really affect the outcomes a ton and it "feels" like quite often one really bad, or really good, player can absolutely make a game unwinnable. And that can then skew the results and over time end up soft locking, a player into more and more of those match-ups. Add in a liberal sprinkling of active smurfs, and the fact that it seems that there's some sort of match-up multiplier for people who play in groups (a multiplier that seems to be very heavy handed) and the game can become really un-fun really fast. I don't really care where I am on the ladder. But I DO care about having games that are fun and competitive and I can easily play 10+ games that are just so lopsided (either way) that it's not even fun. And yeah, I acknowledge I'm probably an outlier who doesn't enjoy stomping my opposition any more than being stomped.
One thing to remember is that within every rank there is a level of competency from the worse 1% to the top 99% with the botton 1% being much worse than the top 99%. This is extremely noticeable in the very low ranks and very high ranks as variation of skill sets such as mechanical skill can vary wildy.
In the lower ranks, players also behave very "randomly" in their performance. They play poorly but not in the same way every game (because they don't really know what they're doing) and so you can get very low quality games in low elo due to the matchmaker not being able to predict accurately the skill of lower elo players.
I have a friend who just climbed from Bronze to Silver after many months of just starting the game. He's an insane Widow shot. Just fast and immaculate. But! Has the game sense of a Magikarp. If the enemy team has a Sombra or good Reaper or anyone who can hide their approach, he's screwed. I've had to aggressively coach him to think about more than his next headshot target.
@ericaltman4087 I had randomly a game last night in comp, and it was bizarre. Like 3 out of 5 on the other team were making such bizarre decisions such as the bastion firing on their own team while sitting near us, their genji trying to fight our tank with just the sword dash and reflect(none of the projectiles or jumping unless by accident it seemed) and their Moira throwing the healing balls at our team. It didn't feel like throwing but rather just putting 3 absolutely brand new players together on a team in game whom haven't played a game at all of overwatch. I have never seen anything like it, even in bronze before.
Losses and wins should count for the same, regardless of rank difference between you and the average rank of the match.
This hidden SR should only affect what average rank you play agaisnt.
Basically the current system results in players who've had a losing streak have an extremely difficult time to regain their status. It inversely rewards lucky players who've played the game little.
More than anything else, all rankings and matchmaking algorithms should be transparent and visible. No hidden "man behind the curtain" bullshit.
Agreed on transparency, but it's just the unfortunately reality of human nature that the more people know about something, the more likely they are to interpret it incorrectly leading to even more confusion, and therefore less engagement.
The reason that taking skill difference into account is necessary is because of the natural bell curve distribution. Unless you're right in the middle of the curve, most of your games will be skewed towards either being lower than your skill or higher than your skill depending on which side you're on. In theory, the game is supposed to give you games of similar skill on average, but that becomes less possible towards the ends of the bell curve.
Nicely explained, great video.
It doesn't change the fact that their matchmaking system is sadistic. Feels as if the main character was the system itself. Trying to constantly figure out where you are, starts juggling and forgets about the most important thing: player's FUN.
I play to learn as I challenge myself through situations that I learn to solve differently. I learn nothing if u drop me in a match with Masters and GM, while I'm a Diamond and u put plats in my team. The game does it and thinks "let's see how u do". Then ofc u lose. And again, and again, and again. And then u end up 5 - 12 and the game says "oohhh...unlucky, u are not good enough, let me put u in a lower level, cause u are not ready".
And the same happens if I get free wins. I learn nothing by stomping the other team, and again I have 0 fun. Absolutely no reward.
What's the point of having such an overworked system if the game has the most random pairings, resulting in matches that, far from fun, are fucking cruel. I have never seen a game that, no matter how incredibly hard u try, crushes you so bad. As I said, sadistic.
It's the balancing act of trying to decide what to compromise on. In this case it would be queue times vs match quality. Higher quality matches = very high queue times, and most people aren't willing to wait 10 minutes for a 10 minute game.
This is why I don't mind Splstoon or Apex's ranking systems.
You win? Go up. In splat? Play a series, gre more points based on how good you won.
Let the rest tbe players fight for their spots, but let the way points are awarded be equal and consistent across players. How much you need is based on current rank, but you get no less or more then others. How high you are shows plainly how better you are.
It's a more transparent system.
One thing with the Splatoon system, and the Apex system is that they tend to be top heavy I believe. Meaning that more players are at the top than the bottom. This is undesirable if the purpose of the system is to be purely an indicator of skill, rather than simply a progression system.
@@Casual314 Precisely. It really makes you wonder why Blizz feels the need to make ranked another progression system, when the game has Like 3 other progression systems now to keep the players carrot and sticked. They don't need ranked to be anything other than a skill placement.
They're telling me they can't fix ranked sooner or go back to a displayed mmr system, when they already have other progression systems to keep players occupied... BP gross as it is, challenges, hero mastery, cosmetics to a lesser extent.
Didn't feel the need to mention since it I just wanted to talk about how much I like apex and splat's better and describe it, but yeah, what they have now is purposeless and needlessly opaque. And they're not even fixing it till like... middle of next year or some shit?
Very dumb.
A lot of people in OW1 got carried OW2 system makes you carry and play your own games to where you rank your self up through performance skill, and wins
Thank you for the video, now i know exactly what had happened to me after playing ranked for he first time.
I got ranked to silver, then to gold and then i had a big loosng streak and went all the way to bronze 1 and felt like crying, but after improving i am now at gold with support (70% wr with ana) and i'm climbing, since i actually got more skilled, can play at a more consistend level and i brought my mmr up through winning games by improving on my playstyle.
I'll ge so happy to start playing again once my unrightful ban is lifed (i got banned for telling my bf to shut up for teasing me in all chat when he was specating me bruhh)
Sorry to hear about the ban. Yeah you have to be careful with the automated chat monitoring, since it can't detect sarcasm or jokes. But I'm glad to hear you're motivated to start ranking up. Best of luck.
@@Casual314 thank you so much for the response to my very sleepy written comment, didnt expect that. Yup i realised that, sadly the moderators didnt even care about my explanation when i send a ticket in, but thats not the topic. But maybe if you have time you could make a video on their ban system aswell ^^
Thank you! :)
I really like the issues you talk about and I appreciate your method.
Some of the things you mentioned are the exact reasons why I don't like the change in the comp matchmaking. Before, you could play against higher or lower ranks, and you could "prove yourself" better against the Elo. But now, I only get matched against players of my exact rank, so the Elo is way slower to readjust.
This is the opposite problem I hear from others, most people complain that the range of skill in matchmaking is "too high". With the system right now, you'll still get into games that are above and below your skill level.
@@Casual314Maybe it’s because I’m silver 4? I guess it probably depends on the rank.
15:52 Bro gave us time to recover from the asteroid hit
I quit competitive a few seasons ago, there no point in climbing when a system doesn't let you climb, no matter if you're on winning streaks or not. Until they actually fix the rank system as they are planning to at this year's Blizcon, I want nothing to do with the mode.
Season 9 supposedly changes a lot of things in OW2, and also resets it.
My Qp games are insanely sweaty with grandmaster ranked players. I don’t play competitive but maybe I should, I could be a lot higher rank than I thought if QP is keeping track.
Potentially yes. Just know that the game can get very frustrating once you're keeping track so try to stay positive as best you can and the experience will be a lot better.
Great Video just a couple of things worth noting specifically for Overwatch 2.
Win/Loss Streaks have been mentioned that they will or already modify MMR change it was mentioned for the S9 Comp 3.0 explanation at Blizzcon.
Also, every old account going into Overwatch 2 kept the OG Overwatch 1 MMR with decay based on the number of seasons since a consistent competitive experience (mentioned in one of the Comp dev blogs). (This was because in a closed beta they saw a marked uptick in stompy games from Diamond+ players (OW1) in games against players they shouldnt) This also duplicates and extraploates the Old Account vs New Account issues, because new accounts have bonuses to stop Smurfing as well as New Player MMR flexibility.
Also another thing Blizzard have tuned the MatchMaker to find people with equivalent-ish MMR in the same role and place them between teams. I.e. your Tank has a Gold Badge but Diamond levelled MMR & Your opponents Tank is Diamond badge and MMR. This means that the teams average is usually less than 3 minor Divisions of MMR (An action of this is to normalise MMR change between rankd cause the teams are eqivalent skill on paper)
i think it's possible to model every type of value in overwatch with the right stats / weights and also not be abusable. i also think this would lead to the best matchmaking / ranking system possible for any team game for multiple reasons. one reason is that it unties your rank from other people (that tie is the source of most frustration and toxicity in competitive games). if you play extremely well in a game you lose, you should not lose rating; the opposite is true if you play badly in a game you win. another reason is that truly good stats would help people learn / improve at the game and focus on their own play. a player should be able to target specific stat improvements to rank up, and if done well, improving a stat would always equal more value (obviously not the stats we have now). if it were up to me, i'd have a person's rank literally be their place on a leaderboard of all players, where your place is determined by all your stats in different ways
In theory yes, but it's unlikely that a human would be the one to figure it out. With extremely high computation power, and some kind of performance tracking AI that has been trained on billions of games, you could theoretically have a system that can accurately correlate how certain players play with their likelihood of winning. You would of course have to track and calculate the game to every last detail, and the game would have to stop getting frequent updates because every update would change the meta making the AI predictions inaccurate. Given that Blizzard already has trouble creating a good POTG system, I wouldn't trust them to create a system that could somehow track and rank every moment of the game accurately.
Something to think about is a lot of those stats don't matter at all. Your ability to manipulate the enemy team into losing is the most important skill after all.
An example is how I sometimes play tracer. We could be near the checkpoint but the enemy can touch and reasonable win the battle. But I start diving alone when they leave spawn and trick them into chasing me. I waste so much of their time they forget about the payload and we cap. You'd be surprised that this works even in masters for me. What are my stats while doing this? Very little damage, zero elims, and I may even die. What was the value of this play? Immeasurable, considering my team captured the objective.
When you think about the game in this way, climbing becomes easy and you're not worried about the ranking system anymore.
As of now I have gotten to the rank change 4 times consecutively and not a single time did my rank change. Feels bad man
I'm proud to say that I am now subscribed to both Casual and Special
I'd like to see them do the same system paladins (and I think League? Not sure) did, where you get the different tiers (diamond I->V for example) but to get to another tier u have to reach a certain number of point (on paladins it's 100). You see a number go up and down so you know if you're making progress within your tier while also being confident you're IN that tier. You can also introduce a mercy system where if you're at 0 point you still need a few more losses to go down a tier to not discourage people.
It kind of works in both directions though, if it's harder to go down, then it'll also be harder to go up. The league system is built more to forced you to grind which personally I'm not a fan of. I'd be fine with LP, but in some ways it just adds a different kind of confusion.
OW1 had a number too. your SR number
The League system is horrible. It's more about grinding and keeping you playing than actually rewarding skill and performance. I hope OW never gets a system like that. At least in OW I can reach GM through individual performance.
Great explanation. Good humor throw in as well. Definitely gonna send this one to my nephews... I tried explaining the first point to them, but they didnt believe me.
He waited for 3 sec to see if the asteroid hit 😅
Had to be sure I didn't jinx it 😆
This was actually a great explanation of the way the statistical side works!
It can be hard to discuss these things in a way the general public can understand but I think you did a phenomenal job breaking it down and building up to comprehensive package.
Thanks, my goal was to make it precise yet easily understandable.
Still looking for explanation of: literally performing the best in the whole lobby with highest elims, damage, healing (yes im a moira main)
Yet STILL HARDSTUCK. I’m in Gold and almost every game I’m watching my team spawn in, then one by one running straight into the enemy team 1v5 and dying. Rinse and repeat.
Im watching DPS (when they actually hit shots) focusing enemy tank that is being double pocketed by their supports.
I was Plat (occasionally touched Diamond - PS5 fwiw) in OW1 and had to quit because of these frustrations. Recently picked up OW2 and I’m experiencing the exact same thing.
It’s like, the better you perform in a game, the worse players it puts you with the next game. So playing well becomes a vicious cycle of frustration.
Im trying my hardest to focus on my own gameplay and not blame my teammates (that top 500’s keep reminding us to do) but come on….
Smurf detection is ironically responsible for a lot of the matchmaking woes. I've been playing with a friend recently who's new to OW (still working on grouping as five), and he's got an inflated QP winrate because I carry so many games. Since he hit 50 wins, all we've been getting is Master/GM lobbies, and based on other folks' testimony, I'm certain that's the smurf detection pushing his MMR way, way higher than it should be.
The game pretty much only counts wins/losses so if you play with your friend, and you guys win a lot of games together that might be something that happens. Overtime, it will correct itself.
Thank you, this was so enlightening. I feel in your videos you talk to a big crowd of people out there who are seldom addressed by the pro streamers and content creators.
OW is my first FPS (away from some Q3 arena and Half Life 2,and that also tells you something about how old I am 😂). I learned a lot since I started playing and while I still mage a humongous amount of mistakes I feel like I shouldn't be b5 anymore (but the who does right?).
I'm super hard stuck even with positive win rates for a while now, but considering how seldom I play compared to the average player, I now have a clear understanding of why I can't even get to a higher bronze rank even if I should really deserve it.
With tank and dps I at least can play a bit of silver though I think I'm worse than on support. But I might be doomed there 😁.
Yeah part of the ranked experience is the grind, which is why there's no reason to feel bad about your rank if you don't have the time to play hundreds of games every month. It's hard to improve without a lot of time investment. No one's paying you to play Overwatch, so my philosophy is just enjoy the game in your rank. There's nothing "more fun" in the higher ranks anyway. Sure some people on the internet might make fun of people in Bronze 5, but are they having fun playing their own game? Probably not, so who's actually winning here?
@@Casual314 many thanks for the kind words. Very much appreciated.
Let’s gooo, I’ve been pumped for this video to come out. Hope your day is going good my man!
My day is going good Pineapple. I hope you're doing good as well. Thanks for watching!
Dropped from GM2 to M5. Currently at M3. This gave me the motivation to just play more ranked, cause if I do a good job over the course of like 200 hundred games. I will most likely climb. Thank you :)
Great Video! This needs to be shared throughout the community!
Trying to model skill leaves out a bunch of factors that people don't tend to count, marks and call outs, pushing enemies back/off of positions but not eliminating them, occupying enemy attention away from the objective which may result in being killed without an elimination but significant capture time for the team etc
Overwatch is the only game where it makes you feel bad for winning. Like even if they don't show MMR anymore, at least have some movement in between the ranks (e.g. you go silver 4 to 2 or 3 depending on those 5 games). I am new to FPS on the computer and I am new to overwatch, so of course I was bad. And I played bad for a while because mouse and keyboard was not natural. It's taking forever to rank up now, where I have been consistently getting more wins than losses (even if it's 5-4). I've done 15 games with no change. Staying the same for me is worse than having a number go up and down that changes each match, especially when it feels like there's forward progress. It just puts fear that the next games will drop me lower.
I've had one time where a bad loss streak didn't drop rank. I've had many win streaks where I've dropped rank. I've had countless 5-0s and 5-2s that kept me the same rank.
They basically try to do that with the percentages that they show when you get your rank update. If you've been playing at the same rank for a while, the percentage will be the best indicator of progress, although it's not perfect.
I really enjoyed listening to your voice- it was really soothing & I appreciate the explanation. Although it won’t stop me from thinking the universe is against me when it comes to choosing my teammates- at least I can look at it from a different lens.
I have liked & subscribed! 😎
I'm glad you enjoyed the video. Thanks for the sub.
Honestly, I miss the old SR system. Everything about Overwatch 1 was much better
This is actually the best video on this I have encountered
Great vid, the statistical confidence is something I hadn't considered before
The rank system is the new suprise mechanic
Still in the progress of trying to figure out how it works (through unkranked to GMs + bronze to GMs) and all I can say is it's VERY stupid.
I miss SR more with each passing day :(
I still have a gripe with this system in that I believe incentive of other players (teammates/enemies) and the quick play experience can skew the data very poorly for the user. Quite a few free to play players just don't care about their MMR, they get placed wherever and do not care where they end up.
Mostly though, QP is a terrible grounds to base skill upon, you can have top 500's playing against new accounts and the new accounts win against them because no one tries in QP, or they can just be screwing around for fun. Now this new account has a wacked out skill value that places them way too high when they go to comp (I have witnessed this as I have been climbing out of gold [now masters] since I was a literal child and now I can drink). So now these new accounts aren't having fun because they're getting toasted in comp, and then they artificially tank the MMR of every single person they get into a lobby with. This is a massive issue I have viewed countless dozens of times firsthand.
Diamond seems to be the worst spot for this. If I had to conservatively guess, one half of all my games climbing out of diamond had at least one brand new account that underperformed either on my team or the enemy team. On top of wacking out the MMR system, it doesn't feel good to win or lose when there's obviously someone being dragged along for the ride. Just as is with a leaver, it is nearly impossible to win when 20% of your team is dead weight. I also understand that statistically over time, these accounts will reach equilibrium between enemy team and my own, but it still does not feel good at all to play like that, and I also have to already work against the MMR system that is statistically confident in my metal rank OW1 days as a teen. Having one on my own team feels worse than beating one on the enemy team, because the statistically confident game likely takes more MMR from a loss than it would give from a win in this scenario.
Yeah I can understand that the way the system works can lead to some very bad games. It could have something to do with the QP estimation they implemented, but I can't be sure about it.
Im a plat/diamond player and honestly i just play for golden guns and also to try and improve :)
After playing many games i've come to realise that it works through a really frustrating system of "Winner" and "Losers" qeues. Once yo start winning you can get into a 5 winning streak and the system places ypu with good teammates that are climbing against those who are deranking from above. As long as you do not lose the game by yourself because you are no longer able to keep pace, you keep winning. And then it does the opposite. It places you in the "deranking queue" where you will be forced to carry your horrible teammates who are also on a losing streak until you derank so low that you can actually carry the game.
Surprisingly, as long as you perform well in these "deranking" games, you dont lose that much SR but it is really tilting and frustrating to see how over and over again you are the best or second best player in your team but carrying that mess of a team is just impossible. When you look it up online people just say" git gud bro" and they are right but 50% of the games are already sealed before they start and thats creates a terrible game experience.
Edit: I know it because when I'm on a losing streak I change to my alternative account and suddenly the system is friendly again.
I'm skeptical of the scientific validity of that kind of experiment. Blizzard has already said on their official blog that they don't do that. If I have two coins and I flip one and I go on a "head-streak", so I flip the other one and coincidentally I get tails, does that prove that the coins are rigged? Now, not knowing doesn't prove that it's not rigged, but I'm not sure the evidence is strong enough to draw a conclusion either way.
I’ve had a similar experience. The first 3 games are always very easy, I just need to not throw. I don’t think I’ve ever lost my first game in ranked in a session. Once I hit that 4th game mark, it gets much harder. My theory is that your mmr goes up from those wins and now you’re expected to carry teammates with a lower mmr, even if ranks are relatively the same. So your enemies get stronger and you’ll get some weaker teammates.
@@Casual314I mean... Bringing up "blizzard said" to prove a point is not exactly the move... Blizzard said many things which were not true.
Blizzard once said that Ashe NOT getting ult charge while Bob is active was a bug. Fast forward 3 years and they said that it was not a bug, and that she shouldn't get ult charge during the time. So I don't see why they can deceive people in terms of their game balancing decisions, but not in terms of their mmr balancing.
Your performance (stats) in the match doesn’t affect your rating. Only win loss does relative to your rank and the lobbies rank
@@129iphone Cynically, I think it might have to do with he way the rank system works, which means they want you play more to see you rank up/down.
Yesterday I won 4 games in a row and then lost 4 in a row before finally winning another. I still ranked up, but it felt like a waste of time and doesn't make a ton of sense given I have a low WR%
I've clawed my way from silver to platinum twice, in overwatch 1 and overwatch 2.
I've also clawed my way from silver to high gold multiple times.
I am not sure what kind of "certainty" the system has but if im good enough to do this multiple times, I think the system has failed to place me correctly.
In all honesty a more pure elo system would feel better and be less of a time waster than this "hidden mmr" garbage, statistics and machine interpretation should never see someone with a %60 WR and do anything but push them up.
it feels so extreme that nothing short of just hopping into kovaks for an absurd amount of hours until I am grandmaster level aim on a hero with a one shot and able to basically "smurf in my elo" would get me out.
Good video, the main frustration I have with the system is the combination of needing hundreds of games to balance out "everything that could happen" and the fact that the more games you play the more confidence the system has in your ability. I guess id add the fact that there's also zero individual performance metrics. All this culminates in me in a place where I know I am higher skilled that the lobbies I'm in and I AM climbing but its so slow because I had a rough start with leavers/throwers/people who look like they've never played an fps before. (Gold2 initially to Plat2 took me 150 games over 2 seasons) I consistently have a better stats than my team win or lose and I thought the system would recognize that and move me up faster, learning that it does not is kind of annoying. /rant
This is so true, I always try to tell people who complain about their rank all these things and they never understand. This is just intuitive, there are tons of bias and statistics in place. You are statistically not the unluckiest person on the planet
They used to have individual performance back in the day, and I miss it. It incentivised paying more attention to your own performance, and gaining +30 on wins and -10 on losses communicated very clearly "keep it up and you will rank up". That W/L is the main metric means people will be so toxic when they think a teammate is bad, and subsequently throw their own games as a result. This feeling those players get is bad for all players in ranked when in colors the perception of super fun losses that were tight games. IMO the solution should be to focus on sub-objectives like "2-3" on a flashpoint map probably meant it overall was even, and if those games don't feel good to lose, what hope do we have of immature players ever taking responsibility for their actions? The frequent draws didn't feel great on 2cp in OW1, but at least there was this feeling of it always being a possibility preventing teammates from tilting. So I'm not sure it helps that draws are so rare now. I'd rather take a 30 min draw over a 15 min loss for competitive integrity. A single 5-6 payload game should be counted as 5 wins 6 losses just imo.
The bad thing about this statistic approach is that you may need an enormous quantity of games, and some players may leave the game by pure frustration by then, because first of all a game needs to be entertaining and fun. Not being able to win 50% or winning a lot and not ranking up is not gonna work in keeping players involved in an already not as played game.
Yep that's a valid point, and some games take that into consideration, and make their ranked system lean towards ranking up so that players will continuing climbing as long as they keep playing. Then you just reset the system every once and a while so that not everyone is at the top rank. It's more of a design decision between feeling of progression or true competitiveness. Neither is necessarily better inherently, it just depends on what you're going for.
I actually never knew elo came from Árpád Élő. It was a great surprise since my country isnt often mentioned anywhere.
This was actually a really nice video. I like being able to think things over and understand a situation better so I can possibly help it. Especially if it is case or time-sensitive. There's a lot of factors that go into wins/losses and sometimes that can be wholly in or out of our control. You can do amazing and still lose on one small error that was made by anyone else on the team, including yourself. Accountability is important but so is making sure you don't go nuts trying to psycho examine things out of your control.
There are going to be bad games. There are going to be good games. The important thing regardless though is that they are just games! Have fun, take it easy, and feel a fire when you get competitive that instead of burning out at the first sign of trouble, gets brighter and bigger for the want to win and commit to the passion even more.
If your games ARE genuinely stressing you out and it's unhealthy, step back and try something new! You can always return later when you feel more patient.
Big fan of this mentality. If you're motivated to improve even when you're losing, it can go a long way. I think it's perfectly ok to say that having a bad team sometimes makes it hard to win certain games. The important thing is to take it easy and make sure you're actually enjoying the things you spend time doing.
Ah, this completely makes sense, so I had an account which I connected to a friend's PlayStation account when looking after their house and got told I couldn't reassociate my account to my PlayStation until a year later. On the account which I can't use, I was stuck in (no joke) bronze 5 as support. I create a new account to keep playing and that gets placed in silver 1... Definitely makes sense that Blizzard has something behind the scenes trying to keep accounts in the same rank then.
It doesn't help that it feels the matchmaking is really broken right now. I love OW and do enjoy comp but I feel like it's never consistent and I'll practice and bust ass only to rank down randomly. I get why the system is how it is but it feels like a lot of rage would ease up if Blizzard fixed matchmaking.
Keep up the vids! So nice to have an Overwatch youtuber talking about interesting subjects without shouting and overreacting all video :)
Thanks, glad you enjoy the videos 💖
It's better to just get a number like in Chess. One problem (aside from transparency) is they have assigned bronze too high. e.g. in chess:
600 is a new player
1000 is a player who can play but isn't good
1400 is an ok player
1700 is a decent club player
2100 boarderline pro
2500 GM
2800 super GM
In OW2:
Bronze is a new player
Bronze is a player who can play but isn't good
Bronze is an ok player
Silver and Gold are decent players
Plat and Diamond are very good
Master and GM are incredibly good, not dissimilar to chess
I've seen B5 players who clearly understand their role, and can crush unranked lobbies, I've also seen B5 players who looked like it was their first time playing and didn't have a clue.
There needs to be more separation in the lower ranks so that players of a lower skill level can still feel like they are competing and getting some progression.
I agree, it's the reason that League of Legends added the Iron rank. I think Blizzard knows that Bronze 5 is too big, but they're hesitant to add an Iron rank because it serves to make players feel even worse if they're sensitive about their rank.
Can someone explain to me how going 15-5 moves me up from silver 4 at 75% to silver 3 at 50%? I went 5-0, then 5-3, and 5-2.
Bro I had the exact problem earlier in the season I Litteraly thought s4 was broken. I am a g1 dpsin console and just switched to pc just to be stuck in s4 with no movement. I went 5-0 and nothing I left no games or anything.
@@Jamilbaker12I've been consistently getting an 80% winrate on DPS. Yet, been stuck on S4 as well. I've seen videos of the top dogs of OW analysing Bronze and Silver lobbies and I'll tell you - there's no way I'm at the same rank as these people.
@@CaesarLvcivs ikr
The individual competitive updates don't matter, it's every single game you've every played that matters. Imagine someone has played 10,000 games at Silver 3 at 50% win rate. Even if they go on a relatively big loss streak, it's probably just some luck.
Me and my fiancé play overwatch together. We play competitive, and I have a higher win rate than him. I also play on my own, he usually only plays with me. I play all different game types, and have a positive win ratio on all of them, along with a higher win ratio to my fiancé. I am currently at Silver 1. That’s currently my top this season. Wanna know where he got to? Fucking platinum. He managed to rank all the way up to platinum while I’m still stuck in silver, despite every single part of my account would tell you that I should be higher. I have more time on different heroes, higher progression level, higher win ratio with more games to make it accurate. All of my statistics are higher than his.
We make a joke that it’s because blizzard is sexist, but I honestly think it’s because I main healers. For some reason, I can’t climb as a mercy main. This makes me think that they rank you similar to deathmatch, on your K/D…and don’t really pay much attention to healing or assists. I maybe get a max of 5 kills in a game, because it’s not my job as a mercy to frontline damage. However, it doesn’t matter how many people I save, resurrect, assist…it feels like the matchmaking system doesn’t look at that…or more that it can’t also put that into consideration. It’s why we don’t get healing POTG, because they don’t know how to judge what’s good healer activity against someone who got a triple kill. It doesn’t matter that you kept them alive which allowed them to do it, that’s not what’s measured.
Blizzard has said that it’s not the case, but they already have that type of shit for POTG, and I don’t think that they’re being entirely truthful.
I love playing healers, it’s my favourite because I like being fast and helping people rather than going in myself. But it feels like I’m gonna have to start choosing between roles that I like, or ranking up.
The ranking system is flawed, but I don't know exactly how to fix it. In season 6, I decided to play both my accounts. My main, which I've consistently sat around D3-4 (3300ish) since ow1, and my alt account, which is only slightly newer than my main account. I played completely through the battlepass on both in S6. In total, on both accounts, I "won" 200+ support games. I played hundreds of games in S6 across both those accounts. It was a long season and I had no other games to grind as I waited for the autumn releases. My main account ended at D4. My alt account ended at P5. I did duo with a few friends on my alt account, but that shouldn't make the SR discrepency so apparent. These are both accounts that have hundreds of hours on them, both played consistently since 2017. You would think the MMR on them would closer? I played a large number of games on both, playing the same heroes. My alt account actually had a negative win-rate overall for S6. Main account was around 55%. It shouldn't be this tin-foil hat thing to figure out your mmr or theorize why it is the way it is. Blizzard can just change it. It's a pain point for every person who plays ranked and turns people off from even using the mode. The queues in ranked are already long enough without people forfeiting the mode all together.
I don't really care enough about rank. Shiny metal badges are nice, i just want to have some fun with a little bit of pressure. What gets me is when you have a terrible match and someone blames it all on you 'cause stats. When I'm pretty sure they've experienced both terrible and fantastic matches, but never think its them. It will never change, it's humanity. But it does wear me out sometimes.
I helped to win a game as Sombra in a cheesy way in a match that was probably Silver - i captured the first point and moved the payload to the end in times where the enemy team was unaware. The first point i had 0 kills and 569 damage. Regardless of what anyone thinks, that was value for the team and had nothing to do with damage, or kills, or deaths.
You may have 1,000,000,000,000,000,000 kills but if you forgot the payload and weren't with your team it just doesn't matter.
Yeah it's just something that you have to learn to deal with. Try to remember that most people are quite unreasonable. If you remember that, then just realize that when they say something silly, it's supposed to happen, because they're unreasonable. Don't try to reason with unreasonable people because then you would be doing something silly as well. This is the mindset I use so I don't get frustrated by silly people.
@@Casual314 Nice way of looking at it :)
Coin flip you say? The system approaches 50%?
That would be accurate if they stopped matching stacks with solo players...
Also, the longer you stay in a rank harder it is to leave, so one can not improve and feel the improvement as the system will tell them they are the same. Yep, amazing system that promotes "smurf" / "boosted" accounts.
I think the system should overlook past data, this will allow faster movement through the ranks. Also it should look for the impact you created within the confines of your specific role. A tank cannot be expected in the current meta state to just create the diff needed to win the game or to get 4k every team fight. So even if they lose the game there should be a way to see if they created the right impact but the team was not able to capitalise on it or just the tank didn't get the right amount of help/heals to combat effectively. Same applies to DPS/Support roles as well, if their tank doesn't create space and hold the lines they can get mowed down very easily. Perhaps if we can measure these, we can give people better ranks.
Lol I just left a comment on one of your last videos about how the rank system works, thanks for this upload casual
Haha that's a coincidence. Hope you find the video helpful, thanks for watching.
I’m happy with the competitive changes that are coming into season 9, I’ve been playing Overwatch since beta and was never really into competitive like that but more into QP. I always placed gold after my placements but recently hit platinum. Hopefully in season 9 I can hit diamond
I still hope for cards with 5 wins or 3 looses🤷♂️ would be way less frustraiding.
I know it works caus splatoon did it and i love it
From my brief research on the Splatoon ranked system, it seems that the system is skewed towards ranking up, probably to keep players motivated to play. Basically as long as you keep playing, you'll rank up eventually, just at different speeds depending on your skill. This is a valid design choice, but it does lead to the distribution eventually becoming "top heavy" meaning that most players will be at the top ranks. This also means that rank is in some ways more of an indicator or time played, rather than purely skill, which is probably why Blizzard wouldn't go in a similar direction.
MMR said I'm mid masters like 6 months ago. Ranked system said I'm plat/diamond. My account is like 500+ hours aged at that point, and I slingshot backwards even though my performance was fine.
I believe that one of the biggest parts of the video that you didn't address very well is just how bad the displayed rank is. I understand that the point of your video was to explain the innerworkings of the ranked system, and to explain all of the hidden mechanics, but when you talked about the displayed ranked system you so quickly glossed over just how bad the displayed rank system is. Speaking in general terms, even if the hidden mmr of a ranked system was PERFECT, meaning it could accurately predict the exact rank you should be in after every game, but the displayed rank system was terrible (akin to what we have in the game right now), then that ranked system would be a terrible system regardless of the back end being perfect. That's because the player would have no clue what they were doing wrong or right, they would have constantly have feelings of not being at the correct rank, and they would feel very frustrated at the system even though it's technically doing it's job perfectly. All that is to say that the inner workings of the ranked system, feelings of players, feedback given to the player, transparency of your rank, and displayed rank all matter. They're all connected and you can't just pick one piece of the puzzle out of the whole system to analyze it separately when considering the efficacy of the ranked system. It's the same fallacy that Blizzard falls into when making balance decisions: they primarily look at statistics like win rates and pick rates in a vacuum without considering how it feels to play a role, how it feels to play a certain character, whether or not mechanics are fundamentally good for the game, how it feels to play against a certain character, why it feels terrible to play against certain characters, etc. Therefore, I believe that most of the grievances that people have with the ranked system mostly lie in the displayed rank system. You mentioned the positives of the new displayed rank system those being less anxiety about ranked and the feeling of going from 2000 to 2500 feeling like more of an achievement, which I completely agree with. However, the negatives of the new displayed rank system are horrible. The benefits mostly arise from moving from a single number SR system to a badge system, gold 1, gold 2, plat 5, etc. But if you also gave the players a number to reference exactly how close they are to ranking up to the next division, then the transparency, feelings of the players, and feedback to the player would all improve dramatically. You could keep the benefits of the badge system while keeping all of the benefits of the old SR system without the ranked anxiety. The displayed rank system matters way more than you give it credit for in the video. Like I said, I understand that you're trying to explain the innerworkings of MMR, but you have to do a better job of acknowledging the problem. Your video frames ranked system frustrations as simply not understanding statistics and how MMR systems work. However, the problem is way way bigger than people not understanding probability and you should really state that better otherwise people will take away the wrong conclusion from an otherwise great video.
Thanks for watching the video, and adding your thoughts.
I understand what you're saying. You are correct that I wanted to mostly stay on topic, which is why I didn't address the perception of the ranked experience as much, especially because how the visible ranked system works could change a lot in the future. I also prefer to keep the descriptive and normative discussions separate, I don't want to add too many of my opinions to a video about something objective, because then it can be seen as dishonest. Which is why all I had to say about the visible ranked system is that it's based on the hidden MMR. Why it's designed that way, and what would be better is a different topic that in my mind would deserve a separate video. My interpretation of your criticism is that I should have taken the opportunity to say that the visible ranked system is bad, and here's how it could be better. But I don't think anything I said would have been particularly surprising to anybody in the online OW community.
Since we're here though, I'll just say that my opinion is that the visible system should be as detailed as possible, letting you see the number, how it's updating, what you can gain/lose, and make it visible in your profile.
In my experience, most people don't understand how probability and statistics works, which leads to a lot of questions in the form of "why doesn't the ranked system just (insert terrible idea)." They now add a percentage to your ranked update which kind of tells you how close you are to the next rank, but my hope was that if one understands how the system fundamentally works, then it's obvious why the visible system is designed the way it is, regardless of what it looks like. It seems that the no.1 problem with the badge system is that players wonder why they can win games but not see progress, but the same problem existed even with the SR system. Players would wonder why they were losing more SR on losses than they would gain on wins, and then claim that the system was broken just like they do now. But if you understand how it works in the background, then you can probably answer both of those questions quite easily.
12:30 The game does take your stats into account though, even if not explicitly for the purpose of ranking you up or down. If you're one-clipping left and right in a bot lobby, gradually you will be matched against people with higher mechanical skill irrespective of your rank or the age of your account. The only way to disprove this is to have Blizzard display everyone's rank again.
If you're one clipping people consistently you're probably hard smurfing winning most of your games, which means your MMR is going to climb faster and faster which is why you get placed with higher ranks very quickly. Blizzard is not purposefully placing people will similar mechanics together.
I wish the game cared more about playstyle and stats instead of just win loss ratios, I can go 50-6 in a game and lose and the game doesnt care how well i played
Yep I can absolutely relate, and I'd be lying if I say I didn't find moments like those frustrating as well. I just try to remember that some games I get carried when I'm performing poorly as well, and it all balances out in the end.
shit is so frustrating. its where im at rn i can do so good but it just doesn't matter if we lose like all the effort was for nothing and there's 0 metrics to aid me for my performance besides a scoreboard
@@RPIIIC you're telling me, i have 4 losses in a row with 30+ elims and 5 or less deaths and it doesn't matter
At the end of the day the W-L based system is a measure of how good you are at carrying games. Number of kills/amount of damage are related but not the same as carrying games.
I agree it is very frustrating to lose or derank when you feel you deserve the opposite, but the best course of action to climb is to simply improve even more. If you widen the margin by which you are the best in the lobby, it will become easier to overcome "bad luck" in matchmaking, and as you continually improve, you will continually climb.
The problem with that is, it would be super hard to make the "playstyle" rating fair across all heroes comparatively.
With some heroes, like widowmaker you could say its straigthforward, just look at number of picks etc. But then you have so many "weird" heroes and matches.
Purely W/L based is the 'fairest' pretty much without a huge development and testing effort, and also try to find all scenarios where it couldnt be abused.
iirc the performance based system was abused at times in ow1, as people found out which stats mattered, they began "stat padding" those, irrespective of other things.
mercys going for "bad rezzes" for example just to pad their rezzes per 10 min stat, and so on.
Never heard of the term "Doomed account" until now but i had a feeling sometimes like that existed and i just proved it.
Just made a new account last season and i threw(in a stack) down to bronze and sat there and marinated in it for a long time.
Recently began climbing the account and i could win 19/20 games and i only climbed 1 rank tier at a time. Despite how well i performed. I could have 30, 40, 50 kills with less than 10 deaths and still only go upn1 tier.
The account is nearing diamond and i finally got to skip a tier from plat 3 to plat 1.
Meanwhile my main account has been hardstuck low master for years.
The reason i threw was to prove to myself that i can actually climb to my true rank and i belong there. If im gonna be a hardstuck master player thats fine butni needed to prove it to myself.
Next gonna make a new account and tryhard from the start to see if i can get higher thannmy main
Yep "doomed" accounts are real. You will still eventually climb to real rank, it'll just take a lot longer than a new account would on average.
Another great video, you’ve been blowing up lately keep it up! 😁
Thanks for the encouragement 💖
@@Casual314 💖
I'm convinced it's my teams a lot of the time. My last games either have over extended tanks, Moria players who think they have high healing means they're doing good and players with overall bad positioning. I've gone games were I only die once and my team dies around 10
Just the other day I lost 7 games in a row, sometimes you'll just get unlucky. I proceeded to win 6 games in a row right after.
@@Casual314 true won 4 in a row then lost 6 right after
Well for whatever reason my rank updates are incredibly linear. I get 1/4 of a tier per win or loss +/- 2%ish per game. So for example, if I go 5-0 I will go up one tier + 25% of a tier basically. It might actually be one tier + 28% or something but it's entirely predictable.
Yes the system on average behaves like that, which is what causes people to expect that it will always behave like that. Once you're no longer climbing steadily, you won't get as large of an increase in your ranked updates.
Hi Casual! Interesting video - thank you! Coming from game development myself, could you argue why do you trust Blizzard has implemented the elo-system ”as is” / without modifications (e.g. coming from business guidance, user experience, etc.) ?
15:30 "somehow the universe itself is working against you, putting all the worst players in existence on your team" I know for a fact its doing this. Every game I'm in there is this one guy named Jing who is throwing my game.
don't worry they gonna fix it on OW3
Happy to see i'm not the only left-handed player
Another huge issue is that unlike Apex or Destiny, players are not incentivied to learn more (game sense) or improve their gameplay. If you compared the game sense and knowledge of a bronze Apex player and a bronze OW player and youll see that players simply dont take the time to learn how the game should be played. Thats why there isnt much of a difference in metal ranks. And sadly for those who are trying to learn and practice playing the game properly they still have to feed into bad bahaviors to get a win because the rest of their teams are not playing the heros right. Unless you have amazing mechanical skills to essentially carry your teams you will struggle greatly to climb regardless of how the comp system judges you because a lose is still a lose at the end of the day.
so recently i used my old blizzard to play ow2 to see what rank i get after 50 wins, i usually buy accounts that are done to bypass the 50 games. i played all my 50 matches one tricking genji. my first few comp games were GM2 and GM3, im a genji player, GM5 lasted me 25 games untill i went 5-14 and went down to M3. i was so excited hitting GM, this makes alot of sense @8:05
ok so if i lose 50 matches in GM5 - M4, with my MMR being in GM2-3 elo, wonder what rank ill get after this season ends.
I think they should go back to just showing the number. Maybe it's just because I'm a chess player and am used to seeing it, but seeing it after every match just helped communicate what was going on so much better.
In the case of OW I think it's got a number of things going against its ranked system. It feels like a lot of the games are so much more imbalanced. I've played everything from Smite to Rocket league ranked and OW has the most number of uncompetitive matches out of any game ive played.
I do think OW could fix a lot of them with a ban system for ranked.
Also it's win 5 or lose X amount of games is just. Especially when you win 5 and lose 1 and go no where. Like I understand how that would Jalen in an Elo system. I'm a 1700 in chess if I lose to 1400 but then beat 5 1500's I'm not going to raise my Elo. However, when they are telling me all of these games are in my rank it visually makes no sense.
It's the weird problem of trying to attract casual players into a competitive environment. Casual players don't like seeing their rank from a psychological standpoint because ignorance is bliss. Truly competitive players don't have this problem because they're in it specifically for the stakes so they want to see them, me included. Chess doesn't have this problem because it's not owned by a company that wants to expand into the casual market. Casual gamers probably aren't playing chess anyway.
My hottest take: The overwatch ranking system is nearly perfect for representing your skill as long as you are Masters or below, and you play enough games. I will leave the GM rank debates to GM players who experience that, which is almost like a different game
I just want to know what the game actually takes into account because I've played roughly 50 matches at this point and am still stuck in Gold 2. I get average 14k healing, 20 or more assists, sometimes (if I'm Zen or Brig) I also get roughly 15 elims and I'm usually good at positioning and dodging attacks.
It's just really frustrating at this point.
good videos homie always informative
That exaplains a lot. Most of my comp games I played 'just like that', 'for fun'. I didn't think about being better. I just played. In one day I made a decision I wan't to be better. I fixed many bad habits, got more skill, better aim... and what? I hard stuck on same level for many many games. Even if I was playing better than before. Eh. It seems it's time to create new account and check what is my real rank currently.
Are you sure that an account that is stuck in a rank for an extended period of time would have to win such an exorbitant amount of games to begin ranking up?
I am confident a player that plays 500 games at silver 3-4 can still climb to silver 1-2 if they win something like ten out of their next twelve games. It would not be a good system if it does not account for a player suddenly improving. It would also mean that there would be some accounts out there than even a top 500 player could not boost out of their rank in a reasonable period of time.
I was exaggerating a bit when I said 150 games, but the general idea is correct. A top 500, although it wouldn't be "difficult" it would take them much longer to boost a stuck account compared to a new account.
Had to win like 60 games in a row to get from silver to diamond lol
It's fucked
the thing i dont understand is im a silver 3-4 support but when i play with my friends they are a low plat on dps we end up getting a lot of gold games and i can usually keep up with the rest of the supports in the game but when i solo que it seems like im playing against dimond players and i cant do anything to provide value to my team. when im queing with my friends i normally average 10-14k heals with usually about 9 to 4 kd but when i solo que i average about 8-9k heals with a kdr of 1 to 1
The mmr versus division difference is very jarring. I could be considered gm 3 by mmr but masters 3 by the divisions and this is clear when you go 5-5 and still go up 2 divisions
Yeah this usually only happens on newer accounts, or accounts that are being hard boosted by a smurf though. The bigger the difference in MMR and visible the rank, the faster it changes your visible rank.