The mathematical best SF6 character is not who you think...
Вставка
- Опубліковано 27 вер 2024
- Check out Capcom's stats here:
www.streetfigh...
www.streetfigh...
and the Reddit thread with tourney stats is here:
/ character_stats_qualif...
Join our Discord server! / discord
Patreon: / jmcrofts
Twitter: / crofts
#streetfighter6 #jmcrofts #sf6
it's funny; you run into so many top tiers that you eventually get used to seeing them, but the moment someone whips out a low tier on you online. . . that's when you know you're in for a challenge.
Be it a knowledge check, a lack of experience dealing with their tool set, or even just encountering a real one; those matches are always the scariest. . . and most fun honestly lol
yeah fr, I lose to a lot of Lily, I know she is bottom-tier, and I only need a knockdown to beat her with Oki, but I don't know how to defend against her lol
then you add on top of the fact that say; you don't encounter them enough to pick up on the new tech those players figure out in the lab lol
Also Dhalsim mains have been playing this character since the 90´s they have a lot of experience
@@PrivateJoker0119 I think she is legitimately hard to defend against, much like Jamie. But they fall apart hard on defense because their reversals are weak at best. They fit the rushdown versions of their archetypes well, provided they can keep the offense rolling.
@@sundevil34 yeah fr, I hate jamie too...specially the ones who know they need to play lame to make him effective
Dhalsim mains rejoice. It's because we're the most enlightened
>(edited)
>"were"
The yoga fundies are no joke
Yesssir
@@yourdagan"enlightened"
@yourdagan couldn't see the grammar because i typed this while my computer was 12 feet away
Mathematically speaking, the greatest Street Fighter character is Twelve, followed closely by Eleven. It's like the rest aren't even trying.
That's numerically speaking not quite mathematically but good try
I would love to see a collab between JmCrofts and BrianF analizing stats.
🧐
ayo
unacceptable im bombing rashid's winrate to force my rhetoric
Bc there are 7 Dhalsims and they're all cracked. Just like Katalina in Gbfvr
As a Dhalsim player, I'll have you know there are 10 of us.
@@RandomRon005 11 of us. I'm the pink one
12 of us if you my ass in silver 4 lol
13 of us, i am Master
This is the power of yoga
Another mechanism is matchup vs top tiers. An interesting case of this is Abigail in SFV. After his initial nerf he was a very interesting case because he had even to winning matchups against quite a few strong tournament characters. For example, he probably beat both Cammy and Karin. However, he had a hard time dealing with Fang and Ed, but at that time those characters were much less popular, so the matchups being bad for him didn’t matter nearly as much.
So basically “how well does your character beat Luke cr.mp” tier list?
It's so cool that fighting game data can be made to be so engaging and thorough. You nailed the pacing man.
JMcrofts called it, dhalsim is a secret top tier! time to hit the lab! Great video.
I love stats like this. I think it also shows that SF6 is actually pretty well balanced overall, given how close most characters are on that chart. (I'm not saying it's perfect. There are things that need to be addressed.) I'll be curious to see how this data shifts as time passes
I swear I’ve seen this exact comment like a thousand times. We all know the game is generally balanced. We know there are problems with it. We’ve been saying that for close to six months now.
Bro we're such nerds cause as soon as he whipped out that plotted chart i was like "YOOO"
The same phenomenon that happens with these hard low tiers like Dhalsim also happens in smash (both Melee and Ultimate) with Ice Climbers. They're like super complicated both to play as well as to play against which means that IF you do play the char, you really know how to play the char. It's always funny so see these outliars be "universal" in games
I don't know about using it as a tier list. But I do think finding outlier matchups or trends like you did by graphing things out is fascinating as far as where to look. Jammerz was talking about something like this the other day on SETS, I think. How should popularity figure in...
I think the excel graph was very interesting. I love that you can see clear outliers in the data. Sometimes there are very simple mathematical relationships that accurately describe complex phenomena. I wonder if Capcom simply adjusted characters according to their "outlier method" if we'd see a better balanced game (Not that it's balance is bad by any means.)
OK so the graph was *super* helpful. Thank you Mr. Crofts.
Jab, jab, jab shoryuken
A typical Ken matchup
Interestingly enough, you see this happen in a lot of games. For example, Dota has statistics pretty much exactly like this where they show the pick and win rates of characters in each different skill bracket. Comparing these statistics to pick and win rates in professional games, there's a huuuuuuge difference. Part of it is correlation vs. causation; top players generally pick characters they perceive to be top tier and even if they're wrong, they're so good they still win anyway. This is especially true in games with frequent patches where there's not much time to see the meta develop, so sometimes top players are just wrong in their initial assessments but then just win anyway. Alternatively, and what I think to be the case in fighting games, sometimes the meta is just very, VERY different with the top 10-20 players in the world than with the top 100-1000 players. Top players are much less concerned with matchup unfamiliarity because a) they lab every MU and b) they can adapt extremely fast.
I'll repeat the same prediction I made at the start of the season because JP turned out to be good more due to having OD Amnesia and a lot of damage from defense: as soon as you nerf those two things, the character disappears from ranked.
As a Honda main with a pocket Sim, I approve of people stopping playing my characters.
I am surprised that there's no actual Theory video of MvC2 Sim, you know with people talking how he is the best character (i just got reminded of that)
So, I'm not a statistics expert by any means, but I feel like that chart could be useful if each character's win rate was normalized against their pick rate for each ranking (maybe with the pick rate for lower ranks factorrd in as well). No idea what that formula would look like, and it probably still wouldn't be applicable to tournament scenarios, but I think it might give a decent indication of how much work and skill each character takes to do well at each ranking.
no, we should take it seriously
the problem is that you're not reading and evaluating the chart properly
sim isn't the best character but rather mid and the chart literally proved it, his MU are nearly 6-4 to 5.5-4.5 on a lot of the top tier on the other hand although his weak MU are around 4.5-5.5 to 5-5 range he lose to a majority of the roster
on the other hand other then sim and kim JP on advantage average is around 5.5 while disadvantage is around 5.0 thats why JP is the best character in the game
and on the other hand all the low tiers MU rating is close to 5.0 on advantage and for ryu his bad MU goes as low as 4.6 and has a huge ton bad MU
as for ken his bad MU isn't that many and nearly all are close to being a 5-5 MU which confirm the pros evaluation of ken not being the top 5 character of the game, he doesn't have any really good MU but on the other hand he doesn't have any real flaws
It tends to be like this with characters that are hard to play and also "don't quite fit" in the typical gameplay.
Dhalsim is Long Range with the slowest Jump, teleports, relative slow drive rush, actually strong mix up in close combat. He plays nothing like any other character.
The same happened in League of Legends with a character named Aurelion Sol. He is a Space/Star Dragon who also played very differently from your average mage. He had stars constantly circulating around him, and that was his main source of damage. He could control them, but it was overall pretty clunky. Almost nobody played this guy....but he sat on a fat winrate of 55-57% not for weeks or months, no...for YEARS! At the same time, he was one of the least picked Champions.
I used him as my main for a whole season, and got a winrate of over 70% in over 100 games with him. It was incredible HOW rewarding he was if you put in the time and effort to learn him. Only maniacs played him, and if you saw an Aurelion Sol in higher Elo, you knew that this guy is gonna fuck you up.
Same goes for Sim. If you see one in 1600+ MR Master...you are almost certainly in for a ride.
There is merit to matchup spread for sure. The problem is basically that online and localplay are different games entirely. So each one cant really represent the other at all. But you can use the data to make your own decisions about what to expect- i think this whole chart should be redone for offline tournaments, since that is the missing half of this
Low population of sim players + regional matchmaking means that these sim stats are in fact for several players and can't be taken very seriously until there's much more data or region-based matchmaking is removed
I have different reactions based on the characters I run into in ranked/casual and realizing this about myself has been quite funny/eye-opening.
*Adorable/No Salt tier: Lily, Chun, Juri, Cammy, Jamie, Manon, Zangief, AKI*
I never get mad playing against these characters regardless of how well or poorly I play. 95% of the time will make faces together on the char screen
*Neutral tier: Luke, E. Honda, Rashid, Marisa, Kimberly, Dhalsim*
May or may not get salty depending on match performance. Usually chill experience, no strong opinions. 40% of the time will make faces together on the char screen
*Corny tier: Deejay, Blanka, Guile, Ryu, Ken, JP*
If you play this character you are corny. I will expect cheese, random DIs, little to no neutral, ODs on wakeup, etc. 0-10% of the time will make faces together on the char screen
It might also be the case that it’s too early to definitively create tier-lists of characters for the game
Man brought the spreadsheets
another aspect that could affect this is you CANNOT get deranked from master
Well here is something to think about. Tournament performance it’s not necessarily the best and ultimate way to determine the “best” because tournaments are “exclusive” events because say the best player of the game can only play online because he/she lives in a country without tournaments or can’t afford to go or has a mobility disability, is too old or too young, doesn’t have the time or maybe just maybe doesn’t care for tournaments. Plus in tournaments it’s also luck based say this player can easily defeat the one that won the tournament but got eliminated early. Sure some tournaments are tight but that’s not always the case hahaha. So long story short I think those statistics have to be taken into consideration too.
I take issue with the assertion that tournament results are luck based. If you're tourney ready in a game that has actual bad matchups, that means having a switch character that is not weak to the same strategies or move archetypes. It's also why you usually play sets (up to 15 rounds) in tourney. If there is any question as to who is better that day, after 15 rounds, GOOD! That's a good spectacle! And the person who won is the correct person even and especially if the result was a sudden turnaround.
@@tobiasvanavelon9684 well what I meant was the bracket placement you know when you are randomised to see where you are. And what you say about players switching characters if you have seen tournaments lately they hardly never change character even if they are losing miserably hahaha. "I'd rather lose with my best character than wing it with the best character for this match up that I have not mastered" and that alone makes the data of the video so valuable. Like 90% of the players will stick to just one character for better or for worse hahahahaha.
@@Glostlyn Bracket placement can't be a factor here, because then you're only talking about the best player 'to fight this specific guy'. If I only lose to my kid cousin and beat everyone else in the world, but my kid cousin can't handle anybody who plays a character other than my main, is my kid cousin who loses to the entire roster the best player in the world? Of course in the back of our mind we always know a tournament could have gone very differently. The POINT is that any of those people are arguably good enough to beat all of the rest of them. And the tournament is the debate they are having.
I think any data of win rate below master is basically worthless to say how strong a character is outside extreme outliers. If a character is broken good, it will increase your rank till you basically can’t be carried, verse you just be very good at the game and getting something similar. I don’t see a way to differentiate between “are these wins player skills or character power”
In ideal world, we have identical players with no bias of how the game works playing against each other to control all factors, but I think the best we can do is just go with “the highest level player pick and win we these” and the data just gets messier the farther down you go.
I wonder if there’s an equation you can run the data through to line up more closely with tier lists or if it’s just too complex to be really useful data
This is why data adjustment is done. There is plenty here to create a good tier list, it's just incomplete. Factor out the role of popularity for example and it'll get much closer.
I don't mean this as a critique, just some advice in case you didn't know but You should be able to add a trend line to the scatter plot if you want to make the downward trend more visible. Just takes a couple clicks
Time to win arguments with stats ;-)
Also known as the Lies, Damned Lies and Statistics strategy XD
Good insights. Thanks for the vid.
Really hope to get online & try ranked (if I can get the Capcom ID & PSN linked).
SF6's online content is so nice.
If people would rematch dhalsim maybe they would learn how to fight him
Oh, it's the return of the match-up list. The real take away is that the game seems very balanced. There aren't a lot of blow-out fights. The worst being something like Blanka losing hard to Honda. Instead, some characters just seem to be consistent, like Luke and JP.
I don't get why people are saying this, the whole point of MMR matchmaking is that it slides you up or down in ranks until you get near a 50:50 win/loss. If a character was bottom-tier they'd still have a close to 50:50 win/loss, they'd just be lower ranked than they would otherwise. If a character was top-tier they'd still have a 50:50 win/loss, they'd just be ranked higher.
Because there's so many Kens it also bring him down. Sim being at the top despite not being popular shows how matchup knowledge is important too.
This table obviously does not make for a better tierlist, but it does make for a very compelling argument on why tierlists do not matter.
I would argue it does make for a better tier list!
Data don't lie man. You just gotta make that correct questions...
Guile being at or near 5 in every ranking, thats honestly pretty impressive, dont play sf6 (yet) but to me that reads that hes pretty well balanced if hes doing roughly the same in button mash central (where id hang out most likely) as well as people wanting to head to tournies.
As a Gief main. Playing against a Dhalsim no matter what rank is tough.
yet another upload from goatcrofts
Mr Crimson alone prob skews the sim numbers. Its pretty cool that in master the top character is 5.14 and the bottom char is 4.88. Very tight balance and only 2 matchups showing worse than 6:4.
As a Ryu modern main… I shall either surpass Chun or become part of the Li
Ryu actually doing better in modern than classic on masters, just saying ahhaha, as a fellow Modern Ryu I approve this message
@@KywPT oh that’s actually nice to hear!!
If you go by CatCammy's 1850+MR chart Luke, Rashid and Dhalsim have the overall highest win rates/best matchup spreads. Interesting stuff for sure.
Ok, hear me out, what if a stage changes tier list?
I'm looking at you, kanzuki beach.
Another thing, most Master players have around 40 to 50% winrate, so that's a major reason for the negative correlation!
What would the tier list look like by the numbers using the tourney win rates as well
U should do a vid attemptimg Dhalsims optimal stun combo. AKA the highest damage combo in the game.
The American fgc, as usual, were quick to complain. I saw the strengths of Aki, Dhalsim and Marissa from the start.
In statistics you always have take into account what your population is and whether or not your selection represents that, which in this case becomes another point against using this data as a good tier list. This data comes from all of master rank which is not the same as top competitive play as there is a big jump in player skill between all of master rank and those who place top 8 in big tournaments. For a tournament relevant tier list you'd have to only look at results from matches played in tournaments. And then there's also the reasons you mentioned which are just as important. So yeah, not a good tier list.
The skill gap within master rank is only larger than it is outside. Capcom really should provide more detailed info.
I actually think looking at data like this can help you understand that maybe paying attention to tierlists isnt really the best idea for a majority of players. If you are in the lower ranks like bronze silver gold area, maybe try figuring out how to deal with the honda army thats smoking you instead of ignoring him cause hes not considered a top tier.
Great analysis
So Luke and JP are objectively the best characters by all metrics. Especially Luke with top 3 play rate, winrate and tournament results. As a Luke main, I’m praying for plot armor in the first patch!
Let's just be clear, win % charts are ≠ to matchup charts. Under represented characters will disproportionately out preform more common characters even with similar matchups. Low pick rate characters are played by specialists.
After seeing the video you said most of this but, while you shouldn't use a win rate chart as a tier list an actual matchup chart is always better than a tier list.
There's also playstyles to take into account. Dhalsim players will have a more solid and stable play style. Which is more reliable in the long run. While most Ken that are fight play like absolute maniacs. And being that random is a double edge sword.
Take Luke for instance, in my opinion he's not better than Ken, they have similar power levels, BUT Luke is more defensive. So Luke players play a more fundemental style : they buffer c.mp ans dish damage.
Ken players will run around do fake resets, dragon lash from full screen, abuse Jinrai, DP on every single oki and then complain when they lose and say the character isn't top 5.
You have to already be good at street fighter to pick Dhalsim. That's a skewed data set.
And this is why DSP only gets to Master with low usage characters
Coming from a Dhalsim player, the best advantage Dhalsim has is he's basically a rare Pokemon that people don't know what to do against nor do people want to learn the matchup. Unless you're in Battle Hub, a lot of your matches will be One & Dones.
As a dhalsim and luke main (both equal), I can confirm online people don't know how to deal with the shenanigans
Also, as a Dhalsim main, the learning curve is really really high. We get beat up in lower ranks more than average until we get good enough. Then at higher ranks we can abuse our privilege of being off meta.
I have always thought the idea of a "casual tier list" is a really good idea because it gives people a stepping stone of who's good. I see so many new players to a game just immediately look up a tier list to help them pick a character to play which means the lower tiers don't get as much play, and then they are lower tier partially because nobody plays them. But the main reason I think this specific data is valuable, is because it shows people what is really going on in the different ranks. Like how Honda is "really good" even though tournament results don't show that, because Honda IS really good at lower levels of play.
The main issue with that is that for a new player the characters power doesn't actually matter that much. Sure there are some newbie killers like Honda, where a new player can probably get to gold by just spamming two special moves. But outside of those, what matters far more than character power is player skill. At low levels, just having consistent anti-airs and being able to hold down back are far more powerful than any character, except maybe Honda lol, he really is a problem for new players.
But yeah, I think a casual focused tier list would be more valuable than an overall one, for new players, but at the end of the day they are both bad and new players should really play who they like, because that is what is going to keep them playing, they are going to suck and lose a lot regardless of who they play, if they choose a top tier then they will just feel bad because they are losing when they think they 'should' be winning. Also, you don't want to pick Honda, spam your way to diamond and then realize you don't know how to play the game and the character you are playing isn't actually good against strong players. Then you just have to start over anyways with another character, or be stuck playing a mid-low tier newbie crusher and never really get good at the game. (obviously this is all exaggerated to get a point across)
Also, I say this from experience. When I first started playing FGs again, after many years away. I simply looked to see who was the strongest and ended up playing a character I didn't really vibe with, and it sucked the fun out of the game for me. I think it was SF4 and I don't even remember the character, but eventually I realized that what mattered was me, not who I played but my ability and my choices and how much time I spent practicing and playing other people.
is it a coincidence that the most POPULAR character (Ken) is lower on the win rate list? Lately I've been feeling like people have figured out how to deal with Ken at the higher ranks, so for him to lose a lot more often kind of makes sense. My opponents have plenty of practice since Ken is everywhere.
-_____- I watched the rest of the video. JM basically said what I was thinking LUL
i think for the most part, the trend is accurate, except for the outliers, which include ken, Cammy, deejay, luke, and JP. I think it's always hard to gather info from master rank because so many ppl switch char's there...pick up new char's etc, just trying to rank up everyone they can. this skews the win rates for those char's... but i don't think the win rates lie...Dhalsim is strong...Lily & Rashid are strong, Ryu and Zangief are weak... Usage def plays a part though.
It's not just that you see Ken and Ryu more often, it's that they are much more likely to be flowchart type players too. I'm an AKI main and whenever I play other AKIs I always get a bit excited because I want to see how differently they interpret the character, but most Kens and Ryus and Lukes play sameish.
I think one thing that needs to be accounted for with your analysis of tournament results is simply the talent of the player. MenaRD is going to be a top player no matter who he chose to main in the game. If he had dedicated himself to Dhalsim instead of Luke/Blanka, Dhalsim would likely have much better tournament results than he does. If AngryBird picked Ryu instead of Ken, Ryu would have better tournament results than he does.
Even in tournament play, the player carries the character. The character doesn't carry the player.
Hmmm..... No.
Yes, everyone knows top players are really good. Thats not the point of these charts
This is every streetfighter though usually ken ryu and chun are the most used and dhalsim the least used regardless of the tier placement
"How can you say Lily is bad when she is so high on the ranked win rate" -jmcrofts
Maybe this can finally shut people up complaining about ken. The problem was luke and JP all along.
Makes sense. As a Master Ryu, Dhalsims are the ones I lose to the most.
Do you think you could normalize Winrate by popularity somehow to get an effective meta rank?
You could, but you'd practically be cooking the data until it represented what looked right.
The fact that the matchup chart is so tight is pretty amazing. The best char is 5.138 vs worst 4.885? That means the strongest char is only 2.8% better than average and worst is 2.3% worse than average. And the best char is only 5.2% better than the worst.
If you subtracted the correlation from the play rate the actual differences would be even less
Ken v Ken matchup has 100% winrate. Checkmate liberals
Fchamps ten accounts is influencing this for sure! Lol
Lmao I dont even understand why he has several accounts all playing Dhalsim
great vid
What's the most important thing in ranked matchmaking? It's your mental game, isn't it? So what do you do when you go up against someone whose moves you don't respect? You try to disrespect them and win the mental game. And what happens when your disrespect is met with an optimized looking YOGA! ass whooping? Tourney players are people who take an optimized ass whooping and react to it with "that was cool. Anyway, c. mK? Yeah? Wombo wombo wombo..."
This likely means they'll never buff Dhalsim even if he can't top8 a tourney to save his life, because he's owning the noobs online and exacerbating that would lose more good will than pleasing competitive Dhalsim fans would gain.
Just look at my boy Jamie. Considered one of if not the worst character in the game. Win rate is one of the worst in the game, popular top 7 picked character even in master. And the toobers just kind of ignore Jamie like completely
I been saying this.
I mean this doesn't mean much without ranks included otherwise these pick rates and win rates being low is pretty meaningless and easily explained. Hopefully no one balances around such a silly chart.
Nobody plays Dhalsim. I have 1750 MR I face Dhalsim daily. I dont see how that metric is accurate. But then again it could be my area.
FChamp boosting the Dhalsim statistics
Why is Dhalsim the least played character in SF6 right now?
I think hes the least played char in every game hes in, or at least most of em. Hes just very weird, different, and hard to play
im in the middle of gold rank and got absolutely bodied by a bronze rank dhalsim simply because i had never fought him lmao
jp its zoner character for non zoner player
YOGAYOgaYogayogayogayogayogayoga
People that are still playing Dhalsim, know how to play Dhalsim.
These are loyalist compare to all the Ken and JP tourists lol
Take this down before they nerf dhalsim
Ken downplayers rejoice
Yes we should use actual data to form our opinions. But we need proper data analysis, not this. This was an analysis of a snapshot in time, most of it only being master. We don't actually know the play rate to win rate corrolation because we didn't look at that relationship throughout time or a deep dive of that character by rank and match up. Dhalsim getting popular suddenly can be solved by rolling averages. Tier lists are literally people's emotions, based on anecodate and not data. That's why they never agree. Jmcrofts has multiple examples of old tier lists being inaccurate throughout a games meta. Nor do tier lists consider match ups.
What tier lists are really really good at, is clickbait, and very easy to boot. And every youtuber will fight tooth and nail for that.
People still play this game?
Lots of people
It can't be ken because hes not even top 5. He might be bottom 5 really after all the pros say hes mid. Mid Ken heavy punch drive rush jinrai kicks corner carry into super 80% trash 😭
first
I guess it goes to show just how little the FGC actually knows about balance.
And yet they'll continue to be myopic.
Gamers as a Collective no very little about design a balance.
Maybe. But there was a story, where two guns for different factions in cod or bf, don't remember- had same stats but different sounds. And it made a noticeable difference in k/d statistics. So, no. Gamers know how they feel, and that's a reality.