@@rk13567 Lmao you have severe deficiency in English comprehension. Phreak said a joke/roast. The commenter LIKED the joke. You're here complaining for no reason.
I'd actually really love to see LS or/and IWD in a call with Phreak to discuss this topic I think both sides actually have a lot of respect for each other just they have very different views, would make an awesome video/stream to just go into this for like 2-3 hours Or I'd love for Phreak to come on the Crackdown and discuss it there, not sure if he has any issues with Thorin tho that would prevent him from not accepting an invite I
Excellent video! Totally agree that you can get a lot of information out of soloQ, however I think it is worthwhile to look at reasons why soloQ might not be 100% accurate. Whenever soloQ shows you a trend you should at least look at it, but for example for Rumble there are a couple of reasons why he might be underperforming outside of being bad: Junglers have played less Rumble than Midlaners, since he was changed to be better in the jungle recently (or well actually the changes targetted toplane, but that's neither here nor there). Additionally Rumble jungle is very much FOTM. Combined this means we expect a lot of players that don't have a lot of experience on Rumble to play him jungle, whereas mostly Rumble veterans will be playing him midlane. This is absolutely a plausible explanation for Rumble mid performing better than Rumble jungle. Or at the very least for why the difference is so large. Whenever we are at a situation like this (we have some data for something and either suspect a bias against it or the current "feel" of the champion is against that data) it is useful to default to common sense. You gave a very nice 2 sentence explanation of why Syndra is so good against Tristana in midlane. I can fairly simply see what might be holding Lillia back (fragile clear, fragile in general, very conditional power in her ult). I am not able to give a nice 2 sentence explanation for why Rumble is weak. Whenever I play him I feel like my clear is fast, it is healthy, my objective control is powerful, both through the zoning/teamfighting power of ult and because of the high damage against objectives, dueling against junglers is good and whenever ult is up (which is often thanks to ultimate hunter) his ganks also feel pretty devastating. I just do not see what makes him a bad jungler. Is Rumble maybe just busted in general? Sure, maybe Rumble mid is the single best champion in the game and Rumble Jungle is only top 10. Is Rumble not the best jungler? Also sure, I could see Morgana like you suggested being stronger. But I can't see him being a bad jungler. I can't point to any part of his kit and say "that is what Phreak is talking about, this is what so many other junglers are doing better than him and this achilles' heel is apparently enough to compensate for his strength and drags him down". If you can please point it out to me. TL;DR Good video, I agree with almost all of it, but data can be tainted with biases. So you should almost always only use the data to ask questions ("why not Guinsoo's before Blade"), potential experiment on it and not just take it at facevalue. Until someone can tell me what Rumble jungle isn't good at (question the hypothesis) I have to continue to consider him a good champion.
Imo I think it's more about the fact that mid Rumble can get more gold and exp than jg Rumble can so he can take over the game better from mid. It doesn't mean that jg Rumble is bad, but rather that it's objectively worse than mid Rumble, and why would you intentionally play a champ in it's weaker role unless you're doing some flex picks for better matchups.
@@henshaw33 actually, here's the thing: does Rumble need gold and items to be useful? Or are his base stats already good? I believe that Rumble is one of those champs that doesn't really need any more items to make his ult a threat; he can build cheap items and still do what he can do best: ult in a tf and burn his enemies. He's in that pool of AP champions like Brand or Velk'oz that are allowed to have good base damage but okay scalings. With that being said, yes, more gold is better, but also, less gold isn't that bad, so playing him jungle, especially in good matchups, is a good choice
Phreak spends all of the video showing why it might be bad, and that pros can often be wrong, but IIRC he doesnt spend any time speculating why they're wrong in this instance. With Lillia, maybe the teams spent a lot of time learning her and were slow to drop her. It's expected that teams stick with what they know too long. But when has something newly meta ever been #1 priority in proplay, and was actually bad? I cant think of an example. Maybe it's only good early game, and stage games go later than scrims. It's such an extraordinary claim and makes pros seem less intelligent, so I'm not persuaded without a reason. Phreak doesnt mention scrims at all which is the best data available and he doesnt have access to it. What happens in scrims filters up to the best teams, they dont need to discover everything themselves by experimenting. Phreak is missing a large part of the data used by pros to decide things, so shouldnt feel so confident he's right. He cites a couple of examples where he knew better, to imply we should trust him more than pros, but you could get things wrong most of the time and still have examples where you were right. Maybe he thinks Rumble is #5 best jungler rather than #1, that would be more convincing.
you missed the point of the video. you can’t trust yours or anyone else’s intuition to figure out what is the best or sucks. use the solo w data and you can end up with the conclusion that rumble jungle is bad, but you don’t know why. you can speculate reasons for that, but you can’t know for certain. all that is certain is that rumble jungle definitively is bad.
@@alexgrover904 nowhere in the video did Phreak say that Rumble in the Jungle is bad; instead, he insisted on the fact that while strong, Rumble tends to be even stronger in the midlane and that is rather bad that pro-teams are using Rumble always in the Jungle when he's even stronger if picked mid, probably because tbf, Rumble's clear isn't one of the strongest and mid Rumble has a lot of good matchups, plus he can easily allow his Jungler to gank with his slow on R.
Great video, but Lillia was really not a good example. She did, in fact, get a direct change, in patch 11.6 her ult CD was increased by 20 seconds at all levels, which matches the time of her decline in pro play.
Also Liandry got nerfed early on, after that Seeker's Armguard and Verdant Barrier got nerfed as well. For support build Moonstone and Staff got nerfs.
As an econ and stats nerd I feel this frustration. It's a bit like Moneyball with baseball. The Old Guard, the professionals, the scouts all knew better than these complex statistical models. Then it turns out they didn't and you could build a playoff team for a fraction of the budget when you use data science and use it well. A big problem you find in polling, within industries, within sports, any group really, is herding. If you try something different and it works you look great but part of trying different things is that it won't always work. When it fails you look like an idiot and people get mad because "you didn't do it the right way" even if "the right way" isn't actually the best way. People are risk averse. If you lose but followed the "meta" you don't get nearly as much flak compared to if you lose when you tried something different. As MIB once said: A person is smart. People are dumb, panicky dangerous animals and you know it. Groupthink is real and impedes innovation wherever it's present.
Love that stats based approach and the amount of work you put into a vid like this. With that being said, Rumble jungle feels great especially with a 45 second ult.
Good video, and good to see you explain yourself in a better format than twitter. However you missed the mark on Lillia: Lillia did in fact get a nerf - in patch 11.6 her ult CD was increased by 20 seconds at all levels, and this nerf lines up with her decline in pro play.
The problem is, there is always more context. For example, if renekton is (considered) really strong in pro, teams first pick renekton. To deny renekton-nidalee the red team picks nidalee. To ensure they have an ap jungler to mix damage with the top side blue team picks lillia. Another problem with lillia is that she is considered a difficult matchup into udyr and hecarim, two of the meta junglers of the split. This reduces her presence even if teams want to pick her, because it is necessary to be able to duel the opposition jungler.
Jungle players are only just now learning Rumble, solo laners are more accustomed to the champion. Someone just learning a jungler is easier to punish at higher elo. Would also like to know how frequently Rumble jungle is played at lower elo - maybe only by autofilled solo laners? Lee Sin should not be a jungler, ganks are far less successful/frequent in pro. He'll be outfarmed and be multiple levels down.
Amazing video that states the obvious in a very well thought out way, and it's absolutely baffling how many people just don't get it. Statistics just represent reality, and while there is nuance involved in absorbing the information responsibly, it's still just completely shocking how much mental gymnastics people do to discredit stats if those stats contradict what their gut feelings are.
Disregarding the why behind the reality of the data is just as bad as misrepresenting the data in my eyes. Data is very important yes, it gives you cold, hard facts, yes, but I am genuinely astonished to hear him say you don't need to know the why behind the data. If league solo Q had voice communication I wonder how different the data would become.
this matches my thoughts with the adc meta at worlds. Even though kaisa and trist were not that good in soloq anymore, pro players have spent so much time practicing them that they're probably better on kaisa and trist (looking at you gala) than adcs that have a higher soloq winrate like jinx or something. The same might go for support. Pro supports have been playing engage supports for so long and are so comfortable with it that they probably perform better on them, despite them not being "objectively" optimal
Hey Phreak, I'm an analytical chemist, and I've been watching all of your videos for the past 6 months, and while I *firmly* believe obsessing over builds and patch notes and win rates isn't a good way to get better at league, watching your videos has made me rethink the way I evaluate data and the importance of data presentation, and as such has made me much better at my own job. So thank you very much for making quality videos like this.
The best way to get better at the game is to play against better people. You learn more when you get exposed more because it makes your fundamental flaws more explicit. I agree that optimizing your runes and item build doesn't *inherently* make you a better player, it just makes you climb slightly faster without directly getting better at the game (insofar as decision making and mechanics), but the important thing is that climbing without getting better *does* actually lead to you getting better down the road, because you're playing against better people. For an analogy, look at how pro play scrims generally operate. The best teams in a region tend to be popular scrim partners. Because the weaker teams *know* they will learn *more* by practicing against people who are better than against people who are at the same level.
as soon as people hear "math,statistics,do the work,research" the shy away... It needs to be titled the new op meta easy counterplay free LP GG EZ to get the crowd
Nice video. I hope people learn from this. As a sidenote: This MSI Lillia has been picked 2 times and Xin not at all. At EU Masters Xin looked absolutely busted. Taric counters engage supports and while Nautilus, Alistar and Leona are the three most picked supports Taric doesn't get played at all, even though he is pretty strong currently. Also Rakan is better than the three tanky engage supports since shurelyas is busted.
I am gushing over this video right now. This might be the best video you've ever made Phreak. Just SO many good points made clearly and concisely. (concisely by Phreak standards)
@Phreak Coming from a statistical background myself (majored in economics and political science with a focus on data-driven voting pattern analysis), I love the video and your points. And I do think that at times, other online personalities can overstate the disconnect between solo q WR and pro play. That said, I think that some of the backlash from a community perspective was in part due to the discarding of RNG's stats with rumble jungle. From a modelling perspective, I think that it's a totally justifiable assumption. Even the most basic equation that we could reasonably expect to model a champions pro winrate would have to include a set of terms to denote relative team strength, perhaps in the form of a non-binary dummy variable. That much is common-sense. In such a model, the effect of RNG's wins would be heavily discounted because of the effect of team difference on the expected outcome. However, I do think that's an important point to make clear, and to communicate to others that what you said in your tweet DOES have a reasonable basis in statistical modeling. While the sample size is obviously too small in just group stage to show significant evidence for any kind of model, its important for the community discourse to understand that the underlying assumptions are, in fact, reasonable to make, and that you're not cherry-picking data.
But the main issue with all of this is how the statistics are being used. For example, only taking into direct nerfs for lillia ( and missing some at that ), she had numerous indirect nerfs and when you're trying to see "Why is this data how it is" you don't just exclude massive factors like that. Secondly, a lot of his pro stats would have to include minor regions + academy to even hit the numbers for his analysis, but the highest level of play is almost completely different, this isn't even mentioned, leading to the same "Why is the data like this". I personally think you use solo q data and pro play data to analyse what could potentially be good and try it out. Obviously, some can be ruled off due to the type of champion or things that are specific to pro play and vice versa. But to claim the data suggests it is the case is terrible, we shouldn't ever do that with such little sample size in pro play on an ever-changing meta due to patches. Another thing with the lillia example, what if a champion is specifically good as a counter to another champ that's in meta, i.e. udyr ( which it has been ). It wasn't really mentioned in the video but the Wr is extremely high against udyr (who was insanely meta ) yet it isn't represented in the explanation for the stats. There are some good points and I agree statistics isn't wrong. But some of the examples in the video demonstrate how you use statistics incorrectly to demonstrate a point that isn't necessarily true.
That was a very nice and detailed video, and I appreciate Phreak for going into such depth. However, I like something Jensen Goh said recently on the True Sight podcast. He said something like it's hard to use stats in league to support your arguments, but rather you can use anomalous stats as starting points for investigation. I think pro teams should be doing this instead - instead of saying "rumble jungle has a worse winrate in jungle compared to mid, we should play him mid", I think there should be a step of questioning why that is the case. IWD has put out several arguments that junglers are doing a really poor job of managing heat, for example. Is there some reason mid laners would on average be more able to manage this, and how does that apply to, say, RNG's Wei and Cryin? Then is it worth it to still say we should put it mid after that
Janna took over my ENTIRE soloQ season for sure. Zyra was insane though for the early part of the year before that. All you need to look at to see Pro Players don't know what they are doing half the time is to look at the ways they have built and played Fiora for almost the entire duration of her rework. Any Fiora main will be able to tell you pros either 1 build her wrong or 2 play her wrong or both. Great video though Phreak :0!
A huge thing you miss here is that solo queue players need to learn the champions too, just as pro players. Players who believe Rumble is a broken OP jungler and start practicing with it in Master+ solo queue have to play versus Master+ players playing Nidalee with hundreds of games played already this season. Same applies for Lee Sin. If the pick requires hundreds of games to get competent with, the solo queue winrates currently mean pretty much nothing, as there are no otp rumble junglers or otp lee sin mids with thousand games played. If current average pro players went back in time to like season 4 they could easily beat everyone, as current players have had coaches with more experience in the game, and have themselves got more experience in the game. Same effect can be seen when Riot releases new champions, people need to get good at them for the champions to be good
Overall agree with all of your points, you're spot on. I'm a little less convinced by your examples at the end (e.g. Lee Sin should be a jungler at MSI). You point out yourself (earlier on) that there is massive space between Masters+ and Pro Play. There is no reason why Lee Sin doesn't have this same Ryze effect. A couple of possible reasons are that reduced early skirmishing in Pro means that efficient clear in the early game is much more valuable. Better team-coordination (i.e. Jungler calls out in comms where Lee should be) also means that Lee can't snowball early well.
Hi Phreak, what websites do you use/that you consider to have the best data? There are a ton of different websites out there but in your opinion which one is the best?
while I agree that this more in depth analysis of win rates and trends across elos is way better than the original tweet, it still relies to heavily on raw win rates. It would have been a lot more interesting if you contended more directly with why ppl believe rumble is strong in the jungle (because no one who believes in the pick is quoting win rates or statistics) and debunked it. For instance, I've heard LS, Dom, etc. discuss how his ult damage scales extremely well for a jungler (who is not typically highest level in the game like a mid is), he does have a fast clear, and he does have strong choke control (if you use his ult correctly). So if he has the tools, we have to look beyond win rates and see how he is being used (and also examine overall team compositions because if hes picked in a lost draft anyways then its not worth counting the loss). I think the most appropriate take is "Rumble is too unique of a champion for junglers to pick up and use at the pro level leading up to/at MSI, even if his ceiling is high." It sounds to me like a few ppl smurfed with him during scrims and it caught on but not everyone can play him properly so it doesn't benefit them more than just banning him or picking udyr.
@One Autumn Leaf He said, and I quote "Champ is giga-overrated in that role. He's a good mid laner. He's a bad jungler. Draft better." from his twitter. Phreak believes he is a bad jungler, and a good mid laner. You think phreak believes he's better as a mid laner, not necessarily bad as a jungle.
I note that in low elo aka my elo nobody calls to help break freezes so freezing as a strong dueling top just destroys if you freeze wave. But try freezing in pro and yes you will do well but not be able to freeze for 5 minutes or longer
go watch doms video in response to this then. You will find actual counter points to many of the rumble/lillia stuff. Of course in many areas there is complete agreement. but phreak dropped the ball on some things
That's excellent video Phreak. I really appreciate how you can put the data in context (e.g. Lillia's WR rising even through the peak of Chemtank meta) - everybody can read a graph but the context is what's imporant and what makes or breakes analysis.
Hi Phreak, Great video - however one possible counterpoint regarding Lee and Rumble sololane vs jungle is player mastery / meta presence of the champs. For a jungler that's been staple before frequently like Jarvan if he suddenly came back into the meta and your argument was the same I would say you have a strong point. However the opposite is true here - Rumble has traditionally been a solo laner not a jungler, and Lee has traditionally been a jungler not a solo laner. This suggests that jungle lee will inherently have a higher player mastery than lane lee just because junglers are more likely to have played lee a lot at a point in the past when he was a meta jungler. The same is true for Rumble - he hasn't been a prevalent jungler before whereas he has been a prevalent solo laner so it's more likely that solo laners already have accumulated mastery from playing the champ in the past. Perhaps this goes some way to explaining why jungle rumble looks bad compared to top rumble and why lane lee looks bad compared to jungle lee.
There are outside factors that also cause win rate to fluctuate. Most players I know have played lee. They know his limits they know how his ult works, and he's one been many peoples flex jungler when they get autofilled. You have a champ that is extremely popular, that is now able to be picked and flexed mid, top and jungle. If all 3 of you players on a team can pick it, it becomes extremely hard to counter pick, and leaves many options in draft. Alternatively, rumble is extremely hard to learn, managing heat is something a jungler is going to have a hard time doing while looking at lane states and also being setting up to be ready for fights and managing cooldowns. While many top laners and even some mids have had years of practice on this even if it hasn't been picked, the last time I remember rumble in the jungle was season 3. Many junglers would need much more time and practice as well as having the teams set up better for it. We when a player can operate it though, the possibilities of this champion are much higher than others. Optimizing runes and team comps around it is also going to fluctuate win rate. The issue here is more that rumble should have the same treatment of lee and be an amazing flex pick yet because junglers are going all in on the champ, it seems like other lanes aren't practicing him and he can't be used the same way. While he might be being over played and we need to see his match ups, counters and best comps to use him in, I think he is better than what is shown here, but players should have more practice before using him in pro and banning him if they know the enemy team is proficient on him.
You do realize LS thinks soraka mid is viable. Phreak will knock LS out in a heartbeat, lets also forget LS straight up looks like a walking corpse lmao.
The message of this video was very good and something that definitely needed to be said, but some examples used were wrong or not the best. Like calling Lillia weak. But thats ok, even if you are Phreak and you deserve a lot of respect, you can also be wrong when making statements about whats strong in the game. Thats irrelevant to your overall point about data, which I agree with. Good video, thank you for making it. 👏
@Phreak Can you explain why Shaco is non-existent in pro play? A lot of champs have "Azir curve" in solo q, but when you add coordination of pro play, their strengths evaporate. Yes, this isn't always the case, but it is also why we have to take solo q data with a huge grain of salt.
shaco does drop off in Master and Challenger, so in that regard Phreak's argument stil works. High Elo solo Q teams are still way more coordinated and better at shutting down champions like shaco than low elo solo Q teams. The rise from Iron to Diamond in win rate can be explained with shaco being an early game champion and games becoming shorter as the elo rises.
Hey Phreak, I have been working to try and build this kind of data out for myself just to increase my understanding of all 'possible' picks and best picks at any given time. However I am having difficulty finding where you found the data for the average winrate of a lane by tier. Maybe I am searching incorrectly or not seeing what is in front of my face but can you point me in the right direction?
Ok, so after watching the full video I just want to give my general thoughts on the whole thing. Data analysis for soloq is interesting to study champions, and analyziing how their performance varies depending on player skill is a good idea on paper. However, that doesn't explain the full picture, as soloq playstyle doesn't translate into competitive at all, so champs that can be great at masters+ are not viable in competitive. Some examples of this are junglers such as elise or kha'zix, that have a higher than 50% winrate on masters+, but are rarely picked in competitive. The other point I think is important to adress is the state of jungle. The role is overpowered, and possibly the strongest one in the game. However, the champs on the role aren't. This is signified by udyr being one of the strongest junglers in the game. The champ is obviously strong in early skirmishes and can oneshot your chickens, but as the game progresses it becomes a point and click stun, which can be good, but most other champs in the game are stronger past 20-25 minutes. The point I'm trying to make is that, even if rumble was better in mid than in the jungle, which is questionable as shove wave then roam style can be punished way easier in competitive, there is no point in sending them there. There is a big amount of ap champs that are viable in mid that can offer what rumble gives in teamfights that also scale better (for example orianna in terms of zoning and damage), but meta junglers can't actually offer the utility and teamfight power rumble can give (morgana, as you mentioned in the video, can, so she is one of the best meta junglers). There are no options in jungle that have a insane ult in terms of terrain control on such a low cooldown, can clear pretty fast and can have a 1v1 or 2v2 win at early crabs, meaning rumble is one of the strongest junglers at the moment (and also lillia isn't that bad as she offers pretty much the same with worse early but better scaling).
Very impressive. Phreaks findings strongly correlate with a hunch I've been having for a long time, this is that analysts hired by pro-teams suck at their job. They don't do the technical analysis, which is why they suck.. They simply aren't nerdy enough. Point seems proven!
Hey phreak, I didn't know where else to put this, but do you think that bruiser mythic items are way over-valued in the current meta? I was making a couple spreadsheets for itemization on Rek'Sai top, including gold values, and gold efficiencies, etc. I know Gold Eff can be quite a misleading statistic if looked at in a vacuum, but when going over passives, as well as overall gain from said items, it honestly seems like a terrible pick. What caught my eye the most was that Goredrinker is the only >100% gold eff bruiser mythic, but IMO its still "meh" because you're paying for 150% health regen, which isn't a TERRIBLE thing, but considering that before the item rework everyone was rushing BC/Steraks, and totally ignoring %hp regen, i think that it's not a stat that i'd want to spend my gold on, especially mid game. I found a lot more success going Bork>Black cleaver/Bramble>(Black cleaver)>Spirit Visage/Deaths dance. If the way i put it makes any sense lol.. I think that wits end is also viable as a 5th/6th item, but I also think that Goredrinker is a lot more viable to pick up late game. Obviously this doesn't go for every role, as ADC, Assassin, Support, and Mages have item passives that are pretty crucial to their gameplay/style. You're the only person i know that wouldn't immediately say that this is stupid without having thought of it already, or looking at it. Is there something that's going way over my head that i'm failing to grasp or see? Am i potentially not good enough at the game to push bruiser mythic items to their limits? But if thats the case, then this still holds water for lower elos.. Anyways, if you read this, I greatly appreciate it. Keep dealing Tons of Damage. :)
But Phreak, and I really want your opinion so I hope you read this, am I supposed to look at the winrates of the master players or platinum players like myself? What if the winrates of my characters like Garen are higher? And what if my mmr is high Plat instead of low Plat, what does it all mean?
Intuition becomes more relevant than stats the more complicated something gets. The human brain is much stronger at considering nuance and different contexts than stats are, since stats only measure one thing at a time.
Usually it's the opposite in cases like these where it's near impossible to quantify all the variables. How do you quantify how impactful rumble's ult is in a coordinated teamplay compared to solo q? (just an example) The best way to even get a understanding would be to try it out in scrims.
Great video but honestly not sure how I feel about the lillia stuff in particular, cause lillia did actually get pretty significant changes, she had a direct 20 second nerf at all thanks to ult cd and fairly big ravenous nerfs, don't think it's fair to say lillia has been a bad champ since last year when she got nerfed a lot earlier this year
20 seconds at all ranks on ult, zhonyas (her main item) nerfed, healstone + staff nerf/remove, the mr + ap item nerfed.. Yeah lillia was butchered. Saying she was good half a year earlier is just dumb. Of course she was better.. she was nerfed because of it..
to be fair about league in soloq there is bias that lee jungle player are good at lee and his mechanics while lee mid/top might have just picked him up Same in reverse might be happening to rumble
Hey Phreak if you have taken a stats course you should know that correlation does not imply causation. The data you provided helps to explain a lot of things that are widely accepted regarding champs like Ryze/Azir with high skill expression/viability in pro play. But then to simply, (and quite emotionally) yell that rumble/lee sin's stats fit this same trend therefore they must be pro viable in X role is a logical leap. As others have already pointed out, you neglect to consider that the data can be tainted by Jungle mains first timing FOTM Rumble vs Solo Lane players who have more experience on the champ. Also when you start to consider champions that can be flexed between roles as opposed to champions that are primarily played in a single role you are suddenly introducing so many more variables. I think the "champion scaling with coordinated play" variable is just one important variable to consider but to claim that alone is enough to dictate that rumble should be mid and lee should be jungle is foolish.
Hey Luca, if you have taken a stats course you'd realize that your comment is irrelevant. Phreak is not saying: since X and Y are similar, my argument is correct. He is saying: here is Data set X, what conclusions can be drawn from this data set. And also your comment of "tainted data" is completely silly. Ofc every data set has some kind of external variable. Do we throw out every point of data because of this? No, we put that in our assumptions and gather more data.
@@gabedurno6780 Finding a champion with an "Azir Curve" from Phreak's solo Q dataset is certainly useful. However it only serves to describe a champion in a vacuum. Azir and Ryze while very good in this isolated metric are not being played either. Does that mean teams are foolish for not playing Ryze and Azir mid right now? Of course not because this metric alone isn't what dictates a champion being a meta pick. It may be one component that can make a champion a strong pick but it is not THE definitive metric.
@@gabedurno6780 Well you just pointed out the issue yourself. "what conclusions can be drawn from this data set" Well to argue why his arguments, in this case, are terrible. Groups pro play stats by data that includes minor regions and possibly academy to reach a larger sample size. I think there's a fair argument to be made that the LJL isn't even on the same level as some of the major regions so you shouldn't really be grouping all the regions together and just call it "pro play". He also said "Rumble jungle is too weak for pro play" - in reference to solo q analysis for different tiers, but if ryze has a similar situation yet wasn't "too weak for pro play" then how does his argument hold for rumble? A lot of people will argue the same fundamental arguments they argue for ryze, that there's more to it when you take it over to pro play. You can't make an argument from data whilst excluding some reasons as to why the data exists. In some cases he does it well, and correctly portrays it, in other times he doesn't.
A lot of really good considerations made in this video, I'm really glad Phreak knows about R^2 correlation, I still think that even with all the considerations made in this video there's room for doubt for some of his conclusions, I mean, even after a well executed Statistical experiment, you can never conclude x is good or x is bad, you have to say something like "with a p-value of .001, we have strong evidence in favor of the hypothesis that x is good".
Phreak has got to feel good after watching DK pick Rumble jungle and lose terribly with it in game 5 of the MSI finals. He can sit there and say "I told you so." If only Kkoma just paid attention to basic math instead of gambling his team's run at MSI in the last game.
I like the effort. Idk if its true or not, but I like a different opinion. I do think rumble might be overrated. Dom was constantly saying it was op. But then in games it didnt look like that, and then he claimed it was op but people werent playing it right. To me that sounds weird, cause if it is only op if u play it somewhat perfect, is it really op? Or is it occasionally op, with the right teamcomp? so not a insta first pick everytime
After IWD made it clear how bad your Rumble argument is, he has been constantly reinforcing your point. Watched parts of his stream during MSI and even in the small windows I was there, I heard multiple times his point that junglers are way worse at Rumble due to less practice on him historically. If he believes that AND the win rate data supports that, then saying Rumble should still be prioritised by junglers is just non-sensical. He basically agrees with you but doesn't realise it aha. Great vid Phreak, really well put together script. P. S. I enjoy Doms content and it's not flame, just has a blind spot on this one.
Also here after watching IWD's vid (and Dong Huap's). Between all of these I now know a fair bit more about Rumble *and* Statistical reasoning, I enjoyed this one too! Have to wait and see, but it seems likely Mid Rumble will be the better performer for MSI specifically, especially as pros get better at him.
Since you use lolalytics, I assume you know Rumble jungle one tricks have a ridiculously high winrate. Clearly Rumble isn't a traditional champion junglers are used to, so I think junglers in soloqueue just aren't used to how Rumble works (weird ult, when is overheating optimal, etc). imo the issue with how Rumble is being use in pro play is not that he is weak in the jungle, but that he is not being flexed as he should be. I think he should only be in the jungle in say 50% of the games, and the other times they should be put mid or top PS Phreak if ur reading this check out moonstone senna (don't take staff or ardent they are noob traps for this build). The 11.8 moonstone changes made her healing insane.
I would agree everything you said in this video except one thing Asol's data.... from the bottom of my heart, I honestly dont think Asol is "physcially" that bad like we complained but neither any close to be good. I do not know why different websites tells different stories of Asol's win rates...the most common one op.gg historically shows a high win rate where league of graph and others shows only a 50% He used to have extreme pros and cons that sometimes in somegames are bit unfair to play against, though in soloque with the meta shift since 3 years ago, hes states should be dropping by a lot which he is dropping, and the voice of frustration from Asol players become louder and louder, YET he still has good win rates.
What I’m hearing is when UA-cam’s claim “pro play is a different game than league” it actually means that the pros are just super meta focused even to the point it becomes detrimental
This probably isn't the be all end all, and granted you don't presume that it is, but you're 100% right that solo queue data needs to be better utilized. Hopefully you can change some hearts and minds with this one, 'ppreciate you Phreak.
I love this video and agree with most of it but I do need to add one potential caveat to the analysis. With respect to junglers, I think there can be massive differences to pro. Don't get me wrong, I agree Rumble is better mid. But ganks work so much better in all of solo queue than they do in pro. The coordination, tracking, and punishment to a failed gank are so much higher in pro. Thus, I'm not sure Lee Sin would be a good pro jungler currently. Even GM+ players frequently disrespect potential ganks that pro teams don't.
Just want to say that Alphari, the apparent best pro top laner in NA, got first blooded 5 games in a row against c9 in the NA finals. Pros don't respect ganks nearly as much as you think they do, and laners also have the ability to manipulate waves to force their lane opponents into bad situations, either give up gold and exp or overextend and die to a gank, either way they lose if the pros play correctly.
I think the reason lee sin isn't played as a jungler is because the nature of the game and the differences between pro and soloq. While i dont think hes a good solo laner the reason behind it lies within one of your other points,its because of pro communication and synergy. Additonally, its to secure prio with sustain to ensure you cant get poked off of it. For example, playing something like kindred ideally wants someone to help with getting marks, so you want something to give you that security. Also, with kindreds trends with game time she generally gets stronger, even more so if you factor in marks. Well great! Lee sin has the opposite of this trend generally with an exceptional early game to help ensure kindred gets marks. Now for the reason lee sins trend for soloq winrate is upward i think everyone agrees upon, lee sin isnt garen and he takes practice and skill, but, he also thrives in an environment where he can obtain early kills with his strength. Something a lot less reliable in pro play than soloq. However you can get marks more consistently than kills in a pro game. Overall, i think that soloq winrates and trends that exist with them are useful and good data points, however, not all are accurate. I agree with the idea that rumble should be a mid laner because of his incredible roaming potential with r and his being able to basically bail himself out of any incoming dive. Do i think he should be played jungle? No, morgana is just better and a good udyr can make things happen before hes online. Something he doesnt have to deal with in the mid lane meta. This is overwhelmingly supported by the data you showed. Now putting this all together, i think certain picks are good because they allow other things to come in, pros dont always have time to keep up with the meta, pro winrates are weird because of lack of data points and team skill dispairity, and lastly rumble should be a mid laner. Great video.
people saying "results based analysis" as if that were a bad thing. Just use the data correctly, if you can't then that's your problem. Yes, RNG would have stomped with taric jungle, giving taric jungle 100% WR, but that doesn't mean that WR is a bad metric in ALL cases.
While I agree that statistics are a measure of the real world, they are just that, measurements without any context. For instance just because Janna was picked after Ardent nerfs doesn't mean she was still the best pick maybe just teams were practicing Ardent supports in the run up to worlds so I was the most comfortable pick. Really enjoyed the rest of the video and I replied to your tweet noting the increase in pick rate in Rumble jungle if you wanna take a look twitter.com/JustPaddy0/status/1392809285223079936 but I do agree that Rumble is a better midlaner.
@@snt03 I tried frozen mallet but I liked IBG better due to being able to proc the slow zone by hitting enemies with Qs through allies and whatnot. It also synergized pretty well with the Auto>Q>Auto playstyle. Now both items basically got merged into frostfire so it's gigabusted.
Excellent analysis. I was screaming about how bad lilia was in last worlds as well and statistically she won more only with tank mid and top. I was so confused when teams drafted her without tank mid and top and most of the times the team lost... Keep it up Phreak
very much agree with you phreak, im a br challenger player and i dont even consider myself a good player, i just look at data and use what i think is op and it works wonderfull
TSM will go 0-6 at worlds regardless of comp. Phreak has me dead lmfao.
Edit: I completely misunderstood what OP's comment meant. My bad
@@rk13567 Bro I can only pray you're joking, because I literally quoted the video.
@@rk13567 It was a joke. I said that line verbatim, though.
@@rk13567 Lmao you have severe deficiency in English comprehension. Phreak said a joke/roast. The commenter LIKED the joke. You're here complaining for no reason.
@@DesertFox36 @Phreak Ohhhhhh. Damn, yeah I completely misunderstood what was going on. My bad
I'd actually really love to see LS or/and IWD in a call with Phreak to discuss this topic I think both sides actually have a lot of respect for each other just they have very different views, would make an awesome video/stream to just go into this for like 2-3 hours
Or I'd love for Phreak to come on the Crackdown and discuss it there, not sure if he has any issues with Thorin tho that would prevent him from not accepting an invite
I
brainlet
@@analysttherapist2929 rude
Dom sure. Not LS. LS doesnt know how to discuss these topics like an adult.
@@ZimzamsXD He is literally defending this video on twitter, stop talking out of ur ass please.
I don't really think Dom or LS respect Phreak. And personally I think that says more about them than it does about Phreak.
Excellent video! Totally agree that you can get a lot of information out of soloQ, however I think it is worthwhile to look at reasons why soloQ might not be 100% accurate. Whenever soloQ shows you a trend you should at least look at it, but for example for Rumble there are a couple of reasons why he might be underperforming outside of being bad: Junglers have played less Rumble than Midlaners, since he was changed to be better in the jungle recently (or well actually the changes targetted toplane, but that's neither here nor there).
Additionally Rumble jungle is very much FOTM. Combined this means we expect a lot of players that don't have a lot of experience on Rumble to play him jungle, whereas mostly Rumble veterans will be playing him midlane. This is absolutely a plausible explanation for Rumble mid performing better than Rumble jungle. Or at the very least for why the difference is so large.
Whenever we are at a situation like this (we have some data for something and either suspect a bias against it or the current "feel" of the champion is against that data) it is useful to default to common sense. You gave a very nice 2 sentence explanation of why Syndra is so good against Tristana in midlane. I can fairly simply see what might be holding Lillia back (fragile clear, fragile in general, very conditional power in her ult). I am not able to give a nice 2 sentence explanation for why Rumble is weak. Whenever I play him I feel like my clear is fast, it is healthy, my objective control is powerful, both through the zoning/teamfighting power of ult and because of the high damage against objectives, dueling against junglers is good and whenever ult is up (which is often thanks to ultimate hunter) his ganks also feel pretty devastating. I just do not see what makes him a bad jungler.
Is Rumble maybe just busted in general? Sure, maybe Rumble mid is the single best champion in the game and Rumble Jungle is only top 10. Is Rumble not the best jungler? Also sure, I could see Morgana like you suggested being stronger. But I can't see him being a bad jungler. I can't point to any part of his kit and say "that is what Phreak is talking about, this is what so many other junglers are doing better than him and this achilles' heel is apparently enough to compensate for his strength and drags him down". If you can please point it out to me.
TL;DR Good video, I agree with almost all of it, but data can be tainted with biases. So you should almost always only use the data to ask questions ("why not Guinsoo's before Blade"), potential experiment on it and not just take it at facevalue. Until someone can tell me what Rumble jungle isn't good at (question the hypothesis) I have to continue to consider him a good champion.
Imo I think it's more about the fact that mid Rumble can get more gold and exp than jg Rumble can so he can take over the game better from mid. It doesn't mean that jg Rumble is bad, but rather that it's objectively worse than mid Rumble, and why would you intentionally play a champ in it's weaker role unless you're doing some flex picks for better matchups.
@@henshaw33 actually, here's the thing: does Rumble need gold and items to be useful? Or are his base stats already good?
I believe that Rumble is one of those champs that doesn't really need any more items to make his ult a threat; he can build cheap items and still do what he can do best: ult in a tf and burn his enemies. He's in that pool of AP champions like Brand or Velk'oz that are allowed to have good base damage but okay scalings.
With that being said, yes, more gold is better, but also, less gold isn't that bad, so playing him jungle, especially in good matchups, is a good choice
Phreak spends all of the video showing why it might be bad, and that pros can often be wrong, but IIRC he doesnt spend any time speculating why they're wrong in this instance. With Lillia, maybe the teams spent a lot of time learning her and were slow to drop her. It's expected that teams stick with what they know too long. But when has something newly meta ever been #1 priority in proplay, and was actually bad? I cant think of an example. Maybe it's only good early game, and stage games go later than scrims. It's such an extraordinary claim and makes pros seem less intelligent, so I'm not persuaded without a reason.
Phreak doesnt mention scrims at all which is the best data available and he doesnt have access to it. What happens in scrims filters up to the best teams, they dont need to discover everything themselves by experimenting. Phreak is missing a large part of the data used by pros to decide things, so shouldnt feel so confident he's right. He cites a couple of examples where he knew better, to imply we should trust him more than pros, but you could get things wrong most of the time and still have examples where you were right. Maybe he thinks Rumble is #5 best jungler rather than #1, that would be more convincing.
you missed the point of the video. you can’t trust yours or anyone else’s intuition to figure out what is the best or sucks. use the solo w data and you can end up with the conclusion that rumble jungle is bad, but you don’t know why. you can speculate reasons for that, but you can’t know for certain. all that is certain is that rumble jungle definitively is bad.
@@alexgrover904 nowhere in the video did Phreak say that Rumble in the Jungle is bad; instead, he insisted on the fact that while strong, Rumble tends to be even stronger in the midlane and that is rather bad that pro-teams are using Rumble always in the Jungle when he's even stronger if picked mid, probably because tbf, Rumble's clear isn't one of the strongest and mid Rumble has a lot of good matchups, plus he can easily allow his Jungler to gank with his slow on R.
Great video, but Lillia was really not a good example. She did, in fact, get a direct change, in patch 11.6 her ult CD was increased by 20 seconds at all levels, which matches the time of her decline in pro play.
Also Liandry got nerfed early on, after that Seeker's Armguard and Verdant Barrier got nerfed as well. For support build Moonstone and Staff got nerfs.
Thank you Phreak! Even though I don't play League anymore I still enjoy watching your analytics videos! Have been learning a lot
Why wound Rumble not be good JUNGLE when Riot gave him the skin Rumble in the JUNGLE ?!?!
Speaking Facts
Clearly this is the pro analysis teams have been doing to play Rumble in the jungle, Phreak has rly lost the thread this time SMH
Phreak is like a more serious, nerdy Sips
why tf did no one tell me phreak could be this sexy
bc we all want him to ourselves
@@hallehuckleberry reporting you so I have less competition
@@hockeydavid9749 it will be a battle to the death
As an econ and stats nerd I feel this frustration. It's a bit like Moneyball with baseball. The Old Guard, the professionals, the scouts all knew better than these complex statistical models. Then it turns out they didn't and you could build a playoff team for a fraction of the budget when you use data science and use it well. A big problem you find in polling, within industries, within sports, any group really, is herding. If you try something different and it works you look great but part of trying different things is that it won't always work. When it fails you look like an idiot and people get mad because "you didn't do it the right way" even if "the right way" isn't actually the best way. People are risk averse. If you lose but followed the "meta" you don't get nearly as much flak compared to if you lose when you tried something different. As MIB once said: A person is smart. People are dumb, panicky dangerous animals and you know it. Groupthink is real and impedes innovation wherever it's present.
Love that stats based approach and the amount of work you put into a vid like this. With that being said, Rumble jungle feels great especially with a 45 second ult.
You can have that in Mid too, tho. Thats no reason he should be in jungle
I remember games when I had 22second cooldown on rumble r and the ult is just too strong to get even close to short cds
He should go wherever his ultimate can be paired with hard cc.
Feels good doesn't mean you win the game more.
this was a really cool video. thanks for making it! Also, nice mandatory tons of dmg
Good video, and good to see you explain yourself in a better format than twitter. However you missed the mark on Lillia: Lillia did in fact get a nerf - in patch 11.6 her ult CD was increased by 20 seconds at all levels, and this nerf lines up with her decline in pro play.
also a ton of other changes happened around other junglers which entirely shifted the meta. Stats don't lie but they can be used in a wrong way.
The problem is, there is always more context.
For example, if renekton is (considered) really strong in pro, teams first pick renekton.
To deny renekton-nidalee the red team picks nidalee.
To ensure they have an ap jungler to mix damage with the top side blue team picks lillia.
Another problem with lillia is that she is considered a difficult matchup into udyr and hecarim, two of the meta junglers of the split. This reduces her presence even if teams want to pick her, because it is necessary to be able to duel the opposition jungler.
Jungle players are only just now learning Rumble, solo laners are more accustomed to the champion. Someone just learning a jungler is easier to punish at higher elo. Would also like to know how frequently Rumble jungle is played at lower elo - maybe only by autofilled solo laners?
Lee Sin should not be a jungler, ganks are far less successful/frequent in pro. He'll be outfarmed and be multiple levels down.
Wait, this channel is criminally underrated. 10/10 video.
I think the low lans lee wr ist due to many people practising and testing him in high elo and the already trained Lees stickig to jungle
Couldn’t the same argument be applied to rumble in the solo lanes?
@@pavango3812 Yes with the fact that rumbles in the jungle are practicing him and learning him while mid laners have played rumble in previous metas
Amazing video that states the obvious in a very well thought out way, and it's absolutely baffling how many people just don't get it. Statistics just represent reality, and while there is nuance involved in absorbing the information responsibly, it's still just completely shocking how much mental gymnastics people do to discredit stats if those stats contradict what their gut feelings are.
Disregarding the why behind the reality of the data is just as bad as misrepresenting the data in my eyes. Data is very important yes, it gives you cold, hard facts, yes, but I am genuinely astonished to hear him say you don't need to know the why behind the data.
If league solo Q had voice communication I wonder how different the data would become.
"Don't be stupid time is limited" reminds me of my high school math teacher
"...and that's a problem with you."
Phreak comin' out swinging. I'm here for it.
this matches my thoughts with the adc meta at worlds. Even though kaisa and trist were not that good in soloq anymore, pro players have spent so much time practicing them that they're probably better on kaisa and trist (looking at you gala) than adcs that have a higher soloq winrate like jinx or something. The same might go for support. Pro supports have been playing engage supports for so long and are so comfortable with it that they probably perform better on them, despite them not being "objectively" optimal
Hey Phreak, I'm an analytical chemist, and I've been watching all of your videos for the past 6 months, and while I *firmly* believe obsessing over builds and patch notes and win rates isn't a good way to get better at league, watching your videos has made me rethink the way I evaluate data and the importance of data presentation, and as such has made me much better at my own job. So thank you very much for making quality videos like this.
The best way to get better at the game is to play against better people. You learn more when you get exposed more because it makes your fundamental flaws more explicit. I agree that optimizing your runes and item build doesn't *inherently* make you a better player, it just makes you climb slightly faster without directly getting better at the game (insofar as decision making and mechanics), but the important thing is that climbing without getting better *does* actually lead to you getting better down the road, because you're playing against better people.
For an analogy, look at how pro play scrims generally operate. The best teams in a region tend to be popular scrim partners. Because the weaker teams *know* they will learn *more* by practicing against people who are better than against people who are at the same level.
Back in the day this man released Blitzcrank as a jungler
thats why phreak is still an underrated content creator and people have still not figured it out completely!
as soon as people hear "math,statistics,do the work,research" the shy away... It needs to be titled the new op meta easy counterplay free LP GG EZ to get the crowd
and more capslock, cause that shows the true skillz from all the streamers/youtubers - phreak simply goes and produces high quality content with style
Nice video. I hope people learn from this.
As a sidenote: This MSI Lillia has been picked 2 times and Xin not at all. At EU Masters Xin looked absolutely busted. Taric counters engage supports and while Nautilus, Alistar and Leona are the three most picked supports Taric doesn't get played at all, even though he is pretty strong currently. Also Rakan is better than the three tanky engage supports since shurelyas is busted.
*econ major huffs car exhaust and starts screaming random league stats*
I am gushing over this video right now. This might be the best video you've ever made Phreak. Just SO many good points made clearly and concisely.
(concisely by Phreak standards)
@Phreak Coming from a statistical background myself (majored in economics and political science with a focus on data-driven voting pattern analysis), I love the video and your points. And I do think that at times, other online personalities can overstate the disconnect between solo q WR and pro play.
That said, I think that some of the backlash from a community perspective was in part due to the discarding of RNG's stats with rumble jungle. From a modelling perspective, I think that it's a totally justifiable assumption. Even the most basic equation that we could reasonably expect to model a champions pro winrate would have to include a set of terms to denote relative team strength, perhaps in the form of a non-binary dummy variable. That much is common-sense. In such a model, the effect of RNG's wins would be heavily discounted because of the effect of team difference on the expected outcome.
However, I do think that's an important point to make clear, and to communicate to others that what you said in your tweet DOES have a reasonable basis in statistical modeling. While the sample size is obviously too small in just group stage to show significant evidence for any kind of model, its important for the community discourse to understand that the underlying assumptions are, in fact, reasonable to make, and that you're not cherry-picking data.
But the main issue with all of this is how the statistics are being used.
For example, only taking into direct nerfs for lillia ( and missing some at that ), she had numerous indirect nerfs and when you're trying to see "Why is this data how it is" you don't just exclude massive factors like that.
Secondly, a lot of his pro stats would have to include minor regions + academy to even hit the numbers for his analysis, but the highest level of play is almost completely different, this isn't even mentioned, leading to the same "Why is the data like this".
I personally think you use solo q data and pro play data to analyse what could potentially be good and try it out. Obviously, some can be ruled off due to the type of champion or things that are specific to pro play and vice versa. But to claim the data suggests it is the case is terrible, we shouldn't ever do that with such little sample size in pro play on an ever-changing meta due to patches. Another thing with the lillia example, what if a champion is specifically good as a counter to another champ that's in meta, i.e. udyr ( which it has been ). It wasn't really mentioned in the video but the Wr is extremely high against udyr (who was insanely meta ) yet it isn't represented in the explanation for the stats.
There are some good points and I agree statistics isn't wrong. But some of the examples in the video demonstrate how you use statistics incorrectly to demonstrate a point that isn't necessarily true.
That was a very nice and detailed video, and I appreciate Phreak for going into such depth. However, I like something Jensen Goh said recently on the True Sight podcast. He said something like it's hard to use stats in league to support your arguments, but rather you can use anomalous stats as starting points for investigation. I think pro teams should be doing this instead - instead of saying "rumble jungle has a worse winrate in jungle compared to mid, we should play him mid", I think there should be a step of questioning why that is the case. IWD has put out several arguments that junglers are doing a really poor job of managing heat, for example. Is there some reason mid laners would on average be more able to manage this, and how does that apply to, say, RNG's Wei and Cryin? Then is it worth it to still say we should put it mid after that
Janna took over my ENTIRE soloQ season for sure.
Zyra was insane though for the early part of the year before that.
All you need to look at to see Pro Players don't know what they are doing half the time is to look at the ways they have built and played Fiora for almost the entire duration of her rework. Any Fiora main will be able to tell you pros either 1 build her wrong or 2 play her wrong or both.
Great video though Phreak :0!
love watching you cast phreak :^)
A huge thing you miss here is that solo queue players need to learn the champions too, just as pro players. Players who believe Rumble is a broken OP jungler and start practicing with it in Master+ solo queue have to play versus Master+ players playing Nidalee with hundreds of games played already this season. Same applies for Lee Sin. If the pick requires hundreds of games to get competent with, the solo queue winrates currently mean pretty much nothing, as there are no otp rumble junglers or otp lee sin mids with thousand games played. If current average pro players went back in time to like season 4 they could easily beat everyone, as current players have had coaches with more experience in the game, and have themselves got more experience in the game. Same effect can be seen when Riot releases new champions, people need to get good at them for the champions to be good
Overall agree with all of your points, you're spot on. I'm a little less convinced by your examples at the end (e.g. Lee Sin should be a jungler at MSI). You point out yourself (earlier on) that there is massive space between Masters+ and Pro Play. There is no reason why Lee Sin doesn't have this same Ryze effect. A couple of possible reasons are that reduced early skirmishing in Pro means that efficient clear in the early game is much more valuable. Better team-coordination (i.e. Jungler calls out in comms where Lee should be) also means that Lee can't snowball early well.
Hi Phreak, what websites do you use/that you consider to have the best data? There are a ton of different websites out there but in your opinion which one is the best?
while I agree that this more in depth analysis of win rates and trends across elos is way better than the original tweet, it still relies to heavily on raw win rates. It would have been a lot more interesting if you contended more directly with why ppl believe rumble is strong in the jungle (because no one who believes in the pick is quoting win rates or statistics) and debunked it.
For instance, I've heard LS, Dom, etc. discuss how his ult damage scales extremely well for a jungler (who is not typically highest level in the game like a mid is), he does have a fast clear, and he does have strong choke control (if you use his ult correctly). So if he has the tools, we have to look beyond win rates and see how he is being used (and also examine overall team compositions because if hes picked in a lost draft anyways then its not worth counting the loss).
I think the most appropriate take is "Rumble is too unique of a champion for junglers to pick up and use at the pro level leading up to/at MSI, even if his ceiling is high." It sounds to me like a few ppl smurfed with him during scrims and it caught on but not everyone can play him properly so it doesn't benefit them more than just banning him or picking udyr.
@One Autumn Leaf He said, and I quote "Champ is giga-overrated in that role. He's a good mid laner. He's a bad jungler. Draft better." from his twitter.
Phreak believes he is a bad jungler, and a good mid laner.
You think phreak believes he's better as a mid laner, not necessarily bad as a jungle.
Really interesting video and topic.
Phreak: Jungle Rumble is bad.
Pro teams: How about we still pick him.
I note that in low elo aka my elo nobody calls to help break freezes so freezing as a strong dueling top just destroys if you freeze wave. But try freezing in pro and yes you will do well but not be able to freeze for 5 minutes or longer
Great content! Maybe I'm getting old but I just love being able to listen to an adult talking stats instead of edgy judgements, thanks a lot!
go watch doms video in response to this then. You will find actual counter points to many of the rumble/lillia stuff. Of course in many areas there is complete agreement. but phreak dropped the ball on some things
That's excellent video Phreak. I really appreciate how you can put the data in context (e.g. Lillia's WR rising even through the peak of Chemtank meta) - everybody can read a graph but the context is what's imporant and what makes or breakes analysis.
FWIW that graph is pro pick/ban rate, not win rate.
Hi Phreak,
Great video - however one possible counterpoint regarding Lee and Rumble sololane vs jungle is player mastery / meta presence of the champs. For a jungler that's been staple before frequently like Jarvan if he suddenly came back into the meta and your argument was the same I would say you have a strong point.
However the opposite is true here - Rumble has traditionally been a solo laner not a jungler, and Lee has traditionally been a jungler not a solo laner. This suggests that jungle lee will inherently have a higher player mastery than lane lee just because junglers are more likely to have played lee a lot at a point in the past when he was a meta jungler. The same is true for Rumble - he hasn't been a prevalent jungler before whereas he has been a prevalent solo laner so it's more likely that solo laners already have accumulated mastery from playing the champ in the past.
Perhaps this goes some way to explaining why jungle rumble looks bad compared to top rumble and why lane lee looks bad compared to jungle lee.
There are outside factors that also cause win rate to fluctuate. Most players I know have played lee. They know his limits they know how his ult works, and he's one been many peoples flex jungler when they get autofilled.
You have a champ that is extremely popular, that is now able to be picked and flexed mid, top and jungle. If all 3 of you players on a team can pick it, it becomes extremely hard to counter pick, and leaves many options in draft.
Alternatively, rumble is extremely hard to learn, managing heat is something a jungler is going to have a hard time doing while looking at lane states and also being setting up to be ready for fights and managing cooldowns. While many top laners and even some mids have had years of practice on this even if it hasn't been picked, the last time I remember rumble in the jungle was season 3. Many junglers would need much more time and practice as well as having the teams set up better for it. We when a player can operate it though, the possibilities of this champion are much higher than others. Optimizing runes and team comps around it is also going to fluctuate win rate.
The issue here is more that rumble should have the same treatment of lee and be an amazing flex pick yet because junglers are going all in on the champ, it seems like other lanes aren't practicing him and he can't be used the same way. While he might be being over played and we need to see his match ups, counters and best comps to use him in, I think he is better than what is shown here, but players should have more practice before using him in pro and banning him if they know the enemy team is proficient on him.
Can we just put Phreak and LS in a boxing ring already
that would be too one sided
You do realize LS thinks soraka mid is viable. Phreak will knock LS out in a heartbeat, lets also forget LS straight up looks like a walking corpse lmao.
The message of this video was very good and something that definitely needed to be said, but some examples used were wrong or not the best. Like calling Lillia weak. But thats ok, even if you are Phreak and you deserve a lot of respect, you can also be wrong when making statements about whats strong in the game. Thats irrelevant to your overall point about data, which I agree with. Good video, thank you for making it. 👏
noooo Phreak! You are giving away my secrets! Numbers!
Huh only 3 min in and sounds like Phreak has watched Jatt's Crackdown episode and Bjerg's Dr. k video. Neat.
Probably face check too
@Phreak Can you explain why Shaco is non-existent in pro play? A lot of champs have "Azir curve" in solo q, but when you add coordination of pro play, their strengths evaporate. Yes, this isn't always the case, but it is also why we have to take solo q data with a huge grain of salt.
shaco does drop off in Master and Challenger, so in that regard Phreak's argument stil works. High Elo solo Q teams are still way more coordinated and better at shutting down champions like shaco than low elo solo Q teams. The rise from Iron to Diamond in win rate can be explained with shaco being an early game champion and games becoming shorter as the elo rises.
Hey Phreak, I have been working to try and build this kind of data out for myself just to increase my understanding of all 'possible' picks and best picks at any given time. However I am having difficulty finding where you found the data for the average winrate of a lane by tier. Maybe I am searching incorrectly or not seeing what is in front of my face but can you point me in the right direction?
Ok, so after watching the full video I just want to give my general thoughts on the whole thing. Data analysis for soloq is interesting to study champions, and analyziing how their performance varies depending on player skill is a good idea on paper. However, that doesn't explain the full picture, as soloq playstyle doesn't translate into competitive at all, so champs that can be great at masters+ are not viable in competitive. Some examples of this are junglers such as elise or kha'zix, that have a higher than 50% winrate on masters+, but are rarely picked in competitive. The other point I think is important to adress is the state of jungle. The role is overpowered, and possibly the strongest one in the game. However, the champs on the role aren't. This is signified by udyr being one of the strongest junglers in the game. The champ is obviously strong in early skirmishes and can oneshot your chickens, but as the game progresses it becomes a point and click stun, which can be good, but most other champs in the game are stronger past 20-25 minutes. The point I'm trying to make is that, even if rumble was better in mid than in the jungle, which is questionable as shove wave then roam style can be punished way easier in competitive, there is no point in sending them there. There is a big amount of ap champs that are viable in mid that can offer what rumble gives in teamfights that also scale better (for example orianna in terms of zoning and damage), but meta junglers can't actually offer the utility and teamfight power rumble can give (morgana, as you mentioned in the video, can, so she is one of the best meta junglers). There are no options in jungle that have a insane ult in terms of terrain control on such a low cooldown, can clear pretty fast and can have a 1v1 or 2v2 win at early crabs, meaning rumble is one of the strongest junglers at the moment (and also lillia isn't that bad as she offers pretty much the same with worse early but better scaling).
I agree with everything you said, however no guarantees on not being stupid
I absolutely loved this video, just fantastic all around and really opened my eyes on some things
Very impressive. Phreaks findings strongly correlate with a hunch I've been having for a long time, this is that analysts hired by pro-teams suck at their job. They don't do the technical analysis, which is why they suck.. They simply aren't nerdy enough. Point seems proven!
Hey Phreak, what website do you get your stats from?
Lolalytics is the site he usually talks about in his streams :)
@@Chaoterich Thank you!
Good vid. I'm too dumb to understand half of it but I like your moustache.
I have to ask, did you purposefully misspell "stupid" at 6:18 for the joke or was it unintentional?
Hey phreak, I didn't know where else to put this, but do you think that bruiser mythic items are way over-valued in the current meta?
I was making a couple spreadsheets for itemization on Rek'Sai top, including gold values, and gold efficiencies, etc. I know Gold Eff can be quite a misleading statistic if looked at in a vacuum, but when going over passives, as well as overall gain from said items, it honestly seems like a terrible pick.
What caught my eye the most was that Goredrinker is the only >100% gold eff bruiser mythic, but IMO its still "meh" because you're paying for 150% health regen, which isn't a TERRIBLE thing, but considering that before the item rework everyone was rushing BC/Steraks, and totally ignoring %hp regen, i think that it's not a stat that i'd want to spend my gold on, especially mid game.
I found a lot more success going Bork>Black cleaver/Bramble>(Black cleaver)>Spirit Visage/Deaths dance. If the way i put it makes any sense lol.. I think that wits end is also viable as a 5th/6th item, but I also think that Goredrinker is a lot more viable to pick up late game. Obviously this doesn't go for every role, as ADC, Assassin, Support, and Mages have item passives that are pretty crucial to their gameplay/style.
You're the only person i know that wouldn't immediately say that this is stupid without having thought of it already, or looking at it.
Is there something that's going way over my head that i'm failing to grasp or see? Am i potentially not good enough at the game to push bruiser mythic items to their limits? But if thats the case, then this still holds water for lower elos..
Anyways, if you read this, I greatly appreciate it. Keep dealing Tons of Damage. :)
Great video. Sound like the Burry of league lol.
LS and enchanter gang approves of this video
nah, I don't approve of this results based analysis andy
what sources can you find to compare champion win rates based on item/rune picks
Can you put the links to where u got the stats from? So we can do our own research
@Brendan Yu the thing is some of those # dont add up so I want to know where @phreak got his #s from
"You" do you mean LS and IWD ? Sure seems like it to me! 🤣 🤣 🤣
Interesting, i think in these pro game position played means slightly less because the coordination is so high.
But Phreak, and I really want your opinion so I hope you read this, am I supposed to look at the winrates of the master players or platinum players like myself? What if the winrates of my characters like Garen are higher? And what if my mmr is high Plat instead of low Plat, what does it all mean?
Tbh the entire later use of it and the obvious winrates as an udyr counter kinda defeats these points
The game just gets more and more complex with time as well so trusting anyones intuition becomes worse and worse.
Intuition becomes more relevant than stats the more complicated something gets. The human brain is much stronger at considering nuance and different contexts than stats are, since stats only measure one thing at a time.
Usually it's the opposite in cases like these where it's near impossible to quantify all the variables. How do you quantify how impactful rumble's ult is in a coordinated teamplay compared to solo q? (just an example) The best way to even get a understanding would be to try it out in scrims.
@@ClickHere4Naruto I only meant it gets worse to their intuition compared to today, but you're right.
Great video but honestly not sure how I feel about the lillia stuff in particular, cause lillia did actually get pretty significant changes, she had a direct 20 second nerf at all thanks to ult cd and fairly big ravenous nerfs, don't think it's fair to say lillia has been a bad champ since last year when she got nerfed a lot earlier this year
20 seconds at all ranks on ult, zhonyas (her main item) nerfed, healstone + staff nerf/remove, the mr + ap item nerfed.. Yeah lillia was butchered. Saying she was good half a year earlier is just dumb. Of course she was better.. she was nerfed because of it..
damn phreak had to let out some pent up anger in this one
to be fair about league in soloq there is bias that lee jungle player are good at lee and his mechanics while lee mid/top might have just picked him up
Same in reverse might be happening to rumble
Phreak presents: There I Said It
Hi Phreak, would it be possible for you to link the statistics you've used for champion win rates by rank?
Hey Phreak if you have taken a stats course you should know that correlation does not imply causation.
The data you provided helps to explain a lot of things that are widely accepted regarding champs like Ryze/Azir with high skill expression/viability in pro play. But then to simply, (and quite emotionally) yell that rumble/lee sin's stats fit this same trend therefore they must be pro viable in X role is a logical leap.
As others have already pointed out, you neglect to consider that the data can be tainted by Jungle mains first timing FOTM Rumble vs Solo Lane players who have more experience on the champ. Also when you start to consider champions that can be flexed between roles as opposed to champions that are primarily played in a single role you are suddenly introducing so many more variables. I think the "champion scaling with coordinated play" variable is just one important variable to consider but to claim that alone is enough to dictate that rumble should be mid and lee should be jungle is foolish.
Hey Luca, if you have taken a stats course you'd realize that your comment is irrelevant. Phreak is not saying: since X and Y are similar, my argument is correct. He is saying: here is Data set X, what conclusions can be drawn from this data set. And also your comment of "tainted data" is completely silly. Ofc every data set has some kind of external variable. Do we throw out every point of data because of this? No, we put that in our assumptions and gather more data.
@@gabedurno6780
Finding a champion with an "Azir Curve" from Phreak's solo Q dataset is certainly useful. However it only serves to describe a champion in a vacuum.
Azir and Ryze while very good in this isolated metric are not being played either. Does that mean teams are foolish for not playing Ryze and Azir mid right now? Of course not because this metric alone isn't what dictates a champion being a meta pick. It may be one component that can make a champion a strong pick but it is not THE definitive metric.
@@gabedurno6780 Well you just pointed out the issue yourself. "what conclusions can be drawn from this data set"
Well to argue why his arguments, in this case, are terrible. Groups pro play stats by data that includes minor regions and possibly academy to reach a larger sample size. I think there's a fair argument to be made that the LJL isn't even on the same level as some of the major regions so you shouldn't really be grouping all the regions together and just call it "pro play".
He also said "Rumble jungle is too weak for pro play" - in reference to solo q analysis for different tiers, but if ryze has a similar situation yet wasn't "too weak for pro play" then how does his argument hold for rumble?
A lot of people will argue the same fundamental arguments they argue for ryze, that there's more to it when you take it over to pro play.
You can't make an argument from data whilst excluding some reasons as to why the data exists. In some cases he does it well, and correctly portrays it, in other times he doesn't.
A lot of really good considerations made in this video, I'm really glad Phreak knows about R^2 correlation, I still think that even with all the considerations made in this video there's room for doubt for some of his conclusions, I mean, even after a well executed Statistical experiment, you can never conclude x is good or x is bad, you have to say something like "with a p-value of .001, we have strong evidence in favor of the hypothesis that x is good".
Tank Senna is cool I was trying Frostfire on Ezreal
tank in this game ? good joke xD
Idk Phreak, I think I need IWD and LS to feed me my rebuttal counterpoints before I can understand why this is wrong.
Phreak has got to feel good after watching DK pick Rumble jungle and lose terribly with it in game 5 of the MSI finals. He can sit there and say "I told you so." If only Kkoma just paid attention to basic math instead of gambling his team's run at MSI in the last game.
What is your LoL Data site of choice? Often times I want to look up non-meta things and the sample size is just super small.
Lolalytics!
Lolalytics is the best if you want to deep dive into the stats
@@Chaoterich @Sepher Thanks!
I like the effort. Idk if its true or not, but I like a different opinion. I do think rumble might be overrated. Dom was constantly saying it was op. But then in games it didnt look like that, and then he claimed it was op but people werent playing it right. To me that sounds weird, cause if it is only op if u play it somewhat perfect, is it really op? Or is it occasionally op, with the right teamcomp? so not a insta first pick everytime
I mean it depends, playing it right doesn't necessarily mean perfect, it can be a lot simpler than that, or somewhere in between.
After IWD made it clear how bad your Rumble argument is, he has been constantly reinforcing your point.
Watched parts of his stream during MSI and even in the small windows I was there, I heard multiple times his point that junglers are way worse at Rumble due to less practice on him historically.
If he believes that AND the win rate data supports that, then saying Rumble should still be prioritised by junglers is just non-sensical.
He basically agrees with you but doesn't realise it aha.
Great vid Phreak, really well put together script.
P. S. I enjoy Doms content and it's not flame, just has a blind spot on this one.
Also here after watching IWD's vid (and Dong Huap's). Between all of these I now know a fair bit more about Rumble *and* Statistical reasoning, I enjoyed this one too!
Have to wait and see, but it seems likely Mid Rumble will be the better performer for MSI specifically, especially as pros get better at him.
@@AEVAN00B Rumble has been mid for awhile and pro's are good at him.
Since you use lolalytics, I assume you know Rumble jungle one tricks have a ridiculously high winrate. Clearly Rumble isn't a traditional champion junglers are used to, so I think junglers in soloqueue just aren't used to how Rumble works (weird ult, when is overheating optimal, etc).
imo the issue with how Rumble is being use in pro play is not that he is weak in the jungle, but that he is not being flexed as he should be. I think he should only be in the jungle in say 50% of the games, and the other times they should be put mid or top
PS Phreak if ur reading this check out moonstone senna (don't take staff or ardent they are noob traps for this build). The 11.8 moonstone changes made her healing insane.
I would agree everything you said in this video except one thing
Asol's data....
from the bottom of my heart, I honestly dont think Asol is "physcially" that bad like we complained but neither any close to be good.
I do not know why different websites tells different stories of Asol's win rates...the most common one op.gg historically shows a high win rate where league of graph and others shows only a 50%
He used to have extreme pros and cons that sometimes in somegames are bit unfair to play against, though in soloque with the meta shift since 3 years ago, hes states should be dropping by a lot which he is dropping, and the voice of frustration from Asol players become louder and louder, YET he still has good win rates.
What I’m hearing is when UA-cam’s claim “pro play is a different game than league” it actually means that the pros are just super meta focused even to the point it becomes detrimental
This probably isn't the be all end all, and granted you don't presume that it is, but you're 100% right that solo queue data needs to be better utilized. Hopefully you can change some hearts and minds with this one, 'ppreciate you Phreak.
Godlike video, Phreak, thanks for doing such an in-depth look at this data!
Great video Phreak. Keep up the good content
You can almost see the progression of phreak talking faster and faster per video nowdays lmao
Also he has a beard here, hard to tell if he looks hotter or not tho. Tough call.
What a good video! very well expleained, it is cool to see how calm can be Phreak be while explaining the topic!
I love this video and agree with most of it but I do need to add one potential caveat to the analysis. With respect to junglers, I think there can be massive differences to pro. Don't get me wrong, I agree Rumble is better mid. But ganks work so much better in all of solo queue than they do in pro. The coordination, tracking, and punishment to a failed gank are so much higher in pro. Thus, I'm not sure Lee Sin would be a good pro jungler currently. Even GM+ players frequently disrespect potential ganks that pro teams don't.
Just want to say that Alphari, the apparent best pro top laner in NA, got first blooded 5 games in a row against c9 in the NA finals. Pros don't respect ganks nearly as much as you think they do, and laners also have the ability to manipulate waves to force their lane opponents into bad situations, either give up gold and exp or overextend and die to a gank, either way they lose if the pros play correctly.
I think the reason lee sin isn't played as a jungler is because the nature of the game and the differences between pro and soloq. While i dont think hes a good solo laner the reason behind it lies within one of your other points,its because of pro communication and synergy. Additonally, its to secure prio with sustain to ensure you cant get poked off of it. For example, playing something like kindred ideally wants someone to help with getting marks, so you want something to give you that security. Also, with kindreds trends with game time she generally gets stronger, even more so if you factor in marks. Well great! Lee sin has the opposite of this trend generally with an exceptional early game to help ensure kindred gets marks. Now for the reason lee sins trend for soloq winrate is upward i think everyone agrees upon, lee sin isnt garen and he takes practice and skill, but, he also thrives in an environment where he can obtain early kills with his strength. Something a lot less reliable in pro play than soloq. However you can get marks more consistently than kills in a pro game. Overall, i think that soloq winrates and trends that exist with them are useful and good data points, however, not all are accurate. I agree with the idea that rumble should be a mid laner because of his incredible roaming potential with r and his being able to basically bail himself out of any incoming dive. Do i think he should be played jungle? No, morgana is just better and a good udyr can make things happen before hes online. Something he doesnt have to deal with in the mid lane meta. This is overwhelmingly supported by the data you showed. Now putting this all together, i think certain picks are good because they allow other things to come in, pros dont always have time to keep up with the meta, pro winrates are weird because of lack of data points and team skill dispairity, and lastly rumble should be a mid laner. Great video.
Great video
people saying "results based analysis" as if that were a bad thing. Just use the data correctly, if you can't then that's your problem.
Yes, RNG would have stomped with taric jungle, giving taric jungle 100% WR, but that doesn't mean that WR is a bad metric in ALL cases.
Some might.
While I agree that statistics are a measure of the real world, they are just that, measurements without any context. For instance just because Janna was picked after Ardent nerfs doesn't mean she was still the best pick maybe just teams were practicing Ardent supports in the run up to worlds so I was the most comfortable pick. Really enjoyed the rest of the video and I replied to your tweet noting the increase in pick rate in Rumble jungle if you wanna take a look twitter.com/JustPaddy0/status/1392809285223079936 but I do agree that Rumble is a better midlaner.
Ads in the vid cause riot doesn't pay him enough
And here I was playing Iceborn Gauntlet / BC senna in 2019 lmao
not frozen mallet :d
@@snt03 I tried frozen mallet but I liked IBG better due to being able to proc the slow zone by hitting enemies with Qs through allies and whatnot. It also synergized pretty well with the Auto>Q>Auto playstyle. Now both items basically got merged into frostfire so it's gigabusted.
Awesome video! Keep it up!
"Draft better"
Mic dropped
Just finished the video, very interesting and informative!
Excellent analysis. I was screaming about how bad lilia was in last worlds as well and statistically she won more only with tank mid and top. I was so confused when teams drafted her without tank mid and top and most of the times the team lost...
Keep it up Phreak
very much agree with you phreak, im a br challenger player and i dont even consider myself a good player, i just look at data and use what i think is op and it works wonderfull
Very interesting video! Would love more statistics content like this, especially regarding late adoption of picks. Super cool stuff.
These mid rolls are killing me man, gotta go get my phone every 5 min
Bro when I read your tweet I assumed you were right.