Jun 28, 2014

Win Rates - What Do They Really Mean?

34 comments

Introduction

There's been a lot of discussion on the tier lists regarding win rates and how much they should be taken into consideration when picking champions for solo queue.

This article will discuss what win rates do and do not mean, and hopefully also shed some light on the thought process behind placing certain champions on the list.
Summary: With enough playtime, a high win rate on a champion indicates *something* generally goes right, and a low win rate indicates that whatever is happening in games with this champion isn't in their favor.

A low win rate may mean a champion is being played incorrectly and is stronger than his win rate indicates, but a top 25 win rate (with 120 champions) almost always indicates that a champion is at least a good candidate for Tier 1 or high Tier 2 if the sample size is large enough.

Background

The current range of champion win rates lies almost entirely within the range of 45% to 55%. This means that while there are certain outliers, the majority of champions are close to balanced when it comes to players of similar skill levels and any unknown variables that change from game to game.

Definitions:

  •          “low win rate” - champions in the lower 25 champion range. This currently comprises of champions at 47% and lower at the time of writing this article.
  •          “high win rate” - champions in the upper 25 champion range. This currently makes up champions at 52% or higher.

Note on Statistics:

For those of you who are unfamiliar with statistics, the reliability of the underlying statistics are always based off of a “good” population size. This means that if a champion isn’t played frequently enough, or is a “niche” pick (see below), then the statistics for that champion will be skewed.

Nerfplz's tier lists generally use data from Lolking.net’s page, and thus uses the data that Riot makes available over a broad spectrum of games. However, it doesn’t account for the following:
  1.          “Troll” Picks: If a certain champion is selected for the sole purpose of losing, then the results should tend towards a losing team. Reasons include poor team morale and low personal and team effort.
  2.          “Niche” Picks: If a certain champion is only selected when a certain other champion is on the enemy team, then the results should tend towards a winning team. Reasons include high morale, higher champion knowledge and effort, and higher chance of winning a specific lane.

Low Win Rates

When a champion has a low win rate it indicates that over the course of the sample size in a range in time, this specific champion has historically underperformed.

What this DOES explain:
  1. Whatever people generally do with this champion may not be optimal;
  2. This champion's strength may be lower than the majority of other champions; and
  3. This champion's core strengths may be heavily outperformed by other, more popular champions in the current meta.
What it does NOT explain:
  1.          The potential overall strength of the champion;
  2.          The potential strength of a specific rune/mastery/skill/item combination; and
  3.         The strength of a champion in a different and shifting meta-game (champion selection, strategies, etc.)
This means that if any of the aforementioned changes or becomes apparent to more of the population, then this champion could experience a sudden increase in win rate and I'll likely experience a sudden increase in complaints in champion placement. (Unless I had the foresight to place them high to begin with, which means I've been getting complaints the entire time - See Wukong)

High Win Rates

These champions have historically won more often than other champions during the times that they're picked.

What this DOES explain:
  1. The current most popular item/skill/mastery/runes on that champion appear to be working; 
  2. This champion's combination of damage, utility, and survivability seem to be line with the majority of other champions currently popular; and
  3. The times when this champion is selected are favorable to that champion (may be a counterpick, but if the pick percentage is high enough, this seems unlikely.) 
What this does NOT explain:
  1. That this champion is a good champion to pick in every scenario; 
  2. That a top 10 win rate champion is necessarily better than a top 20; and
  3. That this champion will win their lane match-up more often then not.

Conclusion

There are numerous ways to win or lose a game. Every day, the metagame shifts and certain champions move in and out of popularity. All the while, Riot nerfs and buffs champions on a consistent basis in the name of balance.

While statistical analysis won't be the end all to champion selection and team synergy, in solo queue at least, a champion's win rate should be looked at if it stands out far ahead or far behind of the pack. This generally signals that there's either untapped potential or meta sheep bias at work.

However, since most win rates generally lie within a similar spectrum, great care should be taken when saying that a 54% win rate champion is significantly better than a 52% win rate champion. Despite the fact that the difference in the two may very well be over 20 places, their overall success is still very similar.

What do you think? Agree/Disagree? Post your thoughts below!




First time to Nerfplz.Lol or not sure where to find everything? Try the Site Map





34 comments:

  1. ShadowSectJune 28, 2014

    Thank you so much for making this SSJ. This makes it a lot more clearer for all of us :D

    ReplyDelete
  2. SquirrelNutkinsJune 28, 2014

    Good analysis, pretty spot-on lesson of statistics 101 in regards to win-rate. Too bad the people who will benefit the most from reading this are likely to skim over it and be up in arms about win-rates in the next tier list lol

    ReplyDelete
  3. ShadowSectJune 28, 2014

    Thankfully there are people like us who can put that into check :D

    ReplyDelete
  4. Just wanted to say: I love your work. It's explanatory, unique and objective. Lots of other websites only present facts. Keep up the good work!

    ReplyDelete
  5. This is off topic, but could someone explain to me why is Kassadin banned in every single "pro" game like in LCS or in OGN? Riot nerfed him many times and he is still the most banned champion. But why?

    ReplyDelete
  6. ShadowSectJune 28, 2014

    Because according to skill, he is one of the most skill reliant champions, which pro people love EXPONENTIALLY.
    As for in regular solo queue him still being banned, is because he's still broken. While still being very skill reliant, if a very good person plays him, he does REALLY well.
    He has insane mobility post-6, built in mana sustain, his kit has a ton of utility which is great for teamfights, he is one of the burstiest champions in the game (another reason why pros love him), he has relatively low CDs especially on his ult, which allows him to ult in, burst, and then ult out (which can also help when under turret).
    Also, he's REALLY good at snowballing like LB.
    The main reason why he's not God is because he's skill reliant. VERY skill reliant. About half as much as LB.

    ReplyDelete
  7. I see. But basically, in competitive play one of the main reasons he is always banned is just because it takes skill to use him? .-.

    ReplyDelete
  8. ShadowSectJune 28, 2014

    No you misunderstand.
    People in LCS (pros) are EXTREMELY skilled, more so than pretty much everyone on this website. Champions that are REALLY OP with enough skill are loved by LCS, specifically LB and Kassadin, which they ban so nobody can have them.

    ReplyDelete
  9. Assassin with high mobility. After all his riftwalk is 8 seconds cd and is almost as long as a flash. He gangks other lanes with ease.

    ReplyDelete
  10. Hmmm now I understand. Thanks for answering my question.

    ReplyDelete
  11. Win rates mean nothing if they are not solely in regard to lane match ups. Win rates are often skewed by popularity.


    Maybe I could explain myself a bit more clearly like this:


    For example, there are 14 ADC's, according to lolking, who are most commonly matched up against one another. 5, of those 14 are in the top 10 in popularity of ALL champions over 1 week's time in ranked soloQ. Stay with me here. Caitlyn, the most popular of the bunch, is strong against Vayne and even with Lucian. Lucian is weak against Vayne. With this in mind, Caitlyn, is the most popular pick for ADC and has the highest win rate amongst them. 2nd most popular pick is Lucian, and 3rd is Vayne. This does not mean that she is the most powerful of the ADC's, she is simply a "niche pick to a larger scale", if you will, because she counters the MOST POPULAR ADC'S.


    So how do we bypass the popularity of Caitlyn in determining how truly strong of a play she is? Well, let's compare her to ALL of the 13 other most popular ADC's. Larger scale analysis = more accurate results. Against the 13 other ADC's, Caitlyn has a sub 48% win rate (I include ~49-50% as being a wash) against 4 of those 13 champs. Pretty nice numbers. Now, let's look at Corki. Not amongst the most popular champions in the game, and is 14th amongst ADC's. Corki has a sub 48% win rate against 5 of those champions. Not much of a difference at all. The only difference is that Corki is not played as often, thus, flies under the radar and remains typically in the "Balanced" category on most Tier Lists. Also, against the 3 aforementioned "most popular ADC's", Corki has a win rate of 54.5% against Caitlyn, 53.4% against Lucian, and 61.9% against Vayne. So, Corki is almost identical to Caitlyn in this regard.


    We can't completely elminate popularity from skewing numbers because there are a million different factors that come into play. But, if we try to mitigate its effects to the best of our ability, we can start to see what champions are *actually* strong. I suggest a total lame match up win rate (add up, divide by 13) to start.




    This could all be gibberish, it's really late. Hopefully you understand some of it. Love the site by the way. Signed up just to post this.

    ReplyDelete
  12. Thanks for the explanation, very helpful.

    Any estimate on when the full tier list will be up?

    ReplyDelete
  13. SquirrelNutkinsJune 29, 2014

    Agreed, good analysis.

    ReplyDelete
  14. RundownmonkeyJune 29, 2014

    Your first explaination for lower win rated champs is spot on when you say that "Whatever people generally do with this champion may not be optimal." ive had account purely dedicated to urgot for a while now and my winrate on that account is about 65% win rate due to me playing him in top where i feel is play style suits him better.

    ReplyDelete
  15. Is being able to dumpster Quinn and Miss Fortune in lane actually relevant to a discussion of current metagame viability?

    ReplyDelete
  16. Kassadin is many things, but 'bursty' is not one of them.

    ReplyDelete
  17. I'll hold you to that.

    ReplyDelete
  18. ShadowSectJune 29, 2014

    He ults in, Qs and Es, ults out.
    If not Kassadin, then the pros are making him bursty.

    ReplyDelete
  19. Sorry, let me rephrase -


    Kassadin bursts really badly.

    ReplyDelete
  20. ShadowSectJune 29, 2014

    I'll take that.
    He's still really good when he snowballs though. IF he snowballs.

    ReplyDelete
  21. MechadreamJune 29, 2014

    To be completely honest I understand and appreciate the need for everyone's analytic arguments. However, I believe that most of it just comes down to the logic...


    All you need to do is look at
    *Pick rate
    *Ban Rate
    *Win rate
    *How they do against counters etc
    *Meta viability.


    They are few more factors but it seriously takes like...Not even a minute to weigh up all the factors and judge their win potential .... Ugh, I feel like I ended up explaining this terribly. But all these analysis and opinions are great, but it just seems like everyone is over-complicating everything.


    Basically, I am for win rates

    ReplyDelete
  22. MoriartyGamerJune 30, 2014

    I'm not english, but I understand almost everything of your posts, i love them, thanks for that explanations, and for the tier list :3

    ReplyDelete
  23. SquirrelNutkinsJune 30, 2014

    I agree, except ban rate isn't the best thing to look at in my opinion. Ban rates are simply a result of 2 of the factors you mentioned (meta viability with win rate). Kind of redundant to look at bans, as it is already accounted for by other factors. Not to mention popularity of bans aren't 100% correlated with the most OP champions...some bans are based on old patches that people haven't adjusted for and sometimes people ban selfishly, or get rid of champs that are very annoying to be against but aren't necessarily OP (evelynn is the 10th highest banned champ at the moment but she's not OP).
    I'd say instead of your factor of ban rate, I'd replace it with "simplicity" or "easiness". And I think that's a factor SSJ does take into consideration.

    ReplyDelete
  24. MechadreamJune 30, 2014

    Yeah you're right, not sure why I included it tbh...

    ReplyDelete
  25. Plz tell me im not the only one who didnt read this whole post. :3

    ReplyDelete
  26. I've watched a lot of the LCS games, and what i heard is the nerfs didnt hit as hard as u would expect and that after 1 or 2 kills Kass can snowball the game, after those kills kass can pick off the slow sheep out the herd.

    ReplyDelete
  27. SquirrelNutkinsJuly 01, 2014

    It makes sense! Your other factors you named were so dead on that it made that one factor look out of place. I agree with you

    ReplyDelete
  28. Agreed on all counts. It confuses me that you understand all of this and are still use win rate to inform your placements to such a huge degree.

    ReplyDelete
  29. "This is exactly what the Tier list is aimed to do - to give less experienced players than Plat an idea of strong, blind pick champs in champ select."


    Wait, what? Blind pick? This is a solo queue tier list. Wouldn't this be for draft pick?

    ReplyDelete
  30. "Blind Pick" was a poor choice of words on my part - I meant as in being top four picks in the drafting process of ranked, and choosing a top/mid/jungler. You are probably going to get counterpicked if you go Renekton/Katarina/Sejuani if they haven't chosen thier lane pick yet.


    Not knowing their top/mid/jungler, this list gives you a good idea of who you can pick and usually still do well.

    ReplyDelete
  31. Ah, I see what you mean.

    ReplyDelete
  32. Where is the 4.10 tier list?

    ReplyDelete
  33. IMHO, win rates should NOT be used to determine champion balance.

    I mean, is an ideal "balanced game"'s goal to aim for all champions to have 50% win rate? That's just ridiculous.

    It's just people using what's popular regardless of how useful it really is.

    Every champion is different enough to bring something different on each game.

    ReplyDelete

Feel free to comment or leave a message :)