Why would you use win rates to judge how good a champion is

SlackerboyTC·Jan 6, 2015·8 votes·785 views

Not everyone in this game is challenger so, I don't see how win rate is a good determinant for much of anything

17 Comments

DJ SiamakJan 6, 20155 votes

The winrate is for every single elo so it is a good indicator.

SlackerboyTCJan 6, 20152 votes

[{quoted}](name=DJ Siamak,realm=NA,application-id=3ErqAdtq,discussion-id=t0oryoTB,comment-id=0001,timestamp=2015-01-06T00:36:55.591+0000)

The winrate is for every single elo so it is a good indicator.

Umm every elo = incredibly diluted data

DJ SiamakJan 6, 20155 votes

{quoted}

Umm every elo = incredibly diluted data

Okay so which elo should they balance around then?

High winrate is a good indicator, you can't base everything off of it, but when you have Rek'Sai who has a 55% winrate with a combined pick/banrate of about 80% then you have a problem.

SlackerboyTCJan 6, 20152 votes

[{quoted}](name=DJ Siamak,realm=NA,application-id=3ErqAdtq,discussion-id=t0oryoTB,comment-id=000100000000,timestamp=2015-01-06T00:44:30.492+0000)

Okay so which elo should they balance around then?

High winrate is a good indicator, you can't base everything off of it, but when you have Rek'Sai who has a 55% winrate with a combined pick/banrate of about 80% then you have a problem.

Thats the thing, they shouldn't balance it around win rates

DJ SiamakJan 6, 20157 votes

{quoted}

Thats the thing, they shouldn't balance it around win rates

Do we base it off of feelings or something?

SlackerboyTCJan 6, 20152 votes

[{quoted}](name=DJ Siamak,realm=NA,application-id=3ErqAdtq,discussion-id=t0oryoTB,comment-id=00010000000000000000,timestamp=2015-01-06T00:48:39.349+0000)

Do we base it off of feelings or something?

You sure as hell don't base it off just win rates. Anyways, it should be based off actual information about a character, I mean, what does a 60% win rate tell you? For all you know, there could've been a bunch of afk and feeders on the enemy teams and such in those matches played by said champion, it does not provide good information

futuresight23Jan 6, 20154 votes

{quoted}

You sure as hell don't base it off just win rates. Anyways, it should be based off actual information about a character, I mean, what does a 60% win rate tell you? For all you know, there could've been a bunch of afk and feeders on the enemy teams and such in those matches played by said champion, it does not provide good information

Depends on the sample size. Given enough games, there will be just as many afks and feeders on the side of the team with the champ with the 60% winrate as against that champ. Over a day winrates are unreliable but tracking winrate from every solo queue game (ranked removes such variables as 2 Lee Sins, or champion free week) over a week, a month or more tends to bring out more consistent data that could be usable.

Winrate is not the end all be all, but with a good sample size is one of many points to look at in terms of evaluating a champ's power level.

SachabotJan 6, 20151 votes

That's the only scenario it which it tells you anything useful, but it still doesn't tell you why. And you need to know why something is the way it is if you want to change it. So winrate still wouldn't be useful as a statistic.

SmokingPuffinJan 6, 20151 votes

[{quoted}](name=DJ Siamak,realm=NA,application-id=3ErqAdtq,discussion-id=t0oryoTB,comment-id=000100000000,timestamp=2015-01-06T00:44:30.492+0000)

Okay so which elo should they balance around then?

The highest elo they have a sufficient data sample to do statistical analysis with. Probably challenger is too small, so either master or diamond 1.

Competitive games should be balanced around correct play. We don't nerf knights because bad chess players can't see the forks.

futuresight23Jan 6, 20154 votes

Not on its own, though winrate does provide a good starting point in terms of discussion. The biggest question after that would be why that is the case.

SmokingPuffinJan 6, 20153 votes

Win rates, used responsibly and as part of a complete data breakfast, are useful information.

For example, Rek'Sai's OP. On the other hand, Heimerdinger is not OP. You can't know that purely from the win rate data, and you need to cross check with other sources, but it's still a good place to look for intel.

kirbyzamaJan 6, 20152 votes

Actually since most players are bad at the game, comparatively speaking of course, win rates are a good way to determine if a champ is over tuned versus expected values. Look at champs like Leblanc or Lee Sin. They have really low win rates overall because the majority of players who play them are low elo and can't play them effectively. As you rise in divisions their win rate improves but the player base is so small they don't elevate the win rate too much.

Now look at Warwick 4.20. He had an absurdly high win rate across all divisions. however as a simple champ he should be effective in less experienced hands however even when up against superior players on "superior" champs in terms of mechanics he still shredded. A 60% win rate is too high on a champ that can be played by a monkey. Meaning that bronze monkey would be playing WW as almost effective as the diamond monkey and reap all the rewards for it.

Now if a champion that is supposed to be a high skill cap champ wins to easily in low divisions that means they are over tuned and all the movement mechanics/cc etc are just bonuses as they don't need that skill cap. Look at Riven in season three. Sure you could be Boxbox and do all these flashy combos but why would you if you could just mash qqqw with no autos but still burst the enemy?

It isn't the only thing to take into consideration when finding ops. I mean Fiora has an above 50 win rate but that is only because Fiora mains exist and they are the only ones playing her. Now if you combine win rates with pick rates you find people gravitate towards champs that win them a lot of games. Champs that win often AND are picked often means that you most likely found an op.

SachabotJan 6, 20151 votes

It wouldn't even matter if everyone was Challenger. Winrate isn't a statistic that should ever be brought up as a statistic. It doesn't tell you anything meaningful.

RebonackJan 6, 20155 votes

{quoted}

It wouldn't even matter if everyone was Challenger. Winrate isn't a statistic that should ever be brought up as a statistic. It doesn't tell you anything meaningful.

It can't be used in a vacuum.

If a champion has a pick rate in the bottom ten and a 45% win rate, then chances are they're really bad.

If a champion has a high pick/ban rate but a low win rate, then there's probably something skewing the data.

If a champion has a massive pick/ban rate and a huge win rate, then there's a decent chance that they're over-tuned.

ValyrianBladeJan 6, 20151 votes

Win rate alone is a useless statistic, I agree.

A really underpowered non-meta champion could still have a good win rate and need buffs because the only people playing it will be people that really know the champion and that role, there won't be any players trying it because everyone says it's op.

In contrast, a champion that's really overpowered is going to get picked by bad players a lot for free-lo, which will dilute the win rate.

Pick/ban rate plus win rate is meaningful though. If champion X has a 1% pick rate and a 53% win rate, he definitely doesn't need a nerf (and may even need a small buff or a rework). If champion Y has a 40% pick rate and a 53% win rate, he needs a nerf. If champion J has a 40% pick rate and a 45% win rate, he really doesn't need a buff but might need a rework. If champion K has a 1% pick rate and a 45% win rate, he definitely needs a meaningful buff or a rework.

There's also the fact that difficult to play champions should be stronger than easy to play champions, and the fact that bad players are drawn to certain types of champs while good players are drawn to different types of Champions. For example, I would imagine zed draws a lot of attention from bad players (good players too though) because he can make really flashy big plays and has a lot of safety and individual stand-out power. So his win rate is probably really a lot less than it would be if he were picked proportionally across skill levels - which means that while he probably needs a nerf it isn't as evident as it should be (I was horrified when I saw he got buffed recently...)

Edge of DaybreakJan 6, 20151 votes

Id say we base it on pick win and ban. Ban shows people dont like playing against it. Pick lets us know how often people play it. Winrate lets us know if its actually winning. Take sona for example. Low banrate so people are fine playing against her. Higher than average winrate so we know she wins a lot. Average pickrate so we know she is picked enough. If anything her winrate shows she is nowhere near as bad as people claim she is. But then again no champion has never been as bad as people claim sona is...