New...Please explain win rate statistics

Tearin True·Mar 2, 2015·1 votes·834 views

So I am a math teacher and I play LOL and as a math teacher I am curious how web sites get their "win rate" statistic.

Is it a mean?
is it a median?

Some other kind of central tendency?

Does this stat take into account frequency champ is played?

What is a healthy win rate stat? I ask because people are mad at the vieger "nerf" and the decrease in win rates. Well I would argue that if I was RIOT my aim would be to make sure that all champs have roughly the same win rate statistic or in other words every champ is equally viable. Depending on how things are measured would tell me what % win rate is considered the "healthy" win rate.

The problem is the time it takes to get it just right because as soon as you tweak just one little thing is changes the entire dynamic.

14 Comments

FHMarshyMar 2, 20151 votes

Is it a mean or a median? I think it's just a pure fraction of #ofgames/#ofwins

I found my stats on na.op.gg, where you can keep track of every games you played. I believe the algorithm they use is just a simple division. They do also provide other stats (Ban/pick rates, etc)

I don't think pick rates would affect win rates.

I think a healthy win rate is somewhere near 50%

TehNACHOMar 2, 20151 votes

Your fraction is backwards. Also, that's a mean.

FHMarshyMar 2, 20151 votes

Yes thanks for point it out the fractions.

But i would argue if you should count that as the mean. Yes, it would be mean if we are only considering it as a binomial stats with just Win and Loss, but I would say there exist just more than just win and loss to determine the ability of a champion.

TehNACHOMar 2, 20151 votes

I'm not here to are what "should" or "shouldn't" be; just that your fraction, in and of itself, is a mean.

Don't start preaching to me.

FHMarshyMar 2, 20151 votes

The fraction in itself is a fraction, mean is actually the average of the population.

Don't start going into technicalities on me.

PoisonedTeaMar 2, 20151 votes

A healthy win rate is around 50%.

taking the win rate is of everyone (that plays ranked) doesn't always = right.

Take Hiem he got the highest win rate of all champion he has 55% in bronze silver and gold. That is 90% of the playerbase. The top 10% it is 51% is still strong. He is an above average champion with numbers there. Of course only 4% of playerbase play him that is not that high.

Thresh has 29% playerbase and only 47% win rate.

although most top picks do have a good win rate right now.

But riot will imbalance it game on purpose and sometimes not. So many times has an unchanged champion just show its ugly face as a major pick/ban

RasvMar 2, 20151 votes

Now that you mention it, for it to be a win rate shouldn't it be the number of wins per unit of time? "Win rate" is actually the percentage of games won.

Champion win rates are typically close to 50%. Something as high as 55% or as low as 43% are considered outliers. A champion with a high win rate does not necessarily mean it's overpowered if it isn't picked very often, because it could be a situational pick.

GaleWinUnleashedMar 2, 20151 votes

Theoretically, every champion's Win Rate should be 50%. Realistically, achieving that is nearly impossible (and we don't want it anyway; perfect balance != everything is identical).

Win Rate is merely a champions number of wins divided by how many games they are played in. Plain and simple. Most Win Rate statistics you find online (such as those on LoLKing) are calculated for a recent time period in order to give you an idea of recent performance. Win Rate stats do NOT account for the popularity of a champion (aka Pick Rate). There are several patterns that can typically be observed by comparing combinations of these statistics. A high Popularity (approximately 10 / 123 = 8% is average, so anything over 20% can safely be considered high) can happen in several scenarios, but is always based on at least the perception of power.

If high Popularity coincides with a very high Win Rate, it's reasonable to assume that the champion in question is overtuned, and needs a few significant nerfs to their numbers. Extreme cases such as a Win Rate of 55% or higher need immediate attention if the champion in question is highly popular.

On the other hand, there are also cases where a champion is highly popular, but has a poor Win Rate. This typically means that the champion is A) new or B) perceived to be very powerful. In both of these cases, the poor Win Rate if often due to a great number of inexperienced players hopping on the bandwagon, greatly skewing the Win Rate.

In the long-term, this tends to settle down some, but it does not go away entirely. If the champion is difficult to play, the trend will remain. Take a look at Lee Sin, for example. There are a great many posts made about how Lee Sin is "OP" (I'm not offering my opinion either way; another time perhaps =P), yet his current Win Rate is only 46.68%. Lee Sin is incredibly versatile and dangerous in the right hands, but he requires a fairly high level of skill in order to properly take advantage of this. Well over half of the playerbase does not possess the necessary level of skill to use him well, but that does not stop them trying anyway because "WOW, LEE SIN OP! DID YOU SEE THAT STREAM LAST NIGHT OMG #LCSBIGPLAYS!"

Win Rate is a great statistic, but has to be taken with a grain of salt; there are a great many factors that go into it, and just looking at the end result doesn't give you a very clear picture.

turtlecoonMar 2, 20151 votes

{quoted}

In an ideal world, every champion's Win Rate would be 50%.

From a mathematical standpoint, maybe.

From a game design standpoint, no. You want picks that are contested. You want picks that are less popular. You want some champions that are more difficult, but offer higher payoffs. You want some champions that are easier, but offer lower payoffs. Flat 50% winrates across the board is not a healthy statistic - then every champion would just be a minion.

GaleWinUnleashedMar 2, 20151 votes

Okay, in a theoretical world? Actually having a 50% winrate for everyone would be boring.

turtlecoonMar 2, 20151 votes

{quoted}

Okay, in a theoretical world? Actually having a 50% winrate for everyone would be boring.

Yep. I guarantee you if people didn't have 55% winrate champs and 45% winrate champs to complain about, they'd be complaining about how stale everything being 50% is instead. People always find something to complain about.

GaleWinUnleashedMar 2, 20151 votes

Fixed it.

SmokingPuffinMar 2, 20151 votes

Is it a median?

Are you sure you're a math teacher? This is nonsensical -- the median win rate for a champion would be either 0 or 1. Really useless information.

Tearin TrueMar 2, 20151 votes

{quoted}

Are you sure you're a math teacher? This is nonsensical -- the median win rate for a champion would be either 0 or 1. Really useless information.

I must be tired. I was thinking in my head of setting u some kind of frequency table taking into account some other variables and then it got all discombobulated in my mind.