Boards›[ARCHIVED] Miscellaneous

Math Doesn't Lie - Your Emotions Do

FarRockBF··16 votes·1,905 views

You all get so emotional about Lee Sin, Zed, etc. First look at my match history. I never play either of them, OK?

Now I posted this on my Lee thread but will post it here again: Win Rate is EVERYTHING. Math says so.

Let's take an extreme example, OK?

100 vitamins. Each has bad and good elements. A BALANCED vitamin will make you neither super strong nor super weak - you will be perfectly normal.

If you take 10 random pills and feel a bit weak there is definitely at least 1 pill in the 10 that is more bad than good, correct? (There could be some that are more good than bad but to get below the middle line you need at least 1 pill to be more bad then good.

But there are other elements involved! Maybe you went drinking last night? Maybe you were already sick? OK no problem. We label our pills 1-100 and give 10 random of these pills to ONE MILLION PEOPLE.

If we notice that 90% of people who had pill #44 in their lot got weak, what does that tell you? That there is a damn good chance that pill #44 makes people weaker.

BACK TO LEAGUE.

Champion X. Amazing stats. Starts will 2000HP. 200 AD. Seemingly ridiculously OP. But... out of 1,000,000 games with him, his team lost 800,000. You see, despite having stats that everyone agrees is OP Champ X is weak as far as the GAME is concerned. Balance should not be about champions; it should be how they affect the outcome of the game. After all, LP is based off of WINS not damage dealt.

The only logical counterargument I have ever heard amongst all the whining here is this: MOST of the player pool don't know how to play most champions well, so it is not fair to count them. This is especially true if the champion is fun to play.

I partially agree here, so to even the score look at win rates amongst pro players only. As a side point these players are probably not very good at other champs either. Additionally, they probably don't play Champ X very often (or they WOULD be better with him) so they probably aren't contributing THAT much to his game count.

With the exception of the above argument all of your emotional whining cause you lost to a pro Leblanc, Lee, Zed, whatever. If the win rates across the board say that player LOSES more games then wins, then math is math - that champ is weak as a GAME champ and you should have won.

/Rant

28 Comments

Jubbinaut12 votes

There's quite a few factors that looking only at win rate doesn't account for. Among those is pick rate. Take the following examples:

  • A low-picked champion with a high win-rate has an arbitrarily high win-rate, as it's likely those picks come primarily from players who focus on that champ. This makes the win-rate less valuable as a determining statistic.
  • A highly-picked champion with a low win-rate has an arbitrarily low win-rate, as it's likely to've fallen into the FotM pit, with abnormally high picks from abnormally unskilled players (on that champion). This makes the win-rate less valuable as a determining statistic.
Trylobyte2 votes

{quoted}

There's quite a few factors that looking only at win rate doesn't account for. Among those is pick rate. Take the following examples:

  • A low-picked champion with a high win-rate has an arbitrarily high win-rate, as it's likely those picks come primarily from players who focus on that champ. This makes the win-rate less valuable as a determining statistic.
  • A highly-picked champion with a low win-rate has an arbitrarily low win-rate, as it's likely to've fallen into the FotM pit, with abnormally high picks from abnormally unskilled players (on that champion). This makes the win-rate less valuable as a determining statistic.

I would argue the second factor applies mostly in Plat and below. The higher tiers aren't likely to change quickly and shift off champions they know to be successful (especially in NA, a region always behind in the meta) so we can assume those folks have put in the time to learn a champion outside of Ranked and aren't just picking it because it's popular. It also becomes a less-valid argument if a champion has been popular for a long time, since by then it's safe to know the highest tiers of competitive play have had plenty of time to learn to get skilled with that champ.

There is, however, a third factor too. Ban rates. A champion that's getting banned a lot will sometimes have an arbitrarily low win-rate when people don't have the chance to learn to play them because they're always banned.

Jubbinaut1 votes

Well, I wasn't really going for a comprehensive list. :P

Though that's a good one to add. I like it.

FarRockBF2 votes

{quoted}

There's quite a few factors that looking only at win rate doesn't account for. Among those is pick rate. Take the following examples:

  • A low-picked champion with a high win-rate has an arbitrarily high win-rate, as it's likely those picks come primarily from players who focus on that champ. This makes the win-rate less valuable as a determining statistic.
  • A highly-picked champion with a low win-rate has an arbitrarily low win-rate, as it's likely to've fallen into the FotM pit, with abnormally high picks from abnormally unskilled players (on that champion). This makes the win-rate less valuable as a determining statistic.

I see what you are saying. I would argue though that when you look over a period of 3 months over millions of games this sort-of washes away too.

But point taken.

corbear0072 votes

{quoted}

I see what you are saying. I would argue though that when you look over a period of 3 months over millions of games this sort-of washes away too.

But point taken.

It has too many variables even in those 3 month span.

Example i would pick, take a new champ, call him Z give him an incredibly op kit, insane ad bonuses, double armor pen, 500% crit, low cd flash, insanely low cd on skills, resourceless, make him the most OP thing in league, ult will kill anyone instantly etc etc. Now make all of his skills at an insane high skill so much so that even those pro koreans mess up quite regularly after years of practice, give him to the public wish to speculate on the win rate? It would be extremely low.

Should be through the roof if math doesn't lie? Your talking about someone who at 6 will instantly kill the whole team if used right... someone with extreme mobility insane life steal pack it out at 100% for all skills 500% if that's not enough, but make those skills incredibly hard to use take morgs Q speed length double it, but make it uncontrollable literally goes where it decides not you, take lees kick again, insta kill but it's uncontrollable with the width of 1/4 a teemo W and E the exact same

This in much less extremes is lee, korean pro players still to this day mess up lee (ward hops mainly) do I agree that he needs a nerf? Yes do I think they need to gut him? No, a few small nerfs would put him back in line with other junglers he's one of the most difficult champs to play perfectly this is a fact, he's also impossible to counter IF PLAYED PERFECTLY problem being we are humans we make mistakes errors in our calculations bad decisions

League has hundreds if not thousands of different steps in skill leading up to lcspro status this being said the top 10% can make lee an unstoppable force rivaled to none, the other 90% can and does skew data favorably thus all realms of data must be explored for a comprehensive picture

AMYS GRAVE1 votes

thank you for making the logical reply. win rates mean something, but they aren't everything.

ResponsibleDad6 votes

Jesus finally someone speaks the truth.

Auz the Ripper5 votes

"There are three kinds of lies in this world: lies, damned lies, and statistics" -Mark twain

Ask any statistician, you can't trust statistics. They're too easily skewed by bias and context.

SciencePrime4 votes

THIS MAN SPEAKS THE TRUTH.

Narasimha4 votes

Close, but not always true. Old Leblanc is a testament to that. Low pickrate, and low winrate. And yet, she was still hit with some heavy nerfs. Why? Because playing against her was extremely anti-fun. But her win rate was still low. Because Leblanc fell off, and didn't carry games. Did that make her any less toxic to play against? Not really. By your logic, she wouldn't have been a problem, because clearly her numbers didn't indicate as such. And you know what? At the end of the day, sometimes, emotions > hard math. Because this is a game. MOST people play it to have fun. If they're not having that fun for a particular and constant reason, then that means something is off.

Tsunday3 votes

Winrate is a bad statistic to go off of. When you're looking at statistics you want a dependent variable, and independent variable, and everything else to be a constant. WInrate depends on your team, the enemy team, trolls, dcs, lag, counters, individual decision making, etc.

If 1,000 people main Fiora for example, and in 70% of those games someone picks Pantheon into Fiora and after they all play dozens of games Fiora has a 34% winrate it doesn't mean that Fiora is underpowered. Especially not if her losing can be contributed to her counter being FotM or just her being countered in the first place.

If Taric is a champion used to troll when somebody gets stuck as Support and 20% of the Taric games in ranked end up being games played by 'full AS Taric' you can't say havin a 42% winrate makes him underpowered.

If there are only 500 Urgot mains on a server and the 500 people who play Urgot only play Urgot and they play him very well, you can't say that a 60% winrate means he's overpowered.

You can't claim a champion like Kogmaw who needs a specialized teamcomp is under/overpowered because of his winrate without thinking about what the meta is. If the meta is favoring carry jungles and assassin mids Kogmaw isn't going to have the peel to do well late game, it doesn't make him underpowered even if his winrate is low. Same could be said for a reverse meta where Koggles would get plenty of peel. He didn't become OP because he has a good winrate, he just consistently got put in teams where he was their main source of damage because of the meta.

Champions who carry early game like Graves and LeBlanc stand a better chance in a 4v5 when they can wipe the early laning and teamfights and then push before everyone is level 18 and approaching full build (aka when statistically a team of 4 doesn't have the damage or the tank to win a teamfight).

We could also talk about low Elo pubstoms like Malphite and Amumu and how individual decision making and individual skill level is the reason they're so strong, not the champions themselves.

I absolutely love math, I've taken classes in statistics, but with something like winrates it's a really easy variable to become confounded.

FarRockBF2 votes

{quoted}

Winrate is a bad statistic to go off of. When you're looking at statistics you want a dependent variable, and independent variable, and everything else to be a constant. WInrate depends on your team, the enemy team, trolls, dcs, lag, counters, individual decision making, etc. I absolutely love math, I've taken classes in statistics, but with something like winrates it's a really easy variable to become confounded.

Partially agree. You will know then that even if it is a bad stat to go off and there are other variables (X, Y, Z) that affect this, at a large enough sample those variables will become irrelevant.

Do you know what a solver is? So, solving for just X is easy. When you are solving for X and Y you need to do X*Y trials. Its just a number game. My point is simply those champions that have millions of games played over a good period of time compensate for the other variables.

Tsunday1 votes

I'd agree with it if there weren't so many variables that define winrates. if there were just a couple of variables that had an impact maybe I'd look at a 55% winrate and claim 'OP' but since there are very little constants (not even builds, skill level, prior experience, your constants are it's the same map, same objectives, same champion) and a larger number of variables while I will agree that winrates mean something they can't be taken as the indicator to a champion's strength/weakness. Once you get under 40% or over 60% I believe it gains a more weight, but for numbers close to 50% you have to look at the statistical significance that number has. In a case with so many variables the statistical significance of the values isn't going to be as high as one would like to justify the change of a champion.

I personally don't believe that Lee Sin is overpowered. I also believe that in my elo most Zeds would rather chase kills than close games . I don't see either of them as 'overpowered' (although I believe Zed is strong, I don't believe he is overpowered), but I still don't see their winrates as being the indicator to their strength.

junglerboy163 votes

You are taking this much much to objectively, to the point of making it no longer about a game, but about a science. Nobody wants Lee nerfed because he is OP (because really he isn't), people want him nerfed because he is atrocious to play against when even moderately strong, like he is now. I don't give a damn if I have a 10% higher chance of winning vs a Lee than vs somebody like J4, because this is a GAME. I don't play because I must win everything I set eyes on, I play because I want to have some fun playing a game. Lee makes this game not fun, in the same way that Teemo makes it not fun. Teemo has been subpar for AGES, but I still never want to play against one, nor do I want to play against a Lee, regardless of whether or not it means I get a free win.

You, Sir, have fallen for an obvious fallacy that because Lee isn't super OP by win rate (which he isn't) he doesn't need changes, because your glorious math and statistics tell you that he won't win as often as most champs. Life and math, while inherently linked and intrinsically intertwined throughout our existence, are not the same thing. Maybe you should consider the fact that people want Lee changed exactly BECAUSE they feel so negatively about him.

TL;DR: NOBODY cares if they win more than they lose against him, he still feels like crap to play against, is super broken (NOT super OP, big difference), and needs changes for those reasons, not because he wins too much.

EDIT: Just in case you were wondering, I haven't lost to a Lee because of him as a champion in months. They are always behind, underfed, and nearly useless by the time the match ends, so I'm not some butt-hurt crybaby who wants more free wins. The fact that he can even get to the point of being that useless late game is in itself another reason to rework him.

FarRockBF2 votes

{quoted} You, Sir, have fallen for an obvious fallacy that because Lee isn't super OP by win rate (which he isn't) he doesn't need changes, because your glorious math and statistics tell you that he won't win as often as most champs. Life and math, while inherently linked and intrinsically intertwined throughout our existence, are not the same thing. Maybe you should consider the fact that people want Lee changed exactly BECAUSE they feel so negatively about him.

I didn't fall for anything.

  1. Reminder this is not a Lee Sin thread.
  2. I am all up for CHANGES to these champions; just not NERFS
junglerboy162 votes

Yes, I know that technically this is not a Lee sin thread, but he was the most obvious example to use since everyone talks about him all the time.

Additionally, I completely agree that he doesn't need nerfs, he just needs a rework to make him more fun when playing against and less of an "I'm either super OP or really pretty useless" kind of champ. I think a rework would actually make him even more fun to play as as well, since it would help him focus on a real niche in the game rather than his identity as being ok at everything.

P.S. I do apologize but on rereading, my first post sounded quite rant-y and that was not my intention. I agree that win rate with enough data really does supply information about how OP or UP a champion is, my point being that being OP or UP is not all that there is to having fun in a game.

Thanks for such a civilized response in the face of my first post.

Sammystorm3 votes

Although I agree that zed and lee are fine your argument is not accurate. You are looking at math in isolation. You can't look at math in isolation. The numbers do not make sense with out context. For example: Cass for the last month has had a 53.1% win rate and by your logic she must be over powered. However that isn't the whole story. If you delve deeper and look at how popular she is. Only 4.5% pick rate and a non existent ban rate. So a legitimate conclusion we could make is that cass is being played more effectively then average by a small pool of people. So we can logically say that cass is probably not op but instead only people that are good at her have been playing her.

Now in comparison lets look at Blitz. He has a 53.1 win rate. The exact same win rate as cass. Now lets look at his popularity; 19.9% popularity and drawing some bans. So blitz is played in 1 out of every 5 games and has an above average win rate. So we know he is seen alot and picked alot and is successful. He would be a case to look closer at to see if he is op.

Now you are correct that math doesn't lie but how you analysis math can lie. You misinterpret the stats whenever you look at win rate or any variable in a isolated scenario. So make sure you remember the context and look at all the variables because other wise you will have a very poor understanding of what the stats are telling you.

Nucleophilic atk2 votes

Stop trolling now. Thanks.

FarRockBF17 votes

[{quoted}](name=SN2 Surpriser,realm=NA,application-id=cIfEodbz,discussion-id=feP5KanK,comment-id=0000,timestamp=2015-01-29T19:12:36.621+0000)

Stop trolling now. Thanks.

Typical. No logical answer so spew stupidity. Continue.

Nucleophilic atk3 votes

Win rates are really stupid to go by. If your champ was so "weak", then they wouldn't constantly be picked. It is that simple.

FarRockBF5 votes

[{quoted}](name=SN2 Surpriser,realm=NA,application-id=cIfEodbz,discussion-id=feP5KanK,comment-id=000000000000,timestamp=2015-01-29T19:39:43.654+0000)

Win rates are really stupid to go by. If your champ was so "weak", then they wouldn't constantly be picked. It is that simple.

What is more correct, FACT (how many times said champ wins) or EMOTION (I think this guy feels super strong I will pick him)?

Tsunday2 votes

Emotion to an extent. If Galio doesn't have a high winrate but you're against a double AP comp he's going to feel super strong and he is going to be a lot stronger than the statistics presented in his winrate as double AP comp (which is where Galio would arguably shine) isn't a very common team composition.

warpenguin5551 votes

so wait, why dont people think win rates matter? if a champion always wins lane but never wins a game, i dont see how thats OP

Nucleophilic atk1 votes

Right, because Zilean is clearly one of the worst champs in the game. That is why he wasn't permabanned in LCS.

Centuros1 votes

Win rates tend to normalize towards 50% regardless of how strong a champion is, if that champion is really popular. If EVERY team in Blind Pick has said champ, it will have a perfect 50% win rate in Blind Pick, no matter what.

junglerboy161 votes

Not how that works. If a champion consistently loses, then they will never have 50% win rate unless they have a 100% pick rate on both sides (which doesn't happen).

Centuros1 votes

Didn't say that could actually happen 100%, but the more teams (comparatively) that have a champ on both sides, the less significance of the other games have. Though obviously I don't know the numbers or if it is actually significant, but this is a greater than zero skew, for better or worse.

Not sure if it has any real bearing on your initial argument, but I decided to put it out there.

agbudar1 votes

yea i don't get the "NERF LEE" as much as i do Zed nerf's i play against plenty of lee sin's who dont know how to play him right and fail miserably. and this has remained to be consistent over the year of 2014 zed on the other hand 90% of the playerbase that mains him is good with him wins a lot and infuriates everyone.

More from [ARCHIVED] Miscellaneous

Open Letter to Parents of League of Legends Players
[ARCHIVED] Miscellaneous860 comments
2,611
votes
Rewards for positive play
[ARCHIVED] Miscellaneous19,712 comments
1,337
votes
I don't like Boards
[ARCHIVED] Miscellaneous324 comments
809
votes
STOP ALLOWING LEVEL 1 ACCOUNTS TO SEND FRIEND REQUESTS!!!
[ARCHIVED] Miscellaneous291 comments
679
votes
Wanting to do something special for an awesome friend. Riot, please help? <3
[ARCHIVED] Miscellaneous220 comments
614
votes
People that are telling others to kill themselves has to stop! Please help me get this noticed!
[ARCHIVED] Miscellaneous605 comments
600
votes

More by FarRockBF

Riot Please Revert This Change
[ARCHIVED] Miscellaneous39 comments
165
votes
Why The Balancing Team Suddenly Sucks:
Gameplay170 comments
151
votes
How To Deal With a Toxic Teammate
[ARCHIVED] Miscellaneous89 comments
105
votes
Upvote VS Downvote Threads
[ARCHIVED] Miscellaneous29 comments
104
votes
Vote me down - I agree with Phreak
Gameplay158 comments
98
votes
New Stat: AP Crit
[ARCHIVED] Miscellaneous75 comments
73
votes