If you flip a coin 1,000,000 times... (Winrate opinions?)

DarkHeadgear·Aug 16, 2016·2 votes·525 views

If you flip a coin 1,000,000 times and divided it into segments of 10,000, and assuming the amount of times heads landed were equal to the amount of tails landed, many of the segments would still not end up with 50% ~~winrate~~ heads landed, some 55% and some 45%. My point here is that in a game of 9 other players, can the winrate of a champion actually hold any ground? For example, a main might still pick in a 90% AP damage enemy team and lose.

I may or may not have confused you a lot, so here's a poll...

14 Comments

Locke DaemonfireAug 17, 20165 votes

Assuming a fair coin flip, the chance of 55% or more heads in 10,000 flips is infinitesimal. 99.9% confidence interval is between 48.45%-51.55%, if I'm using the online tool for binomial distributions correctly. With enough data, even a small deviation in percentage is significant.

DarkHeadgearAug 20, 20161 votes

That's impressive. Do remember this is an analogy and that in reality, it is even more random being 5v5 in league.

EmbertineAug 16, 20163 votes

But coin flips are terrible as a comparison. The outcome of the coin flip can often be determined by the amount of force used to flip the coin, as well as the height of which the coin was flipped before it lands.

Actually now that I think about it the same can be said of matches as well. Matches are influenced by skill level, choices made, and other factors.

Angry MonsterAug 17, 20161 votes

I like the way you are thinking here

The difference between a coin flip and the game are the amount factors that the flipper controls verses the amount of factors that a player in a 5v5 control.

A person flipping the coin can control where, when, weather (inside/outside), speed, force, ect. I am guessing that ~95% of the conditions are controllable by a skilled person.

While there are so many other facets on of league where players can not control a majority percent of the game.

This is why it is a bad example.

DarkHeadgearAug 20, 20161 votes

I see my error in my thinking. However, do you understand the point that i believe is still valid? Are the champion win rates reliable? We are humans, fallible. We play these champions so can we expect the win rate to be a solid indicator of the champion's worth?

Angry MonsterAug 20, 20161 votes

Win rates mean something, at higher skill levels they mean more than at lower skill levels. The lower you get the more in consistent your play is going to be. Also it matters how big a gap is from 50%. 1 or 2 percent is not a big deal. 3% or more of deviation from 50% usually mean there is a problem; strength of the problem goes up as the numbers get more one sided.

But if a champ as a 45% win rate in bronze and a 49% plat higher, it would mean there is a problem that less skilled players have at bronze level compared to plat +. What is the problem can be up for a debate: mechanical skill, team synergy, or too complex kit.

Everything has to be taken in context of what each number means and is compared to. When you do the numbers are not random or meaningless. They are hints about what is going on with the champ but are not absolutes.

Most people look at numbers out of context or fail to understand what is really being said. For a casual number reader the numbers are probably worthless.

DarkHeadgearAug 21, 2016

Is there a way to view these seperate stats?

Angry MonsterAug 21, 20161 votes

through 3rd party sites? then yes. OPGG allows you to view stats per division. Championgg shows plat and higher since that what teh balance team tends to focus on.

There are plenty of other sites that pull data from the riot API to put together stats. You just need to find one that is providing the info you need.

DarkHeadgearAug 21, 20161 votes

Thanks, appreciate it.

DarkHeadgearAug 20, 20161 votes

I suppose. Maybe I should go physician mode and use the rotation of an atom instead (Or was it molecule?). Think about Blitzcrank. Works great in low elo due to the skill required to play him, you only need hooks and he can even work well through mid-elos, but dies at high elos. Or so I think. Either way, would the data be large enough to accept a champion as good or bad?

TreeofwarAug 16, 20161 votes

and that statistic is still higher than me getting a girl friend :^)

catastrophicSkilAug 16, 20161 votes

winrates are saturated with about 80% of the games belong to silver and bronze players and although not everyone is bad i good majority of them are. usually high win rate champs have overpowered in the case of reworked sona and some are just cheesy pubstomper picks that if play against correctly aren't really that op i'm talking about champs like amumu in this case

LordGeovanniAug 20, 20161 votes

you are right win rate doesnt really mean much because people forget there are multiple factors (coin flips) in every game. 1)Did you pick an op broken or meta champion 2) did your allies pick an op broken or meta champion 3) do your allies champions complement yours 4) do you coutner any of your enemies 5) do your allies counter any of your enemies 6) do any of your enemies counter you 7) do any of your enemies counter your allies 8)

I'm bored i'm gonna play a game and maybe update the list in between games