Percentages are hard

[Snowcrash000]

I'm not sure if this has been brought up before and maybe I'm really the idiot here for missing something really obvious, but is it really that hard to calculate win percentages correctly? Obviously all of those are all over the place, but some of these are nothing short of hillarious, why can't the site manage this?

[GTSchemer]

It's been that way as long as I can remember.

I suspect the reason is something based on rank in the game, like being 2nd place in a 4-player game might be considered a fraction of a game? Regardless, it's either very difficult to fix, or no one has been bothered enough by it to ever fix it.

[turtler7]

Edit: I was 3/4 right but confusing. Read Kayvon below.

NOTE: Friendly (or at least alpha) games do something strange with their counts. Winning 3 of 3 games in an alpha game has some REALLY strange stuff to those stats.





Note it doesn't say I have any victories even though I won them all. I blame friendly and or alpha for that.
But it also tries to say a win percentage... that doesn't make sense either.

If I am expected to win 1/3 in 3 player games... it is supposed to multiply by 3 to remove the 3 player feature and then divide by 2 to normalize to a 2 player game... so winning 2*3/2 is supposed to count as 3 victories along with another 1v1... It should be counting me as 4 victories in 3 games and presenting 133% (what we see often). Instead it is saying 83%. I am a bit confused by that one.

The only alpha game that it makes me a little sad for is Tichu. Since it say 0 victories for all the alpha games it makes it look like my win percentage isn't anywhere near as good as it actually is. My self worth isn't entirely based on winning games thankfully. But I would like to be able to show off a little sometimes.

[Jellby]

As far as I know, only full victories are counted: 1st position, no training, no abandoned games. But victories against more players count more, basically: 0.5 times the number of players at the table. So, in 2-player games, each victory counts as 1, but in a 7-player game each victory counts as 3.5: if you play a single game and win, you'll have 350% victories. The reasoning is that, in the long run, if you just win the "average" amount of games (i.e. 1 out of 7 games in 7-player games), your rate will be 50% no matter the number of players.

[Kayvon]

That's correct. It's percent of expected wins, normalized to 50%. If you only play four player games and always win one in four games, that's 50%. If you win twice as often, that's 100%.

Percentage = (numberOfWins / numberOfGames) * (numberOfPlayers / 2)

Percentages aren't that hard, though they could be clearer about what they're really measuring.

Edit: Corrected my mistakes based on Jellby's kind suggestion. (I had previously misstated that it was normalized to 100%, which was simpler but also wrong.) The equation's goal seems to be that the expected win percentage should always be 50%, regardless of the number of players

[Jellby]

That's correct. It's percent of expected wins. If you only play four player games and always win one in four games, that's 100%. If you win twice as often, that's 200%.

Not exactly (at least I don't think so), if you win one in four, that's the "expected", and you get 50%. Same as winning one in two in 2-player games (the only case where it means what it says).

[Kayvon]

You're absolutely right. I knew something was off as I was writing that. I'll edit my post to clarify.

[RicardoRix]

How is 'Just One' 52% ?
Something weird with co-op games?

[Kayvon]

Yeah, they changed coop scoring recently. I believe the intent was to account for the difficulty of the coop scenario, but I didn't follow it closely enough to know the details.