How can a player have a wins percentage that is over 100% in a game?

[tarongot]

Yes. (Though co-op games might cause inaccuracies; I'm not sure.)

Its not really inaccuracies as completely different formula. I play coop ranked games, lots of 2 players but also fair amount of 3 and 4 players, and I have around 75% wins on those. So, since many of those games were 3 and 4 people percentage should be higher then that. And in reality it is half that.

[Barberserk]

BarnardStar is right, it has nothing to do with elo. Only 1st place counts. If I remember correctly, tied 1st place counts as half (I believe independently of number of players tied), and only ranked victories count as victories (but friendly games still count as number of games to divide by).

Bottom line: it's a pretty useless stat.

I agree, but if it is useless and so tricky to calculate, as this conversation here proves, why not just get rid of it?

[FrankJones]

BarnardStar is right, it has nothing to do with elo. Only 1st place counts. If I remember correctly, tied 1st place counts as half (I believe independently of number of players tied), and only ranked victories count as victories (but friendly games still count as number of games to divide by).

Bottom line: it's a pretty useless stat.

I agree, but if it is useless and so tricky to calculate, as this conversation here proves, why not just get rid of it?

For the same effort to get rid of it, they could probably just fix it.

[tbhp]

It's always going to be problematic when you try to mix in winrates of different player counts.

If someone who only plays 4-p games had a 25% winrate because they win them 25% of the time, people would complain that the number is misleading too.