Forged Alliance Forever

Re: Why are ladder map stats wrong?

2013-05-26T17:14:10+02:00

This is the only viable way to count them, as we have 4 factions, not 2.

Statistics: Posted by Ze_PilOt — 26 May 2013, 17:14

Re: Why are ladder map stats wrong?

2013-05-26T10:37:21+02:00

I did some calculations. Aren't the games in the win ratios being counted twice? If you have a UEF vs Cybran, naturally, one of the players will contribute to the win ratio and one will contribute to the loss ratio of their respective faction. If you add the number of no mirror matchup games, you get 1750 of them (28% of the games are mirrors, which is consistent with the chance of a mirror matchup if we assume the factions played are distributed evenly across the entire leaderboard) and 802 of them have been won.

This would be correct, because if it's higher than 50%, at least one game should have resulted in a win for both players. By this counting method, there are 73 games that resulted in a draw, which is 3% rather than 0.03% (You sure you read the latter, FunkOff?)

In short, all wins and losses get counted from all games and all games get counted twice: one time from player A and one time from player B.

Statistics: Posted by Plasma_Wolf — 26 May 2013, 10:37

Re: Why are ladder map stats wrong?

2013-05-26T08:37:52+02:00

Blasted Rock

2432 games played on this map [...]

34.25 % UEF (833 occurrences)
38.08 % Cybran (926 occurrences)
11.97 % Aeon (291 occurrences)
15.71 % Seraphim (382 occurrences)

Win ratios :
UEF : 43.83 % (245 games won in 559 no mirror matchup games)
Cybran : 49.48 % (286 games won in 578 no mirror matchup games)
Aeon : 43.68 % (121 games won in 277 no mirror matchup games)
Seraphim : 44.64 % (150 games won in 336 no mirror matchup games)

of these 833 uef games, 559 appear to be uef vs another faction (the others are obviously mirrored then), but i agree that it's weird that the factions don't end up with an average of 50% win

Statistics: Posted by Myxir — 26 May 2013, 08:37

Why are ladder map stats wrong?

2013-05-26T05:54:03+02:00

I think this is something worth looking into. I haven't any guesses why they are wrong, but they clearly are. Just look at them:
Blasted Rock: All factions have under 50% win despite 0.03% draw rate? And there have been ZERO mirrors in over 2000 games? Those things seem to defy logic.
Canis river has the exact same issues, as does vale of Isis.
NO maps have ANY recorded mirror match-ups! that's wildly improbable!

Statistics: Posted by FunkOff — 26 May 2013, 05:54