Description: Showing a player's percent games won is misleading since that player could be playing anywhere from 1 to 7 opponents on average, so if this player is equally matched by his opponents he could end up with anywhere from a 13% to 50% win rate. Conversely, a player with a 50% win rate could be demolishing his opponents, or just as good as them. A value that compensates for the typical number of opponents that player faces per game would be much more informative.
Details: The name doesn't have to be "win rate", but whatever it is the point is that you give a number which represents not how often a player wins, but how often this player wins relative to how often he would win were his opponents all exactly as skilled and lucky as himself. So if the player always plays 4-player free-for-all and wins 25% of the time, their win rate would be equal to a player who always plays 1v1 and wins 50% of the time. The math is simple: games_won / games_played * avg_teams_per_game yields 1 if the player is usually equally matched, 0 if the player has never won, and is higher for players who more often outmatch their opponents.
Benefits: This is *not* the same thing as score - score also accounts for opponents' scores, order the games are played in, etc. I believe this number would make the score more meaningful - if a player has a 3.00 "win rate" as calculated above, 500 games, and a score of 2000, then they are probably farming. If they have a 3.00 win rate, 100 games, and a score of 1500, then they are probably very good at what they play. If they have a 1.05 win rate and a score of 2000 they are probably moderately skillful and very persistent. If they have a 1.00 win rate, 50 games played, a score of 1500, and joined two months ago, then chances are they are as good as your average colonel but won't earn major for another year. I could go on, but I think I've made my point - win rate is useful, and is a drop-in replacement for percent won, which is not.
I apologize if this has been suggested before - I tried searching for it, but admittedly I could have tried harder.
Last edited by /boost
on Wed Jul 13, 2011 12:10 am, edited 1 time in total.