- Add a "Win Rate" feature as part of the site, or a user script. Really simple to code, especially since we have code for map rank already.
Specifics:
- Here is how the percentage system could be adjusted:
Games won with Y players (2v2 counts as 2 player, etc.) x Y x Relative Rank for those games
Add all of these up and divide by total # of games that person has played
This gives us a percentage that shows how far one of us is above or below the normal. For example, a 1.00 would mean that you win as many games as probable. 2.00 would mean that you win twice as many games as probable. - NEW FEATURE (edited into the original post): Terminator games will be counted as "number of opponents eliminated / ( total players in the game - 1 ) " This way, if you eliminate all people, you get full credit, but you get credit for kills, not winning. This simply means that you double the equations and run the first ones on standard-assassin, and the next batch on terminator.
- Imagine a player who does the following:
Has won 750 out of 1000 two player games against people of half his rank.
Has won 10 out of 100 six player games against people of his own rank.
Plays no other games for simplicities sake of this calculation (seeing as I have accidentally deleted this entire post twice already)
His win ratio is currently 69% (Good, right? If you did not know anything else about him)
His Win Ratio would be the following:
( 750 (2 player games won) x 2 x .500 + 10 x 6 x 1.00 ) / 1100 (Total Games played) = .74
What does this mean?
Well, if all games were determined by luck, and the score system was accurate, then everyone's win ratio would be exaclty 1. However, some people are better than others, and so you do not win exactly 50% of your two player games. So what this statistic does is take the normal win percentages, but take into account that you are more likely to win a 2 player game than an 8 man one, that you might play more 2 player than 8 man games, and that you might be farming (which only affects your score at the moment, and helps your win rate).
My equation would say that the person wins only .74 of the games he should win. Take the next example: - A player has the following statistics:
Has won 25 out of 100 two player games against people double his rank.
Has won 33 out of 99 three player games against people of his own rank.
Has won 25 out of 100 eight player games against people of half his rank.
25 x 2 x 2.00
33 x 3 x 1.00
25 x 8 x .500
Added and Divided by a total of 299 games played = 1.00
Although he wins only 25% of his games in two player, he plays against people of double his skill, so it does not matter. Although this player farms on 8 player games, his win rate is not positively affected by that 25% win percentage on them. He wins double the games he should against people of half his rank, half the games he should agains people double his rank, and the exact amount of games he should win against people his own rank. Therfore, his win rate is exaclty 1.00 - Before, the player would have had a win percentage of 28%. This stuff tells you much more that simply 28%, doesn't it? I suggest you run it on yourself, even if it does not get updated to the site. I am a major right now, but my win rate is only 1.08, so not that many more games won than I should win, is it?
This will improve the following aspects of the site:
- A way more accurate statistic than win percentages
- Low rank farming will no longer positively affect your win rate (let's face it, a lot of them are really good at it. Possibly square the relative rank?)
- Another statistic. Always good.
- I was kind of interested to find out about this myself. Run it on myself every so often, interesting to see how good you actually are rather than how many points you can accumulate.
