I recently made a Discord bot for a leaderboard around a game called PUBG. I made my own fun "score" to promote team-based stats (versus just win rate) but I believe I have a fatal flaw in my desired logic.
My intention is to use three stats and take their average to give you this score. The stats are your win rate, KDA (kill death assists), and your teammate revival. All of these being your average per game.
I planned on weighing:
- Win rate at 3x
- KDA at 1x
- Teammate revival at 0.25x.
I also at the end multiply by 10,000 to make the number look more leaderboard-esque versus a single digit number.
My problem, is that I believe KDA is getting a massive sway over this average as you will see in my example below.
Player 1 has a score of: ~6,570
Player 2 has a score of: ~4,220
Player 1's stats and current calculations are:
Win/Matches (0/ 31) = 0(0%) Weighted at 3x = ~0.000
KDA (54 / 16 / 31) = ~1.914
KDA as ((Kills+Assists/3)/Deaths)
Teammate revives per match: (7 / 31) = ~0.226 Weighted at 0.25x = ~0.057
Winrate + KDA + Revives is divided by 3 (for the three factors) to get an average of: ~ 0.657.
We multiply it by 10,000 for a total score of: ~6,570
Player 2's stats and current calculations are:
Win/Matches (3 / 98) = 0.030 (3%) Weighted at 3x = ~0.090
KDA (93 / 45 / 98) = ~1.102
KDA as ((Kills+Assists/3)/Deaths)
Teammate revives per match: (29 / 99) = ~0.293 Weighted at 0.25x = ~0.073
Winrate + KDA + Revives is divided by 3 (for the three factors) to get an average of: ~ 0.422.
we multiply it by 10,000 for a total score of: ~4,220
From looking at this Player 2 has a better win rate than Player 1, yet Player 1 has a far better score. I noticed that both the revive and winrate stats are numbers such as 0.0xx, while KDA is a 1.xxx number. This means KDA will always have a much stronger sway on the combined score, when really it should be the win rate.
My question, is how can I go about correcting my calculations to meet my desired weights? Is there a mathematically correct way or is it just to start inflating winrate to a point where it looks like it is getting the 3x weight it deserves.
Background:
These stats are based off 4-man squads in a 100-player battle royal. Only 1 team per round can win (so roughly 25 teams per match). Maybe this will help in figuring out a better score/weight for my use. As maybe giving win rate a 3x weight isn't doing enough justice for how hard it is.
Thank you!