This is a simple average of all the ratings you currently hold.

This average is weighted by the number of games you have played.

This rating ranks players based on the percentage of their games that they win.

This ranks players based on the total number of wins recorded.

This statistic is more complicated. Essentially, every rating you hold is examined. Based on your standing for a given game, that rating is “normalized” to a number between 0 and 1. Your score for the game is actually a fraction of this normalized rating based on the number of games you have played. Your total Well-Rounded score is the sum of these per-game scores. This score is most meaningful with a wide player base for each game. The maximum score you can ever have is just a little less than the total number of different games you have played. (Eg. If you only play Chess, then you will never get a score equal to or greater than 1.0. If you play 6 different games, your max score would be just under 6.0.)

Here's how it looks in (pseudo-)pseudo-code:

  totalscore := 0
  FOREACH game
    min := 15thPercentile(game.allratings)
    max := 85thPercentile(game.allratings)
    div := max - min
    FOREACH rating
      rating := min if (rating < min)
      rating := max if (rating > max)
      normalized := (rating - min) / div
      score := normalized * (1 - (1 / numGamesPlayed))
      totalscore := totalscore + score
    END
  END

This ranks players by their highest rating. Ties are broken by looking at how many different games they hold this rating in.

This rating is only for games. It looks at how many times the first player wins as opposed to the second player. The ideal parity is 0, meaning there is no difference in the odds of the first or the second player winning. A positive parity represents a percent advantage held by the first player. A negative parity represents an advantage to the second player.

This simply lists each and every registered game and the player who currently holds the highest ELO rating.

IAGO's ELO algorithm has been modified to allow for more than 2 players, or even games with multiple teams of uneven sizes.

A game with 3+ players is treated as a series of duels:

1. Start with players A, B, C, and D
2. The final result is: B, C, A, D
3. For rating, you get the following:
   • B over C
   • B over A
   • B over D
   • C over A
   • C over D
   • A over D
4. You process each of these duels using standard ELO
5. The final ELO is the average of the individual ELOs

Here are some 4-player examples:

Start:
---
- 1200
- 1200
- 1200
- 1200


Final:
---
- 1216
- 1205.33333333333
- 1194.66666666667
- 1184


Deltas:
---
- 16
- 5.33333333333326
- -5.33333333333326
- -16

Start:
---
- 1200
- 1400
- 1400
- 1150


Final:
---
- 1220.77899451613
- 1399.27397760856
- 1388.60731094189
- 1141.33971693341


Deltas:
---
- 20.7789945161337
- -0.726022391439074
- -11.3926890581058
- -8.66028306658882

With teams of multiple players, you perform the exact same calculation except the team's starting ELO is the average of all the team members' ELO ratings. Once all the delta's are calculated, they are applied equally to each member of the team.