Closed cscorley closed 6 years ago
Currently, the suggestion algo uses a three step process based on player rank. Assuming a set of 4 teams with 5 players each, ranked 1..20:
(1, 2, 3, 4) (5, 6, 7, 8) (9, 10, 11, 12) (13, 14, 15, 16) (17 18, 19, 20)
(1, 2, 3, 4) (8, 7, 6, 5) < (9, 10, 11, 12) (16, 15, 14, 13) < (17 18, 19, 20)
(1, 8, 9, 16, 17) (2, 7, 10, 15, 18) (3, 6, 11, 14, 19) (4, 5, 12, 13, 20)
This is fine, as it produces teams with averages within 1 point of another.
(1, 8, 9, 16, 17), mean = 10.2 (2, 7, 10, 15, 18), mean = 10.4 (3, 6, 11, 14, 19), mean = 10.6 (4, 5, 12, 13, 20), mean = 10.8
However, we'd really prefer if we could pair the worst players with the best players. An algorithm for this could work on a set of 20 players as such:
Pool: (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17 18, 19, 20)
(1) (2) (3) (4) Pool: (5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20)
(1, 20) (2, 19) (3, 18) (4, 17) Pool: (5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16)
(1, 20, 5) (2, 19, 6) (3, 18, 7) (4, 17, 8) Pool: (9, 10, 11, 12, 13, 14, 15, 16)
(1, 20, 5, 16) (2, 19, 6, 15) (3, 18, 7, 14) (4, 17, 8, 13) Pool: (9, 10, 11, 12)
(1, 20, 5, 16, 9) (2, 19, 6, 15, 10) (3, 18, 7, 14, 11) (4, 17, 8, 13, 12) Pool: (empty)
This should produce teams that are equivalent in terms of means, but instead explicitly pairs the worst with the best:
(1, 20, 5, 16, 9), mean = 10.2 (2, 19, 6, 15, 10), mean = 10.4 (3, 18, 7, 14, 11), mean = 10.6 (4, 17, 8, 13, 12), mean = 10.8
Another alternative to the current approach is to always flip the last group.
Currently, the suggestion algo uses a three step process based on player rank. Assuming a set of 4 teams with 5 players each, ranked 1..20:
This is fine, as it produces teams with averages within 1 point of another.
However, we'd really prefer if we could pair the worst players with the best players. An algorithm for this could work on a set of 20 players as such:
Pool: (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17 18, 19, 20)
This should produce teams that are equivalent in terms of means, but instead explicitly pairs the worst with the best: