Daily tournaments: improve allocation of players to groups

Sort:
Avatar of MGleason

Currently, when dividing up players into groups, they are sorted by rating and then allocated as follows (using a 16-player tournament with groups of four to illustrate):

Group 1 Group 2 Group 3 Group 4

Player 1
Player 5
Player 9
Player 13

Player 2
Player 6
Player 10
Player 14

Player 3
Player 7
Player 11
Player 15

Player 4
Player 8
Player 12
Player 16

 

This has the rather obvious effect of making Group 1 considerably stronger than Group 4.  The difference could even encourage small-scale sandbagging, as Player 5 might try to drop a couple games to chop his rating down so he becomes Player 8.

 

A better method that would result in all groups being about the same average strength would be to use a snaking system:

Group 1 Group 2 Group 3 Group 4

Player 1
Player 8
Player 9
Player 16

Player 2
Player 7
Player 10
Player 15

Player 3
Player 6
Player 11
Player 14

Player 4
Player 5
Player 12
Player 13

 

This system will result in all groups being roughly equivalent.  It also discourages sandbagging, as, if anything, you'll tend to get a stronger group.  For example, if Player 5 drops down to become Player 8, the bottom two members of the group will average about the same, but his top opponent will now be Player 1 instead of Player 4.

Avatar of Hitchy04

I think players should be in the group with other people with about the same strength/rating

Avatar of MGleason
11dex-11 wrote:

I think players should be in the group with other people with about the same strength/rating

That's not really ideal.  If you have 16 players rated from 1000 to 2000, you don't want all the 1800+ players to go in one group, the 1500-1800 to go in another group, etc.  That results in weaker players going through to the next round while strong players miss out.

Pairing strong players with weak players in the early rounds will tend to result in the stronger players making the later rounds unless a weaker player earns it by pulling off some upsets.