This algorithm still achieves the goal (stated in https://support.chess.com/customer/portal/articles/1444961-how-is-seeding-and-group-placement-determined-in-multi-group-tournaments-) of the best players generally making the later rounds because they don't tend to run into each other too early.
My suggested (snaking) algorithm is often used in sports. In the NBA or NFL playoffs, the top team plays the bottom team, the second top plays second bottom, etc. Now, that's groups of two, but easily generalized to larger groups.
Competitions like the Champions League in Europe use an alternative method - seeding teams, then allocating them randomly - to accomplish the same purpose. That would be a decent option too.
Tournament allocation of players to groups is flawed
Sort:
I never knew chess.com tournaments followed the first algorithm. That definitely is flawed. I support the snaking one.
In your example thre are 4 groups. How is the number of groups or number of players in a group determined? In the "Create A New Tournament" form I can see no provision for the tournament director (TD) to specify the number of groups or the number of players in a group. There is a place to specify the maximum number of players in the tournament (Max Players) but that is all (from what I can tell). This may be due to the fact that the choices in the form for me are limited, as I am a Gold Member, whereas higher membership levels have more choices/options.
Originally posted in http://www.chess.com/forum/view/tournaments/allocation-of-players-to-groups-in-tournaments-is-flawed, but I think this is actually the correct place:
The algorithm used to allocate players to groups within a tournament is flawed, resulting in some groups being considerably stronger or weaker than others.
For a tournament with sixteen players in four groups, the current algorithm is as follows:
Group 1
Player 1
Player 5
Player 9
Player 13
Group 2
Player 2
Player 6
Player 10
Player 14
Group 3
Player 3
Player 7
Player 11
Player 15
Group 4
Player 4
Player 8
Player 12
Player 16
Notice that the first group is considerably stronger than the last, having the best player instead of the fourth, the fifth instead of the 8th, the 9th instead of the 12th, and the 13th instead of the 16th.
This means that if you are the 4th best player, you have an excellent chance of winning your group - approximately as good a chance as the best player does. If you are 5th best, your chance of winning your group is immensely lower, as you have to beat out the best player.
Being one spot lower in the rankings has no negative impact on your chance of winning the group as long as you don't drop into a lower tier (from being best in the group to second best), as being in a lower group means you just have weaker opposition to match. But drop down from 4th to 5th, from 8th to 9th, or from 12th to 13th, and it suddenly gets much tougher. This is especially true in tournaments with a large number of groups.
One could also imagine somebody losing a game or two deliberately to be dropped into an easier group.
Instead of sequentially allocating players to the different groups, a "snaking" algorithm would be much better:
Group 1
Player 1
Player 8
Player 9
Player 16
Group 2
Player 2
Player 7
Player 10
Player 15
Group 3
Player 3
Player 6
Player 11
Player 14
Group 4
Player 4
Player 5
Player 12
Player 13
Now all the groups have roughly equal strength. Comparing groups 1 and 4, the best player in group 1 is better, but the second best player in group 2 is better.
Also, the 4th and 5th best players now have a roughly equal chance of winning their group, with the 4th best having the edge. Being one spot lower in the rankings now has a small negative impact on your expected change of winning the group. That's as it should be.
Also, the only way to have an easier group is to be a stronger player and move up the ratings; group 1 is roughly as tough as group 4, but Player 1 certainly has an easier time of it than Player 4, because Player 1 is the reason that group is tough, and he doesn't have to play against himself. So there's no longer an incentive to throw a game to get an easier group.