Direct Comparison
Direct Comparison (EGF), also called Face to Face result (AGA), is a tiebreaker used in go tournaments. The simple and most common case is for two tied players. If these players have played each other and one of these players has beaten the other during the tournament, then that player wins the Direct Comparison.
For larger groups, Direct Comparison tries to break ties by the (Swiss) score from only the results between the tied players, but only if they have all played each of the others.
Direct comparison may not break all ties. In groups of four or more players, it may break the tie partially, and can in that case be applied iteratively.
Table of contents |
When to use Direct Comparison
Direct Comparison does not always split ties; its greatest usefulness is in cases of two tied players. It is among the recommended tie breakers of the EGF Tournament System Rules. The AGA considers this tiebreaker acceptable only if the Swiss Score, SOS and SODOS are all tied.[2]
Advantages of Direct Comparison
- Easily understood by players in the case of two tied players.
- Easily calculated manually. (= Doesn't require tournament software.)
Disadvantages of Direct Comparison
- Not always applicable, all direct games need to have taken place.
- Less easily understood in the case of more than two tied players.
- It does not consider performance over the whole tournament, instead considering only some of the games that have been played.
Iterative Direct Comparison
When four or more players are tied, it is possible that direct comparison partially breaks the tie (i.e, some players are still tied, but not all players). In that case, it is possible to iterate Direct Comparison by applying it again to the players that are still tied. This is called Iterative Direct Comparison in the European Go Federation Tournament System Rules.
How to do Iterative Direct Comparison.
Iterative Direct Comparison needs to be done in a prescribed way. some shortcuts that superficially seem okay are in fact mathematicaly unsound because game results are intransitive
The right way to use is to make a virtual round robin tournament between the tied players from the tournament results. (all results not between the tied players are discarded)
If some games in this virtual tournament haven't been played then this tiebreaker cannot be used.
There are now three possible results:
- One players has more wins than any other player: He is declared the winner.
- All players have the same winscore: This tiebreaker cannot decide on the winner. and other tiebreakers needs to be considered.
- More than one player has the highest winscore: the same procedure is done but now only between the players with the same winscore. (this can also be done to find 2nd, 3rd ... prize winner)
Example
In the European Go Congress 2009 the four top players were all tied with 8 wins.
The Final Wall list was: (unnecessary data removed)
Name R 1 R 2 R 3 R 4 R 5 R 6 R 7 R 8 R 9 R 10 Wins Player 1 14+ 10+ 7+ 5- 25+ 11+ 2+ 6+ 3+ 4- 8 Player 2 27+ 11+ 4+ 3+ 5- 9+ 1- 13+ 23+ 7+ 8 Player 3 33+ 8+ 6+ 2- 14+ 4+ 5+ 15+ 1- 13+ 8 Player 4 24+ 23+ 2- 17+ 12+ 3- 26+ 11+ 10+ 1+ 8
For breaking the tie using IDC only the results against each other are used:
First direct comparison
Name | P 1 | P 2 | P 3 | P 4 | DC-1 | Next |
---|---|---|---|---|---|---|
Player 1 | -- | 7+ | 9+ | 10- | 2 | To comparison for 1st and 2nd |
Player 2 | 7- | -- | 4+ | 3+ | 2 | To comparison for 1st and 2nd |
Player 3 | 9- | 4- | -- | 6+ | 1 | To comparison for 3rd and 4th |
Player 4 | 10+ | 3- | 6- | -- | 1 | To comparison for 3rd and 4th |
(the numbers refer to the round number)
The column "DC-1" is the result of the first iteration: i.e, the Number of Wins Score of each player relative to the tied group.
It is important to check that all games have been played if some games have not been played this tiebreaker can not be used.
The players with the same result are now compared against each other. The players in the highest result group are eligible for the highest prizes. the second highest group for prizes that cannot be awarded to the first group, because there are more prizes than players in that group.
Second direct comparison for first and second place
Name | P 1 | P 2 | DC-2 | Placement |
---|---|---|---|---|
Player 1 | -- | 7+ | 1 | First |
Player 2 | 7- | -- | 0 | Second |
Second direct comparison for third place
Name | P 3 | P 4 | DC-2 | Placement |
---|---|---|---|---|
Player 3 | -- | 6+ | 1 | Third |
Player 4 | 6- | -- | 0 | Fourth |
The second iteration (DC-2) is done between the players with the same score from DC-1 and is calculated using only the results from games between players in the same group. In other words, DC-2 for players 1 and 2 is the Number of Wins Score relative to the group {Player 1, Player 2}, and DC-2 for players 3 and 4 is the Number of Wins Score relative to the group {Player 3, Player 4}.
See also
- The EGF description of Direct Comparison.
- this post
- The AGA description of Face to Face Result
- http://www.usgo.org/EJournal/archive/20050905.htm this link points out that the AGA do not approve of face to face result
- Robert Jasiek's opinion on the Quality of Direct Comparison (widely disputed[1])
- isd's satirical opinion on the Quality of Direct Comparison
- Bass' opinion on the Quality of Direct Comparison
References
- [1] See discussion at t:2110
- [2] http://files.usgo.org/usgo/interface/howtd.html