Sunday, April 21, 2013

[ybdqefpn] Candidates tiebreaker

The recent Candidates chess match used a tiebreak system of 1. Head-to-head result, 2. Number of wins, 3. Sonneborn-Berger, 4. Rapid playoff match.

By putting Sonneborn-Berger so late in the considerations, it seems that it reduces the probability of a kingmaker, a player eliminated from contention being able to determine the outcome of the tournament by throwing a game.

Test this assertion by simulation.  The simplest case is to focus on the final round only, testing whether in any game in that round someone can be a king maker.  We can also look at earlier rounds, a probabilistic king maker, with players who have already been eliminated with several rounds left to go.  Or probably eliminated.

Is there another round-robin tiebreak system that has even less probability of collusion and fraud? Wins as black has been used in tournaments such as Linares.

No comments :