Wednesday, 17 October 2012

A proposal for a new tie-break system

GM David Smerdon has been a busy boy lately. Apart from holding down Board 1 for Australia at the last Olympiad, he is also working towards his PhD in Economics. As part of his studies he has been looking at the fairness and accuracy of various tie-break systems, and is proposing an alternative system.
His proposal is based in the concept of  'indirect wins' which relies on a transitive set of results (ie If A beats B, and B beats C, then you can assume that A would have beaten C). . So a tiebreak score would be the number of direct wins, plus the result of a function of 'indirect wins'. In the case of 'indirect wins', there would be a scaling factor applied, as assumed results aren't as good as real ones. At this stage part of his proposal is to investigate tournament data to see what would be the best estimate for a scaling factor (shades of Duckworth-Lewis!).
David has posted a link to his proposal paper, as well as some thoughts on the topic, in this article at Chessvibes.

1 comment:

Denis Jessop said...

Dear Oh Dear! The idea that if A beats B and B beats C then A should beat C is about the oldest fallacy in the book. How can David be serious? It's good to know also that he has a fine sense of humour.