Wednesday, 11 February 2015

Good tie-breaks

Lets start with the idea that there is no perfect tie-break system. Some are probably better than others, but not all tie-breaks work the same way in different situations. Having been involved in various technical discussions about tie-breaks over the years, what I have mainly learned is that tie-breaks are often a matter of preference, and depend on how you weight the pros and cons.
For example, head-to-head seems to be a fair system. And yet some people object on the grounds that you don't know which games are important and which are not, until towards the end of the tournament. Others like systems like Sum of Progressive Scores, which do not depend on the results of other players, while just as many prefer SOS (Buccholz), on the grounds that you need to measure a players performance using the rest of the field.
One tie-break that I have been utilising recently is APRO - Average Performance Rating of Opponents. This was drawn to my attention by Roberto Ricca (Secretary of the FIDE Pairings Commission), as part of his work for the FIDE Rules Commission. You simply calculate the Performance Rating of each player, and then take the average of everyone that a player has played. On the upside it seems to work ok, even with unrateds (although not too many), as they will also have a TPR, even if it is based on a few games. It also happily deals with bye/forfeits/missing games as it is an averaging system. On the other hand it is difficult to work out on the fly, meaning that players find it difficult to calculate their own tie breaks (NB some organisers think this is a positive).
Whether it catches on probably depends on which arbiters push it. If it turns up in a high profile event, like the European Championships, and solves the tie-break problems that have occurred in the past, then I can see it becoming more popular. Otherwise it may just float around in the background, used for small events, but nothing too crucial.

No comments: