Latest revision as of 13:44, 28 February 2015

With the return to a playoff model in the 2015 North American SCRABBLE Championship after several years of choosing winners based only on final standings after modified Swiss play, it is instructive to consider how past National Championships would have played out had they included playoffs. In what follows, we use a simulation methodology.

Summary

If we had had playoffs in 2009, and Nigel Richards had played to expectation in them, he would most likely have successfully defended his title rather than finishing in second place to Dave Wiegand in Dayton.

In 2010, he would have been favoured to win in Dallas, as actually happened.

In 2011, because of his lifetime 1-6 record vs. Joel Sherman, Richards would only be the third most likely playoff champion, behind Sherman and Kenji Matsumoto, rather than winning the title for the third time.

In 2012, because of David Gibson's historical domination of the rest of the semifinalists, Gibson would have been much more likely to make the finals than Richards, and therefore more likely to end up victorious than losing in the final round to his evenly matched opponent.

In 2013, facing a more even field, Richards would have been most likely to duplicate his actual win, thanks in part to fortuitously being paired to eliminate Komol Panyasophonlert in the quarterfinals.

In 2014, we would most likely have seen an exciting finals match between 3-time champion Joe Edley and actual 2014 champion Conrad Bassett-Bouchard, with Joe Edley favoured to prevail.

In each case, the actual champion makes it into the quarterfinals, never seeded less than 5th place, and in five of the six years examined, the actual runner-up makes it into the quarterfinals too. The quarterfinals typically consist of four or five players who would have finished at the top of the field, with the rest having finished barely out of the money.

The effect of returning to a playoff format is therefore to increase the chance for players who typically finish in the top 20 to become the champion, while preserving the ability for the most deserving experts to reach the playoff stage.

Methodology

Historical records were reviewed to determine Division 1 final standings and standings after Round 21. Head-to-head player records were used to estimate the probability of game outcomes in playoff matches, using historical win ratio where at least three head-to-head games were recorded, game point ratio where only one or two games were recorded, or rating-based estimates where there was no record of the players playing each other. The probability of a player advancing through a best-of-4 or best-of-5 playoff round was computed using elementary probability theory.

Details