1985-1990: Nick Inglis gradually developed the first ratings program based on the Elo ratings system used in chess. As he was Winking World editor, he got sent the scores of all tournaments, and so he was in a position to enter all of them into his BBC microcomputer (starting with the ETwA Singles in 1985). The results were initially only reported to the Cambridge club, and the algorithm was refined as the number of tournaments played increased. By 1990, the program was stable and reliable numbers for most players were produced. At that time, ratings began to be reported in Winking World and thus became available to the whole world of winkers.

1990-1995: Nick continued to maintain the ratings. However, the number of players and scores in the database gradually became too unwieldy for the original program, and several quirks began to become apparent. By 1996, other pressures on Nick meant that he no longer had the time to devote to maintaining and improving the ratings program.

1996-1997: Tim Hedger resurrected and took over the ratings program. He used essentially the same calculation algorithm as Nick, but converted the database into nice friendly Microsoft Access format. This is quite important, as the database currently has over 14,000 game scores involving more than 500 players. However work pressure meant that he became unable to keep them updated from 1998.

Jun 1999-Dec 1999: Patrick Barrie volunteered to take over maintenance of the ratings in June 1999. He decided to implement a new calculation algorithm in an attempt to eliminate some of the anomalous results that Nick's and Tim's program occasionally produced. In particular, he introduced the concept of a Ratings Reliability Factor (RRF) into the calculation, and removed the concept of "rated games". (In Nick's and Tim's programs, games involving players who hadn't played in the previous year were only used to rate the unrated player, and didn't influence that player's partner's rating).

Jan 2000-Jul 2002: Various changes were made to give a sounder statistical basis to the calculation of ratings changes for each player in a tournament. The idea was that a valid statistical approach would lead to a method that would work sensibly for all the different tiddlywinks formats, even allowing for the fact that many tiddlywinks players have only played in a limited number of tournament games.

Jul 2002: All ratings were recalculated using the new algorithm taking
into account all games since November 1985. A paper describing the
calculation method was published in the *Journal of Applied Statistics*
in 2003.

There is an article describing the current algorithm in detail.

Back to the ETwA ratings page. |