Since an earlier notice was posted here (Possible Changes to Future ECF Grading Lists), an outline of the proposal on ECF Grading changes has now appeared on the ECF website at

https://www.englishchess.org.uk/monthly-grading-proposal/, with some further comment to be found at

https://www.englishchess.org.uk/monthly-grades-faq/.

The possible changes relate to three independent considerations:

1) frequency of publication of a “list” (monthly rather than present 6-monthly lists),

2) method of calculation of grades (ELO-style rather than present ECF-style),

3) number of digits (4 rather than present 3).

1) Recognition is given to the problems possibly encountered by organisers needing to get results up to date and reported every month. This would be eased by results being taken directly from LMS (meaning data enterers might need to be no alert regarding the identity of apparently “new” players, as an ECF “Grader” would no longer double-check this independently), but there are still many non-league events, especially congresses, which are reported by other means, which is where the main problem might lie.

2) Unlike the present ECF system, an ELO-type method would not involve retention of earlier results, but instead apply “statistical” methods to the current grade, comparing actual performance in the new batch of results against an “expected” performance based on the previous grade. As regards “accuracy”, each system works better in certain cases than in others, such as juniors improving rapidly (or not), players with periods of inactivity, and so on. No grading system is wholly “accurate”, and different systems are differently “inaccurate” Operationally, the problem of lateness of reporting of results can be better handled by the ECF-style system. The proposal recognises that this is a problem, and envisages initially recalculating each month of a grading season in each new month, of the grading year, in order to knit in the late results correctly. As the system bedded in, that flexibility might be lost, increasing pressure on the dwindling number of data processers, who sometimes go on holiday or are taken ill.

3) ELO-style calculation does not have to be based on 4 digits, but 4 digits afford easy comparison with other ELO-style grades/ratings, assuming they are comparable as regards value of the grading point, so to speak. With 4-digit ECF grades/ratings there will still be players with two grades asking why their ECF grade/rating is not the same as their FIDE rating!

Grading is of course an attempt to numerically quantify something that is not actually defined, i. e. “playing strength”, and people tend to lose sight differences between current on the one hand short-term “form” and on the other hand “average strength” over a period of time, the latter being more appropriate for various organisational purposes. ELO-style systems are geared more to the former, and ECF-style more to the latter.

Steve Mann

16/02/2019