January 2012 ECF Grades

As you are no doubt aware, the preliminary grading list has been published, with the welcome inclusion of the underlying game results data (from 2003 onwards) and a snazzy chart to boot.

In-form David Weston deservedly climbs the Trowbridge table, and he is now the highest graded of our currently active players. Also improving is Dave Marshall, whose grade now more accurately reflects his playing prowess. Louis Burton continues his inexorable climb to the world of 3 digit ratings and Anthony Ransom is also up a couple of points. However everyone else has suffered a set-back. Roy Ludlow in particular appears to have been hard done by. He had below-par results in 2011 and I'm sure he will bounce back in the next grading list.

Indeed it is surprising how easily a handful of results can affect the rating calculations. So that you may judge the accuracy of the ECF's new grading assignments, I have reproduced the January figures below. In the right-most columns I have compared the January 2012 grades with the known performance of each player, using the data available on the Trowbridge website (up to and including the 26th of January).

	2011	Standard ECF	Performance	Diff
David Weston	149	155	162	-7
Mark Leonard	153	151	154	-3
Derek Rothwell	152	147	144	3
Dave Marshall	141	143	149	-6
Tim Woodward	155	137	143	-6
Gareth Williams	131	129	128	1
Roy Ludlow	131	111	120	-9
Louis Burton	82	91	92	-1
Anthony Ransom	73	75	73	2
Nigel Crucefix	71	67	64	3
Richard White	ung	59	56	3

Of the 14 games I've played this season, 11 took place in 2011, but only 6 have been submitted for grading. For some reason the Wiltshire league and the Frome internal club games do not appear in the list of interim results. The grade given to me by the ECF happens to be 3 points less than my actual performance over the last 30 games.

I think the conclusion we should draw is that the ECF grades should be taken with a pinch of salt. They are just numbers, after all, and are not meant to be viewed as an absolute indicator of someone's playing strength. I've long held the opinion that the values should carry a confidence interval of ±10, and this fits the numbers above rather well.