Re: ECF Grading Proposals
Posted: Mon Mar 04, 2019 8:11 pm
No one has a constant playing strength - our performance changes regularly. No one should be given grading points for free - we need to earn them by our performance.
The independent home for discussions on the English Chess scene.
https://www.ecforum.org.uk/
Underlying playing strength certainly can be constant, performance varies around it. Why should a player be rewarded solely for variability of performance?Maurice Lawson wrote: ↑Mon Mar 04, 2019 8:11 pmNo one has a constant playing strength - our performance changes regularly.
But that is precisely and deliberately what this isn't.Maurice Lawson wrote: ↑Mon Mar 04, 2019 8:39 pmBecause that's exactly what ratings should be about! Measurement of performance!
By the same token, why should a player be punished solely for variability of performance?NickFaulks wrote: ↑Mon Mar 04, 2019 8:36 pmUnderlying playing strength certainly can be constant, performance varies around it. Why should a player be rewarded solely for variability of performance?
It is freely acknowledged that initial ratings are the one area of the current FIDE rating System where significant improvement is required. No simple solutions seem to help, and there may be no alternative to something more complicated. This is a major decision.David Sedgwick wrote: ↑Mon Mar 04, 2019 9:18 pmBy the same token, why should a player be punished solely for variability of performance?
After spending many months looking at the data, I think the hypothesis that juniors tend to improve is conclusively proved against the alternate that playing strength is constant. Based on this conclusion it then becomes a question of how to recognise this in the system where a lot of junior results are against other juniors. Earning the grade improvement seems to be a good idea as opposed to "just give them extra points".NickFaulks wrote: ↑Mon Mar 04, 2019 7:35 pmMy impression based on a few seconds of mental arithmetic is that a junior whose playing strength is constant and who plays ten games each month might be expected to gain 100-150 points per annum from this rule. If that is our desired result, why not just give them the extra points and be done with it?The k factor will be 20 except that in a month where a junior player has outperformed expectation, then the k factor will be 40.
On the other hand, you have high-graded juniors leaving the system at age 18, and low-graded juniors joining the system very young. So don't they cancel each other out?Kevin Thurlow wrote: ↑Tue Mar 05, 2019 8:44 am"After spending many months looking at the data, I think the hypothesis that juniors tend to improve is conclusively proved against the alternate that playing strength is constant."
Are you sure? Admittedly it was many years ago, but I took a small sample of juniors (about the first 30 pages of the grading list) and discovered that on average juniors went up slightly. Obviously some went up (some dramatically so), some went down, but what tended to happen is that the ones who started to get lower grades, gave up, therefore removing themselves from the calculation.
I am sure you thought of that!
The key point, as perhaps established by the BCF graders 50 years ago when 5 and later 10 points started to be added to junior grades as directly calculated, is that when playing adult players, there needs to be an inflationary measure so that the players in the pool of adults don't find their grades reducing for no good reason as at least some of the juniors are going to have grades below their current playing strength. Most national Elo systems have felt the need to adjust for this in some arbitrary manner. Even the international system attempts to do so with the K=40 rule.Alex Holowczak wrote: ↑Tue Mar 05, 2019 9:59 amOn the other hand, you have high-graded juniors leaving the system at age 18, and low-graded juniors joining the system very young. So don't they cancel each other out?
If your are suggesting this was intended as an inflationary measure, then you are mistaken. The k factor represents a tradeoff between two risksRoger de Coverly wrote: ↑Tue Mar 05, 2019 10:17 amEven the international system attempts to do so with the K=40 rule.
Being sure is not a concept fitting with statistics theory!Kevin Thurlow wrote: ↑Tue Mar 05, 2019 8:44 am"After spending many months looking at the data, I think the hypothesis that juniors tend to improve is conclusively proved against the alternate that playing strength is constant."
Are you sure? ....