ECF Grading Proposals

General discussions about ratings.
Nick Burrows
Posts: 1704
Joined: Sat Aug 14, 2010 12:15 pm

Re: ECF Grading Proposals

Post by Nick Burrows » Sun Aug 25, 2019 2:37 pm

Alex Holowczak wrote:
Sun Aug 25, 2019 2:15 pm
I do think it's a barrier to some people playing over the board. If they knew the sections were U1500, U1750 and U2000, I think a few people would feel a bit more comfortable.

So actually I view it as a change that would make it more user-friendly for people to join over-the-board chess in the first place. Online chess is an area where people in England are playing chess, and I think it would be good if some of them transitioned to playing over the board. So this change is something that makes no difference to the existing playing base at a practical level, while potentially being more welcoming to new players.
I really don't see why you would have an interest in playing otb chess and then decide against it because one aspect of it is new. The whole experience is new. Their online grade will be 4 figures but that is the only comparison, as they will be quite different anyway.

The million dollar question is how accurate will the new system be? If it fails, it will be a disaster, as accurately measuring progress is an enormous part of the whole allure of chess, and so there is the potential of it making a massive practical difference to the existing player base.

I really hope it works.

NickFaulks
Posts: 8452
Joined: Sat Jan 02, 2010 1:28 pm

Re: ECF Grading Proposals

Post by NickFaulks » Sun Aug 25, 2019 5:17 pm

Nick Burrows wrote:
Sun Aug 25, 2019 2:37 pm
I really hope it works.
I can think of only two factors here. These are the algorithm being used and the timeliness with which results are reported - including never. If these are known, the accuracy of the resulting grades ( judged by predictive power ) can itself be predicted with confidence.
If you want a picture of the future, imagine a QR code stamped on a human face — forever.

Kevin Thurlow
Posts: 5803
Joined: Wed Apr 30, 2008 12:28 pm

Re: ECF Grading Proposals

Post by Kevin Thurlow » Sun Aug 25, 2019 6:56 pm

"The million dollar question is how accurate will the new system be?"

er, it won't be "accurate". No chess grading system is. The system gives a measure of past performance, but there are far too many variables. Going back nearly 50 years, my grade dropped one year, but I knew I was playing far better than the previous year, and had played stronger opposition. You can't just believe the numbers. Richard Clarke warned of this when he created the system.

But gradings/ratings provide a useful guide to past performance.

Nick Burrows
Posts: 1704
Joined: Sat Aug 14, 2010 12:15 pm

Re: ECF Grading Proposals

Post by Nick Burrows » Sun Aug 25, 2019 7:41 pm

All rating systems are measures of past performance. The debate is about relative accuracy. At my rating level the Ecf system is more accurate than the Fide system. I am quite reasonably concerned about breaking the only measure that currently works for imo a very marginal upside.

NickFaulks
Posts: 8452
Joined: Sat Jan 02, 2010 1:28 pm

Re: ECF Grading Proposals

Post by NickFaulks » Sun Aug 25, 2019 8:05 pm

Nick Burrows wrote:
Sun Aug 25, 2019 7:41 pm
The debate is about relative accuracy. At my rating level the Ecf system is more accurate than the Fide system.
Nick, over the past year 23 of your games have been submitted for FIDE rating. How would you expect to get an accurate result out of that?
If you want a picture of the future, imagine a QR code stamped on a human face — forever.

J T Melsom
Posts: 1294
Joined: Wed Aug 12, 2009 11:12 pm

Re: ECF Grading Proposals

Post by J T Melsom » Sun Aug 25, 2019 9:12 pm

Somebody up-thread explained at great length why the mantra that 'progress is good' is a nonsense. I agree but (I) I'm not sure that any of the modernisers believe this to be self evident and (ii) conservatism has its own pitfalls. You can become a ranting dinosaur. Its really all about management of change. The current way of doing chess in the broadest sense may have its merits but if it doesn't adapt it will surely die. And I don't understand why so much of the criticism is directed at the former Director of Home Chess given the policies are being carried forward by his successor. Perhaps its ageism?
Last edited by Carl Hibbard on Mon Aug 26, 2019 8:02 am, edited 1 time in total.
Reason: Name removed due to confusion

David Sedgwick
Posts: 5249
Joined: Mon Apr 09, 2007 5:56 pm
Location: Croydon
Contact:

Re: ECF Grading Proposals

Post by David Sedgwick » Mon Aug 26, 2019 8:58 am

J T Melsom wrote:
Sun Aug 25, 2019 9:12 pm
And I don't understand why so much of the criticism is directed at the former Director of Home Chess given the policies are being carried forward by his successor
To my mind it is to Alex Holowczak's credit that he posts on here when other Directors do not. People criticise him and he defends himself forcefully.

With the possible exception of Mike Truran, criticisms of other Directors have to be made privately if they are to elicit a response.

Kevin Thurlow
Posts: 5803
Joined: Wed Apr 30, 2008 12:28 pm

Re: ECF Grading Proposals

Post by Kevin Thurlow » Mon Aug 26, 2019 9:29 am

"To my mind it is to Alex Holowczak's credit that he posts on here when other Directors do not. People criticise him and he defends himself forcefully."

I agree and even on the occasions I think he is talking complete rubbish, at least I know what he is trying to do!

Nick Burrows
Posts: 1704
Joined: Sat Aug 14, 2010 12:15 pm

Re: ECF Grading Proposals

Post by Nick Burrows » Tue Aug 27, 2019 3:11 pm

NickFaulks wrote:
Sun Aug 25, 2019 8:05 pm
Nick Burrows wrote:
Sun Aug 25, 2019 7:41 pm
The debate is about relative accuracy. At my rating level the Ecf system is more accurate than the Fide system.
Nick, over the past year 23 of your games have been submitted for FIDE rating. How would you expect to get an accurate result out of that?
I agree that this is a large part of the problem. Fide ratings are so slow to capture players real strength, because so many games played in Fide rated tournaments are against unrated or drastically underated opponents so it's a vicious cycle that improves at sub-snail pace.

If all the ecf rated players begin with a converted rating this shouldn't be an issue, but can't help but feel there will be problems ahead.

At least I will be able to pass 200 ecf :P

Alex Holowczak
Posts: 9085
Joined: Sat May 30, 2009 5:18 pm
Location: Oldbury, Worcestershire
Contact:

Re: ECF Grading Proposals

Post by Alex Holowczak » Tue Aug 27, 2019 4:12 pm

Nick Burrows wrote:
Tue Aug 27, 2019 3:11 pm
Fide ratings are so slow to capture players real strength, because so many games played in Fide rated tournaments are against unrated or drastically underated opponents so it's a vicious cycle that improves at sub-snail pace.
I'm not sure which tournaments you play in, but you're based in Oxfordshire? If you play in the 4NCL Congresses or in London, you'll struggle to find an unrated player.

In terms of underrated players, actually the problem is not with the unrated players, but with the conversion formula. On average, it has been overestimating people for 18 months now, but it is not being changed because given our impending move over to a different rating system, there seems little point.

NickFaulks
Posts: 8452
Joined: Sat Jan 02, 2010 1:28 pm

Re: ECF Grading Proposals

Post by NickFaulks » Tue Aug 27, 2019 5:40 pm

Nick Burrows wrote:
Tue Aug 27, 2019 3:11 pm
If all the ecf rated players begin with a converted rating this shouldn't be an issue
How else would you expect it to be done?
If you want a picture of the future, imagine a QR code stamped on a human face — forever.

Roger de Coverly
Posts: 21291
Joined: Tue Apr 15, 2008 2:51 pm

Re: ECF Grading Proposals

Post by Roger de Coverly » Tue Aug 27, 2019 5:57 pm

NickFaulks wrote:
Tue Aug 27, 2019 5:40 pm
How else would you expect it to be done?
It would be theoretically possible to run it the same way as the grade recalculation of a few years ago. Namely start with data a few years back, run the new player estimation method on everyone and roll forward from there. That ranks players, but an arbitrary absolute scale would have to be chosen. Another method would be to have assumed the reliability of some or all of the players existing FIDE ratings either now or in the past.

The outcome of the process is that everyone is going to have at least two Elo ratings. How widely they will vary remains to be seen. As mentioned above there's now a distrust of the ranking order implied by the FIDE Elo ratings for players below perhaps 2000. Whether an ECF Elo rating including more results would work any better remains to be seen and may well hinge on the hack proposed for handling improving juniors.

David Robertson

Re: ECF Grading Proposals

Post by David Robertson » Tue Aug 27, 2019 6:28 pm

Prior to drafting my ECF Delegate report last June to my League AGM, I had been under a little local pressure to explain where we were up to on "this grading business ". I never like to admit that I haven't a clue because it rather tarnishes my brand, colleagues naturally looking to me as a font of clarity, sagacity and good humour in these matters. But I confess to working somewhat in the dark when I had this to say.
David Robertson, ECF Delegate wrote:On Elo-style 4-figure grades, I can’t get too fussed here. The current model is an outlier: everyone else, in the UK and more widely, uses Elo-style. There’s scope for some confusion though, not least for players with a reliable Elo, and for players (like me) with an unreliable/partial Elo. We could end up with three numbers. In my case (for example), these could be:
• 1870 (my actual ECF with an added ‘0’)
• 1930 (my actual Elo, but derived from a historic low base from partial results)
• 2100 (my ‘reliable’ Elo, converted from my ECF grade)
Moreover, since league chess cannot be FIDE-rated, those players with a reliable/trusted Elo who remain active in FIDE-rated events (eg 4NCL) will soon find divergence growing between local and non-local grades. Never mind. We’ll cope
Looks like I wasn't too far wide of the mark

Brian Valentine
Posts: 574
Joined: Fri Apr 03, 2009 1:30 pm

Re: ECF Grading Proposals

Post by Brian Valentine » Tue Aug 27, 2019 6:44 pm

Roger de Coverly wrote:
Tue Aug 27, 2019 5:57 pm
...Whether an ECF Elo rating including more results would work any better remains to be seen and may well hinge on the hack proposed for handling improving juniors.
I'm taken aback by the elegant solution has been demoted to a "hack"!

There are two other levers that give me confidence that the proposal will work better than a copy of FIDE in our circumstances.
1. The requirement to grade as many results as possible means that we get nearly 10% more of results graded , i.e. those results where an rated player plays someone who would not have a grade under the FIDE system.
2. (And this is where the current proposal is being refined.) There are clear benefits in keeping initial ratings as just the TPR of results to date, rather than using the Elo updating formula starting on an often unreliable first submission. As always, this is not a free lunch and I'm looking at the implications based on a 5 year dataset.

Brian Valentine
Manager of ECF Grading

John Swain
Posts: 412
Joined: Sun Jun 20, 2010 10:35 pm
Location: Nottingham

Re: ECF Grading Proposals

Post by John Swain » Sat Aug 31, 2019 1:06 pm

I'm very encouraged by Brian Valentine's remarks and approach on this and on other threads, including the July 2019 grades one. He seems to be open to discussion and not dogmatic or doctrinaire …

I thought that it might be a good idea to compile a list of the advantages and disadvantages of monthly grading. All too often, the issue is seen as a right or wrong issue or modernisers v conservatives, when the truth is that the issue is far more complex. On balance, I prefer the present system, but if the software could be improved without creating extra work for graders and organisers (I am an organiser and not a grader) then I would be a lot happier to move to a monthly system. My list is as follows:

Advantages or alleged advantages of monthly grading:

- monthly grades would be far less out-of-date than six-monthly; this is especially valuable for juniors who are often fast-improving and so would be particularly helpful for determining board orders, allocating folks to the correct tournament for their ability level, etc.
- monthly grades bring us into line with what FIDE does.
- new players, juniors especially, will be encouraged, rather than having to wait six months.
- online results are instant, after each game, so it makes sense for ECF grades to move in this direction (but why not go the whole hog and have the Yorkshire 'Chessnuts' daily experience?)

Disadvantages or alleged disadvantages of monthly grading:

- if the grading software can be improved to the point where matches are reported under an LMS system and results are then "beamed up" to the grading database without further intervention by congress, league and club organisers, local graders etc. that would be ideal, but until this can happen there will be a lot of regular extra work for local graders and organisers to make monthly deadlines, rather than the six-monthly targets at present. No one should be under any pressure or expectation whatsoever to work 24/7 for the ECF, nor have to explain themselves to anyone or apologise for results not reaching the next monthly list.
- most grades will not change over a month by more than one or two, so it is a lot of input for little output.
- organisers will either have to declare one monthly list as key (like the current August list) or risk upsetting people by reserving the right to transfer folks to another section (e.g. a 168 player enters an U170 event, but by the time of the event another list has appeared and she's 170 and been transferred to the Open, a tournament she has no wish to play in and would not have entered if she had been 170 originally).
- it may encourage cheats to throw games to bring their grade down quickly, maybe with a key lucrative tournament in mind, knowing that they don't have to wait six months for their nefarious practices to bear fruit.
- juniors especially, but really all players, should be encouraged to enjoy chess for its own sake and not become obsessed by grades. How many times does a junior ask an opponent "What's your grade?" even before shaking hands? I saw a game fairly recently where a junior graded 124 was thrashing one of my team-mates graded 192 but accepted a draw in a winning position because he had already mentally banked the extra grading points and was afraid he might not see his way through to victory. The match was drawn as a result. If he had been unaware of his opponent's grade or been playing someone graded at his own level, I'm sure he would have played on and won. In the unlikely event that he had lost, he would have learned far more from playing out the ending than he did by the half-finished game. A cop-out draw doesn't just apply to juniors - I have been guilty myself on several occasions!
- some juniors become disillusioned when they "plateau off" in grading terms and lose interest, eventually drifting away from the game. They have grown up in a culture of improving numbers and can't cope with reaching a ceiling. This particularly applies to stronger juniors who are required to reach a particular grade by a certain age for purposes of England selection. One 160ish England player told me that he realised that he wasn't going to get a lot better, certainly not without a huge amount of study to make 180-190 and he preferred to spend his time on academic studies, rugby, socialising etc. Monthly grading will not arrest this phenomenon.

Post Reply