ECF Grading Proposals

General discussions about grading.
User avatar
Chris Goodall
Posts: 471
Joined: Sun Oct 10, 2010 6:40 pm
Contact:

Re: ECF Grading Proposals

Post by Chris Goodall » Thu Sep 19, 2019 12:00 pm

John Swain wrote:
Wed Sep 18, 2019 4:16 pm
Has someone turned off the internet in Devon?

That's... not actually too far from the truth. Results used to be posted on the WECU/DCCA site. Unfortunately, this was Bill Frost's personal project, and when he died, there was no way for anyone to carry on updating it.


Um... admins? I didn't write the text above. I think John Swain hit the Quote button on this post, and somehow instead of quoting it, has managed to edit it :?
Last edited by Chris Goodall on Thu Sep 19, 2019 3:16 pm, edited 1 time in total.
Chris Goodall, formerly known as Chris Wardle. Northumbria League hand-cranker; ECF Grader for Bernicia and the NCCU.
Newcastle is not in Scotland!

User avatar
IM Jack Rudd
Posts: 4079
Joined: Tue Apr 17, 2007 1:13 am
Location: Bideford

Re: ECF Grading Proposals

Post by IM Jack Rudd » Thu Sep 19, 2019 3:00 pm

Ack, sorry, that was me. I hit the Edit button on your post instead of the Quote one.

User avatar
Chris Goodall
Posts: 471
Joined: Sun Oct 10, 2010 6:40 pm
Contact:

Re: ECF Grading Proposals

Post by Chris Goodall » Thu Sep 19, 2019 3:12 pm

Repost:
John Swain wrote:
Wed Sep 18, 2019 4:16 pm
The move to monthly and four-digit grading seems to be proceeding; I wish Brian Valentine and others the best of luck in making both work with minimal inconvenience to graders and organisers.
I will press ahead with WikiGrades. I've now checked the 155 "Leagues etc" from the grading list, not counting club internal competitions. 42 are on LMS, though I'm not sure whether Suffolk league is being updated.

Of the remaining 113, I think I can write Scrapy modules to pull the results from all but 23 of them.

Of those 23: Bristol, Cumbria indivs and Herts league are visible and can be entered manually if necessary. Fenland, Richmond juniors and Wiltshire don't have the latest results visible and Pimlico Summer and Northants KO don't publish individual results - nothing I can do about those. EACU Clubs and North Circular are question marks since they don't currently have any matches visible. I'm not sure if Leicestershire Individual is the same thing as Leicestershire County Champs, or Norfolk Rapidplay the same thing as North Norfolk Rapidplay.

I cannot find, possibly because they no longer exist: Derby Individual, Devon Individual, Hammersmith League, Exeter and District, Hertford Chess Challenge, UKCC Mall School Quals, WECU Counties. Has someone turned the internet off in Devon? (Yes - see Jack Rudd's quote of my original post.)

If I can get some of those modules working, next step is to contact the graders who are concerned about monthly workload, and offer to take the Leagues etc. off their hands. They would still need to submit tournament reports as they happen.
Chris Goodall, formerly known as Chris Wardle. Northumbria League hand-cranker; ECF Grader for Bernicia and the NCCU.
Newcastle is not in Scotland!

Mick Norris
Posts: 8209
Joined: Tue Apr 17, 2007 10:12 am
Location: Bolton, Greater Manchester

Re: ECF Grading Proposals

Post by Mick Norris » Thu Sep 19, 2019 3:17 pm

There would be GDPR (data protection) issues here; my club members agree to their data being shared with the leagues they enter, and the ECF, but not with anyone else
Any postings on here represent my personal views and should not be taken as representative of the Manchester Chess Federation www.manchesterchess.co.uk

User avatar
Chris Goodall
Posts: 471
Joined: Sun Oct 10, 2010 6:40 pm
Contact:

Re: ECF Grading Proposals

Post by Chris Goodall » Thu Sep 19, 2019 3:20 pm

Mick Norris wrote:
Thu Sep 19, 2019 3:17 pm
There would be GDPR (data protection) issues here; my club members agree to their data being shared with the leagues they enter, and the ECF, but not with anyone else
They don't need to share any data. Results of chess games are a matter of public record.
Chris Goodall, formerly known as Chris Wardle. Northumbria League hand-cranker; ECF Grader for Bernicia and the NCCU.
Newcastle is not in Scotland!

Brian Valentine
Posts: 476
Joined: Fri Apr 03, 2009 1:30 pm

Re: ECF Grading Proposals

Post by Brian Valentine » Thu Sep 19, 2019 4:08 pm

Chris,

I have been reluctant to join this debate while most of my current activity is around administration. However when you are thinking about manual entry to try out your system, I think I ought to add my own thoughts. I'd hate you to be too far into your adventure without considering the drawbacks.

I cannot comment on the best way of holding records as I have to defer to experts. You seem to have found a nifty way to keep records and then found a grading method to fit it.

It is not that the ECF hasn't considered a (30 result) moving average model, it is that an Elo system (amongst several) is statistically superior (especially for recreational chess). Also you seem to miss two key requirements given to us by the Board:
1. Players should receive a grade (of some reliability)as soon as possible and
2. Results submitted up to 7 months in arrears will be graded

I have no wish to impede your project, but make clear some obstacles. You may want to have a chat with Jon Griffith at chessnuts about the workload in maintaining a separate rating site.

Enjoy your life!

User avatar
Chris Goodall
Posts: 471
Joined: Sun Oct 10, 2010 6:40 pm
Contact:

Re: ECF Grading Proposals

Post by Chris Goodall » Thu Sep 19, 2019 6:02 pm

Brian Valentine wrote:
Thu Sep 19, 2019 4:08 pm
It is not that the ECF hasn't considered a (30 result) moving average model, it is that an Elo system (amongst several) is statistically superior (especially for recreational chess). Also you seem to miss two key requirements given to us by the Board:
1. Players should receive a grade (of some reliability)as soon as possible and
2. Results submitted up to 7 months in arrears will be graded
Both Elo and Clarke feed your grade and your opponent's grade into a formula, and the formula tells you how many points to take off the opponent and give to you. You can have that formula model expected results as a line, a logistic curve or anything in between - the important part is that you aren't making use of any information that isn't already encapsulated in the grades. If you wanted to wring every last drop of information out of the two grades, you could create a 3-dimensional actuarial table: my grade vs. my opponent's grade vs. result, and individually calculate every cell in that table. Throw White/Black in for good measure.

Rating reliability is a whole science of its own. One always runs into the mathematical certainty that not every game gives you the same amount of information. If an unknown player, let's say newly arrived from Russia, started playing in your local league, you would pay more attention if he beat one FIDE Master than if he lost to 10 FIDE Masters, because most people lose to FIDE Masters. So if you have [number of games played] and [information content] on two axes of a graph, the line on that graph that represents the boundary of "reliability" will be a diagonal line. That means you might have to have a situation where two players play 5 games each, one gets given a grade, the other doesn't. If you have a constraint that for no players A and B where A has played more games than B, A does not have a grade but B does, then the problem is trivial, it's just a matter of counting to 5, or 10, or whatever.
Chris Goodall, formerly known as Chris Wardle. Northumbria League hand-cranker; ECF Grader for Bernicia and the NCCU.
Newcastle is not in Scotland!

Dragoljub Sudar
Posts: 127
Joined: Wed Jun 16, 2010 10:34 pm

Re: ECF Grading Proposals

Post by Dragoljub Sudar » Tue Oct 22, 2019 6:23 pm

One of the major advantages of the current ECF grading system is that it ignores results more than 3 years old so reflects current or recent form.

The flawed FIDE system retains inactive ratings that can take many years (if ever) to bring up to a level that reflect's someone's standard of play.

A club mate played FIDE rated chess a decade ago. He is now at least 160 (1900) standard but has an inactive FIDE rating in the 1400s.

I would appreciate it if someone could answer the following:-

1. Why doesn't FIDE delete inactive ratings that are more than 'n' years old?

2. How many 5 round FIDE rated events would my club mate have to play with TPRs of 2000 to get his FIDE rating up from 1450 to 1900?

3. Will the new ECF system retain a 3 year cut-off or will it adopt the flawed FIDE model?

4. Will a player be able to ask the ECF to delete a historic grade more than 'n' years old so that he can start afresh?

NickFaulks
Posts: 5887
Joined: Sat Jan 02, 2010 1:28 pm

Re: ECF Grading Proposals

Post by NickFaulks » Tue Oct 22, 2019 7:03 pm

Dragoljub Sudar wrote:
Tue Oct 22, 2019 6:23 pm
2. How many 5 round FIDE rated events would my club mate have to play with TPRs of 2000 to get his FIDE rating up from 1450 to 1900?
It doesn't work quite like that, but if his earlier rating was not based on too many games, then three should get him to 1800, two more after that.

In England, where most results are not submitted to FIDE for rating, this might take a while. You have to remember that the rating system is designed to suit federations which do generally submit results, not those that do not.

edit : Of course, if he played elsewhere in Europe, the games which got his grade to 160 would probably have been FIDE rated anyway, so he would not face this problem.
If you want a picture of the future, imagine a boot stamping on a human face — forever.

Brian Valentine
Posts: 476
Joined: Fri Apr 03, 2009 1:30 pm

Re: ECF Grading Proposals

Post by Brian Valentine » Tue Oct 22, 2019 9:22 pm

I can only comment on the last two points.
3. I don't think one anecdote makes a rating system flawed. Nevertheless we are looking at whether there should be some cut off. GDPR is one issue.
4. This is not something we envisage. Organisers want the evidence for section limits.

Brian Valentine
Manager of ECF Grading

User avatar
Chris Goodall
Posts: 471
Joined: Sun Oct 10, 2010 6:40 pm
Contact:

Re: ECF Grading Proposals

Post by Chris Goodall » Tue Oct 22, 2019 10:23 pm

Dragoljub Sudar wrote:
Tue Oct 22, 2019 6:23 pm
4. Will a player be able to ask the ECF to delete a historic grade more than 'n' years old so that he can start afresh?
Brian Valentine wrote:
Tue Oct 22, 2019 9:22 pm
4. This is not something we envisage. Organisers want the evidence for section limits.
That's a good reason to delete historic grades of players who were last active before 2009, or at least convert them to new money. If organisers take pre-2009 grades at face value then someone who fell off the grading list in 2008 with a grade of 129 can enter an Under 130 section today, whereas someone who stayed active with a grade of 110 probably can't because their grade is probably 135 now.
Chris Goodall, formerly known as Chris Wardle. Northumbria League hand-cranker; ECF Grader for Bernicia and the NCCU.
Newcastle is not in Scotland!

User avatar
Matt Mackenzie
Posts: 3375
Joined: Tue Mar 31, 2009 11:51 pm
Location: Millom, Cumbria

Re: ECF Grading Proposals

Post by Matt Mackenzie » Tue Oct 22, 2019 10:33 pm

On this occasion, Chris makes a very valid point.
"Set up your attacks so that when the fire is out, it isn't out!" (H N Pillsbury)

NickFaulks
Posts: 5887
Joined: Sat Jan 02, 2010 1:28 pm

Re: ECF Grading Proposals

Post by NickFaulks » Tue Oct 22, 2019 11:42 pm

Matt Mackenzie wrote:
Tue Oct 22, 2019 10:33 pm
On this occasion, Chris makes a very valid point.
Really? Is it actually true that 110 in 2008 is equivalent to 135 in 2019?
If you want a picture of the future, imagine a boot stamping on a human face — forever.

Roger de Coverly
Posts: 19000
Joined: Tue Apr 15, 2008 2:51 pm

Re: ECF Grading Proposals

Post by Roger de Coverly » Tue Oct 22, 2019 11:47 pm

NickFaulks wrote:
Tue Oct 22, 2019 11:42 pm
Is it actually true that 110 in 2008 is equivalent to 135 in 2019?
The ECF, or perhaps just its directors, were talked into believing that the grading system needed recalibration. AS a consequence, the median and mode grades moved from the 110 to 120 zone up to the 130-140 zone.

NickFaulks
Posts: 5887
Joined: Sat Jan 02, 2010 1:28 pm

Re: ECF Grading Proposals

Post by NickFaulks » Tue Oct 22, 2019 11:55 pm

Roger de Coverly wrote:
Tue Oct 22, 2019 11:47 pm
The ECF, or perhaps just its directors, were talked into believing that the grading system needed recalibration. As a consequence, the median and mode grades moved from the 110 to 120 zone up to the 130-140 zone.
I was not in England and missed all this excitement. Is there documentation of how the desired result was achieved, and why?
If you want a picture of the future, imagine a boot stamping on a human face — forever.

Post Reply