Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Debate directly related to English Chess Federation matters.
User avatar
Adam Raoof
Posts: 2720
Joined: Sat Oct 04, 2008 4:16 pm
Location: NW4 4UY
Contact:

Re: Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Post by Adam Raoof » Wed Aug 25, 2021 1:32 pm

John Upham wrote:
Wed Aug 25, 2021 12:41 pm
NickFaulks wrote:
Wed Aug 25, 2021 10:43 am

Except that many countries have then taken the final step and gone over entirely to the international system. This would be unthinkable in England, where many club players continue to believe that the ECF grading / rating system is the primary benefit for which they pay their membership fees.
How do other NGBs "sell themselves" to their nation's players if they are not offering a national grade / rating?

What are they offering that entices their nations players to maintain their memberships?
Many countries make membership compulsory. The USCF for example, and they dont have a fraction of the opportunities to play in FIDE rated tournaments.
Adam Raoof IA, IO
Chess England Events - https://chessengland.com/
The Chess Circuit - https://chesscircuit.substack.com/
Don’t stop playing chess!

Alex McFarlane
Posts: 1757
Joined: Sat Aug 02, 2008 8:52 pm

Re: Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Post by Alex McFarlane » Wed Aug 25, 2021 1:57 pm

John Upham wrote:
Wed Aug 25, 2021 12:41 pm
What are they offering that entices their nations players to maintain their memberships?
Since FIDE only accepts tournaments for rating from the national body, nations which do not offer their own ratings instead offer access to the FIDE system.

As, currently, FIDE ratings start at 1000, there is at least one country which offers national rating up to that figure for young and inexperienced players. However, once they reach the magic number and get a FIDE rating their national one is 'dropped'.

Other things which can be offered include resolving disputes over the Laws (including suspensions for cheating), training arbiters, training coaches and offering coaching (usually only to juniors).

As Adam indicated, some federations simply insist that membership is a licence to play competitive chess.

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

Re: Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Post by Brian Valentine » Wed Aug 25, 2021 2:53 pm

Hok Yin Stephen Chiu wrote:
Wed Aug 25, 2021 8:49 am
The best feature of the old system was that your grade was calculated from your last 30 games, and (as I understood) it reset, if you haven't played in several years. It allowed players coming back and improving after a break, to have their grade accurately reflected within one grading cycle.
It's not going to be clear for some time what the effects of a year or so hiatus will do to the ratings. SEveral of the reasons have been set out above. All I can say is that we will be watching developments.

At this stage, I want to allay the mis-conception in the quote above. Note that any improvements on the old system would take at least six months. The new approach can be faster or slower than that. It can be as quick as one month, there is already an example in the system.

For those geeky enough to want to check it out look at the rapid rating for Quixiang Han 328538J for July. Here he started at 1063 and had his K-factor restricted to gain only 122 to give 1185 (his performance for the month).

User avatar
Chris Goodall
Posts: 1057
Joined: Sun Oct 10, 2010 6:40 pm
Contact:

Re: Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Post by Chris Goodall » Fri Aug 27, 2021 1:29 pm

Hok Yin Stephen Chiu wrote:
Wed Aug 25, 2021 8:49 am
I don't know how many people feel the same way, but the new ECF ratings, does not make sense, in context of our English chess circuit.

The best feature of the old system was that your grade was calculated from your last 30 games, and (as I understood) it reset, if you haven't played in several years. It allowed players coming back and improving after a break, to have their grade accurately reflected within one grading cycle.

I currently know several 1500 players at university, who in the almost two years of the pandemic have reached 2000 online, and was consistent beating/drawing with 160s/170s in our local online chess league. With issues like the k-factor of 30/20/10, it will be a distant future until these players will ever reach an accurate reflection of their chess playing ability, if (and is true in many cases), on average, they play in one or two rated-rapidplays a year, and 5 to 10 long play games (if you are lucky to fit that many in).

In the old system, the effective range of grades was from 0-280, and wins and losses costed you +/- 50, which is about 20% of the full range. Whereas single digit changes (or a low double digit change) in a four figure system, is about 1% of the full range.

This is not much of a responsive system, and ruins how we can accurately assess the strengths of these university players / young people, in general. And creates a disruptive effect on the intermediate and minor sections of local congresses and rapidplays.
With modern computing it's possible to take a set of players who have played a set of games, and assign a set of ratings to them that exactly reflects every piece of information you have about them. (A "model-free" system, the machine learning buffs call this.) So if Joe 1600, of unknown age, drew a game 7 months ago with Black against a 1450-rated 13-year-old, you can ask the computer what that tells you about Joe 1600's current skill level. Based on everything the computer has learnt about "1600 player of unknown age", "1450 13-year-old", "Black" and "7 months ago" from its training data, it might tell you to adjust Joe 1600's grade downwards by 3.5116 points.

The problem, as I discovered from my adventures with rating systems, is that players think they know themselves better than any computer does. They don't want a number that accurately reflects their expected score against other players. They want a number that looks right, relative to everyone else's numbers. A rating system that actually provides the players with some insight that they didn't already have, will be seen as "random numbers". That's why the whole world isn't currently using the Glicko rating system with its dynamic RD variable, and is instead persisting with the Elo system with its arbitrary k-factor.


Regarding the granularity of ratings, I think even 3-figure Clarke (ECF) grades are more precise than is justified by the formula that calculates them. By a factor of at least 10 and possibly 25. That my grade is 174 one year instead of 171 is statistical noise. If it were 17, it would mostly stay at 17, which would be a true reflection of what my ability tends to do over time. But I'm told juniors get terribly excited about their grade being 36 and their friend's grade being 34, which... fair enough.

3-figure Clarke grades have the nice property that if you played opponents with an average grade of 150 and scored 38%, your grade will be 138; if you scored 92%, it will be 192, etc. One grading point is one percentage point.
Donate to Sabrina's fundraiser at https://gofund.me/aeae42c7 to support victims of sexual abuse in the chess world.

Northumberland webmaster, Jesmond CC something-or-other. Views mine. Definitely below the Goodall Line.

Hok Yin Stephen Chiu
Posts: 148
Joined: Sun Dec 31, 2017 7:52 pm

Re: Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Post by Hok Yin Stephen Chiu » Tue Aug 31, 2021 11:30 am

Chris Goodall wrote:
Fri Aug 27, 2021 1:29 pm
3-figure Clarke grades have the nice property that if you played opponents with an average grade of 150 and scored 38%, your grade will be 138; if you scored 92%, it will be 192, etc. One grading point is one percentage point.
I think you have nailed it, there. This property is something that is broadly very helpful for organisers.

Tim Spanton
Posts: 1204
Joined: Thu May 01, 2008 11:35 am
Contact:

Re: Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Post by Tim Spanton » Wed Sep 01, 2021 8:38 am

Hok Yin Stephen Chiu wrote:
Tue Aug 31, 2021 11:30 am
Chris Goodall wrote:
Fri Aug 27, 2021 1:29 pm
3-figure Clarke grades have the nice property that if you played opponents with an average grade of 150 and scored 38%, your grade will be 138; if you scored 92%, it will be 192, etc. One grading point is one percentage point.
I think you have nailed it, there. This property is something that is broadly very helpful for organisers.
I don't see it - calculating a percentage is not much quicker than using, for example, Fide's rating calculator

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

Re: Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Post by Roger de Coverly » Wed Sep 01, 2021 9:42 am

Players at the Northumbria events which finished on Monday can now see the effect on their ratings. That's quicker than FIDE which hasn't been updated and presumably won't be until 1st October. It would have been dependent on the organisers submitting a rating file before the ECF's cutoff which is evidently later than FIDE's.

I don't think we have yet seen how the new ECF system will handle tournaments that cross a month end.

User avatar
JustinHorton
Posts: 10364
Joined: Mon Aug 04, 2008 10:06 am
Location: Somewhere you're not

Re: Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Post by JustinHorton » Wed Sep 01, 2021 10:32 am

The chief advantage of the old ratings was perhaps the one given to a friend by a fellow-member of his club, which was to the effect that he would give up the game when his rating no longer exceeded his age.
"Do you play chess?"
"Yes, but I prefer a game with a better chance of cheating."

lostontime.blogspot.com

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

Re: Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Post by David Sedgwick » Wed Sep 01, 2021 11:58 pm

JustinHorton wrote:
Wed Sep 01, 2021 10:32 am
The chief advantage of the old ratings was perhaps the one given to a friend by a fellow-member of his club, which was to the effect that he would give up the game when his rating no longer exceeded his age.
Will you give up chess when your FIDE rating no longer exceeds your year of birth?

User avatar
Joey Stewart
Posts: 1857
Joined: Wed Apr 11, 2007 2:35 pm
Location: All Of Them
Contact:

Re: Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Post by Joey Stewart » Thu Sep 02, 2021 1:20 am

With the rate fide ratings are deflating it might not be so long before this joke of being "older then your rating" actually becomes a reality.
Lose one queen and it is a disaster, Lose 1000 queens and it is just a statistic.

User avatar
IM Jack Rudd
Posts: 4815
Joined: Tue Apr 17, 2007 1:13 am
Location: Bideford
Contact:

Re: Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Post by IM Jack Rudd » Thu Sep 02, 2021 1:57 am

Heh, there's a moving target for me: try to keep my FIDE rating higher than my age in weeks.

dejan_lekic
Posts: 29
Joined: Tue Dec 10, 2019 8:23 pm

Re: Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Post by dejan_lekic » Thu Sep 02, 2021 11:49 am

I think the ELO adoption by ECF should have been done many years ago... It is funny to see that even after so long people still come up with all sort of excuses, like one of the recent ones about percentage IF opponents were graded around 150 ECF.... Seriously guys... Out of 11997 members who got grades changed in January 2019 ONLY 780 had grades between 145 and 155 so that "convenience" is mostly useless as only 6.5% of all registered ECF members have average grade of 150. Sure it is possible that in some divisions of some leagues average opponents grade may be 150, but I would say this is extremely rare case.

How do I know? Because I imported all CSV files that were available on the grading website and now I can do simple "select count(*) as total from ecf_full where yyyy = 2019 and mm = 1 and s_grade > 145 and s_grade < 155;" SQL query to find this information...

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

Re: Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Post by Brian Valentine » Thu Sep 02, 2021 12:05 pm

dejan_lekic wrote:
Thu Sep 02, 2021 11:49 am
... Because I imported all CSV files that were available on the grading website ...
Dejan,

The downloads are still available on the new website. Go to the help tab and select section 10, and look in "all rated players" for how to do it.

Not picking on you because it happens with other "customers", but as a general point for those who yearn for the historical grading data: all the historical data is on the new site.

I think it's more accessible, but do recognise that it's all a bit strange to start with.

dejan_lekic
Posts: 29
Joined: Tue Dec 10, 2019 8:23 pm

Re: Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Post by dejan_lekic » Thu Sep 02, 2021 12:11 pm

Brian, it was badly worded statement... I apologise. I should have written something like "... were available on the old website..." :) For the record, I just love the new website! :)

User avatar
Stephen Westmoreland
Posts: 447
Joined: Tue Jul 06, 2021 4:55 pm
Location: Holmfirth

Re: Have the ECF replaced the old grades with the new ratings, at the wrong moment?

Post by Stephen Westmoreland » Thu Sep 02, 2021 6:50 pm

May as well just go for it. Never a good time to do it but it is needed, especially in this online age
HDCA President

Post Reply