ECF to scrap its grading system

Debate directly related to English Chess Federation matters.
User avatar
Michael Farthing
Posts: 1828
Joined: Fri Apr 04, 2014 1:28 pm
Location: Morecambe, Europe

Re: ECF to scrap its grading system

Post by Michael Farthing » Sun Dec 09, 2018 12:57 am

Kevin Thurlow wrote:
Sat Dec 08, 2018 6:36 pm
I can see the point of juniors having frequent recalculations of grade, but is it wise to change the system just to suit players who are generally only active for 5 - 10 years? Of course, new arrivals to the country would benefit from having frequent recalculations of grade.
Most, but not all, and retaining 10% of 3000 is better than retaining 10% of 2000. (made up figures of course, but surely attracting juniors must be a major objective?)

J T Melsom
Posts: 571
Joined: Wed Aug 12, 2009 11:12 pm

Re: ECF to scrap its grading system

Post by J T Melsom » Sun Dec 09, 2018 1:56 am

It is clear to me that these changes are proposed entirely for the benefit of new players. if they play enough games against relevant opponents within a shorter period than six months then let them have a number to amuse them. But just as F grades are generally out-with most competition rules I suspect these grades will be as well, and chess requires patience so this rush to get a rating should be properly moderated. I understand the demand to get a number but I don't think the system should be changed just for that unless it retains a level of robustness.

Brian Valentine
Posts: 439
Joined: Fri Apr 03, 2009 1:30 pm

Re: ECF to scrap its grading system

Post by Brian Valentine » Sun Dec 09, 2018 7:50 am

J T Melsom wrote:
Sun Dec 09, 2018 1:56 am
It is clear to me that these changes are proposed entirely for the benefit of new players. if they play enough games against relevant opponents within a shorter period than six months then let them have a number to amuse them. But just as F grades are generally out-with most competition rules I suspect these grades will be as well, and chess requires patience so this rush to get a rating should be properly moderated. I understand the demand to get a number but I don't think the system should be changed just for that unless it retains a level of robustness.
While I have some ideas on what constitutes a good system, I'm open to anyone adding to the list of statistical features that constitute "robustness".

Kevin Thurlow
Posts: 3048
Joined: Wed Apr 30, 2008 12:28 pm

Re: ECF to scrap its grading system

Post by Kevin Thurlow » Sun Dec 09, 2018 9:01 am

I should have added that twice a year, I have sent internal club results to our tame grader (an ex-member of the club), and he has kindly processed the results. I don't know about him, but I won't be keen about doing that 9 times a year (we don't meet June - August), instead of twice.

If the idea is just to give new players a grade (any grade) quickly, then perhaps the idea is useful.

Roger de Coverly
Posts: 18188
Joined: Tue Apr 15, 2008 2:51 pm

Re: ECF to scrap its grading system

Post by Roger de Coverly » Sun Dec 09, 2018 10:50 am

Kevin Thurlow wrote:
Sun Dec 09, 2018 9:01 am
but I won't be keen about doing that 9 times a year (we don't meet June - August), instead of twice.
A replication of the international Elo system can cope with late results. It doesn't have any problems with the notion that two players meeting more than once in the same reporting period can have two different calculations applied.

Where a simple replication of the international system would struggle is with rapidly improving players and the stability of the ratings for juniors.

I note that according to the ECF grades I'm marginally better than one well publicised junior player. According to the current FIDE ratings, I'm marginally better on those by the same amount. This hides that on previous lists, a month or two before, he was 150-200 points ahead.

MartinCarpenter
Posts: 2454
Joined: Tue May 24, 2011 10:58 am

Re: ECF to scrap its grading system

Post by MartinCarpenter » Sun Dec 09, 2018 12:03 pm

Kevin Thurlow wrote:
Sat Dec 08, 2018 6:36 pm
I don't really care one way or the other. I do wonder who would use monthly grades.
More or less no one in Yorkshire when we had continuous grades. They were either profoundly unstable - new people! - or fairly obviously noise (everyone else). All the leagues etc just used the grades achieved at the end of August(?) for board orders & stuff.

The claimed use in Yorkshire was never accuracy. It was much more about motivating people to play more/get enjoyment etc. Can't see why it'd work myself, but it is possible :)

David Williams
Posts: 203
Joined: Sun Mar 29, 2009 8:37 pm

Re: ECF to scrap its grading system

Post by David Williams » Sun Dec 09, 2018 1:41 pm

As I understand the current system your grade is the average grade of all the people you have played, plus or minus one point for each percentage point your record is above or below 50%. So if you have, say two wins against a 160, a draw with a 150, and a win and a loss against a 140, you have scored 70% against opposition with an average grade of 150. Your grade is 170. Also calculated as (210+210+150+190+90)/5. The arithmetic holds good for any number of games.

The only difference if you do it strictly on the last 30 games instead of in batches is that it becomes a little more volatile. If you beat someone graded 10 points higher than you, you might expect your grade to go up by 2. But if the game that is dropping out of the calculation was a win against someone graded 40 points higher than you, it will actually go down by 1. But that's as extreme as it gets.

I just don't see where statistical soundness comes into it.

Roger de Coverly
Posts: 18188
Joined: Tue Apr 15, 2008 2:51 pm

Re: ECF to scrap its grading system

Post by Roger de Coverly » Sun Dec 09, 2018 9:31 pm

David Williams wrote:
Sun Dec 09, 2018 1:41 pm
The only difference if you do it strictly on the last 30 games instead of in batches is that it becomes a little more volatile.
It hasn't been done that way for around 4 to 5 years. What happens now is that a year's worth of results is always used. If that doesn't total at least 30, earlier scores are added in by proportion. Thus if in the most recent year you play 25 games, 5 games at the average of the previous season's performance are added back.

The statistical underpinning is that it's a measure of performance, used than to give a calculation value to assess the performance of your opponents in the next time period.

Neill Cooper
Posts: 1240
Joined: Sun Feb 24, 2008 4:43 pm
Location: Croydon
Contact:

Re: ECF to scrap its grading system

Post by Neill Cooper » Mon Dec 10, 2018 7:20 am

My year 7 pupils find it frustrating that having taken up graded chess in September they do not get a grade until January. Those with grades would love to get monthly grading updates. In view of on line ratings for computer games (including chess) being updated after every game they are not used to the long wait.
Presumably as with the 6 monthly updates (which many objected to at the time) the system of monthly updates would just be for rapidplay. This would include mainly junior and other one day events but few leagues.

Edit: I should have said "the system of monthly updates would initially be trialled just for rapidplay."
Last edited by Neill Cooper on Mon Dec 10, 2018 7:51 am, edited 1 time in total.

Mick Norris
Posts: 7595
Joined: Tue Apr 17, 2007 10:12 am
Location: Bolton, Greater Manchester
Contact:

Re: ECF to scrap its grading system

Post by Mick Norris » Mon Dec 10, 2018 7:36 am

The 6 monthly updates apply to longplay, not just rapidplay
Any postings on here represent my personal views and should not be taken as representative of the Manchester Chess Federation www.manchesterchess.co.uk

Neill Cooper
Posts: 1240
Joined: Sun Feb 24, 2008 4:43 pm
Location: Croydon
Contact:

Re: ECF to scrap its grading system

Post by Neill Cooper » Mon Dec 10, 2018 7:48 am

Mick Norris wrote:
Mon Dec 10, 2018 7:36 am
The 6 monthly updates apply to longplay, not just rapidplay
Sorry, I was unclear in my post (now edited)- I meant it should be trialled in rapidplay.

There were 5 years of January rapidplay grades before January long play was introduced.
In 2007 to 2011 the January grading list was only rapidplay.

Richard Bates
Posts: 2933
Joined: Fri Nov 14, 2008 8:27 pm

Re: ECF to scrap its grading system

Post by Richard Bates » Mon Dec 10, 2018 8:06 am

My instinct (happy to be contradicted with statistical evidence!) is that the current ECF method of calculation arguably lends itself better to a system of monthly updates “for fun”, but an anchored “official” grade that can be used reliably for things like year long competitions. The point being that for the vast majority of players that the system needs to cater for in its primary purpose (ranking people in order of strength), it is possibly less volatile and less affected (except at very low levels) by the volume of games played. ELO systems (when recalculated regularly as they are now - unlike how FIDE ratings originally operated) may drag in a greater element of very recent form (which it compensates for over time by being self-correcting) which undermine the case for treating a particular arbitrary month’s grade as “official”.

In general you can start producing ECF calculated grades on a monthly basis, but doing do will not have a material effect on the eg. August grade that might be treated as the official one.

Richard Bates
Posts: 2933
Joined: Fri Nov 14, 2008 8:27 pm

Re: ECF to scrap its grading system

Post by Richard Bates » Mon Dec 10, 2018 8:14 am

Neill Cooper wrote:
Mon Dec 10, 2018 7:48 am
Mick Norris wrote:
Mon Dec 10, 2018 7:36 am
The 6 monthly updates apply to longplay, not just rapidplay
Sorry, I was unclear in my post (now edited)- I meant it should be trialled in rapidplay.

There were 5 years of January rapidplay grades before January long play was introduced.
In 2007 to 2011 the January grading list was only rapidplay.
Although it could be pointed out that what happened was they were trialled in rapidplay, introduced on the same basis in slowplay, and then within a very short order the method of calculation was changed (I think because there was a perception that the system had become too volatile and unstable). Which may indicate that for many people rapidplay grades have always been treated as a bit of fun , and/or otherwise are not a great basis for trialling changes - parallel running is possibly better. Although, as per above, I think the risks are far less if the core method of calculation remains unchanged (but that doesn’t appear to be what is being proposed).

Alison Bexfield
Posts: 23
Joined: Mon Nov 02, 2015 12:30 pm

Re: ECF to scrap its grading system

Post by Alison Bexfield » Mon Dec 10, 2018 9:32 am

It is disappointing reading the replies to date to see how stuck in the past the ECF is. Thirty years ago I played a lot of competitive chess in leagues and in weekend tournaments. I found the grading system frustrating then due to its annual nature but assumed that was how it was.

Since that time I took up the game of Go and found it did not have to be that way. The European go grading database is updated whenever any tournament results or league results are submitted. Generally UK tournaments are submitted electronically a day or so after a tournament so that within a few days of the event players can see their updated grades. Players enter tournaments at their most recent grade. If they enter several months in advance and their grade then changes, their entry grade is updated at registration. If someone has improved rapidly without playing a tournament game they can adjust by entering at a new grade at the tournament organiser's discretion and effectively they are reset. This is done for example with new players who are quite experienced and entering their first tournament - they are given an estimated grade and the system then adjusts once results are submitted. And for league teams a set cut off date is used to select teams. Of course Go also uses McMahon for most tournaments so everyone plays in the same event with around 20 sections that players move fluidly up and down in according to their cumulative score (so that players winning can meet progressively harder opponents and players losing can meet weaker opponents). I find it sad that chess is still so wedded to the swiss system which is not suitable for juniors entering their first events. (But that's a different topic).

I am now back playing competitive chess again and running a junior chess club. I am a great fan of FIDE grades as they are updated monthly. ECF grades are just not suitable for juniors and I still find the system annoying and outdated. I enter tournaments and want some immediate feedback. If you play a lot of rated games in a year there is less volatility but of you don't play many then there will be volatility but this is not a bad thing. I understand that I have up days and down days with my play. I don't mind my grade reflecting that.

I sincerely hope the ECF does manage to bring in monthly grade updates. If they cannot do this, I would vote to abolish ECF grades and move entirely to using FIDE grades.

Alison

J T Melsom
Posts: 571
Joined: Wed Aug 12, 2009 11:12 pm

Re: ECF to scrap its grading system

Post by J T Melsom » Mon Dec 10, 2018 9:57 am

I think the vast majority of posters just want a system that is robust and doesn't impose substantial additional burden on volunteer graders. And it does have to be recognised that whilst there is a group of people predominantly but not exclusively juniors who are keen to get numerical feedback on performance as soon as possible, many more are either disinterested in the grade , or relatively unconcerned about whether it is monthly, six monthly or annual. Even those who obsessively track grading performance during the year don't necessarily want monthly data to unduly inform competition eligibility or board order. Some of us also recognise that it is only a statistical measure, often misunderstood, and perhaps whilst the number is important it should be set in its proper context - its a terribly blunt form of feed-back.


So if the ECF sees this as a priority and can satisfy the tests in my first sentence then I'd be relaxed. But as to the intrinsic merit of the change, I'm less convinced, that this is a demand that it is desirable to meet even if there is the competence to do so.

Post Reply