ELO vs ECF

[John Upham]

Oh dear! a raw nerve has been touched!

"ECF Council decided that the ECF didn't want standard play lists more than once a year."

Maybe the membership does? Who is the customer? Who cares?

Let's continue to stick our heads in the sand.

John

[Chris Majer]

Oh dear! a raw nerve has been touched!

Yes, I confess It. First my proposals were turned down by Council. Then I have people telling me that I should change something that I wasn’t allowed to.

ECF Council decided that the ECF didn't want standard play lists more than once a year."
Maybe the membership does? Who is the customer? Who cares?

I don’t know who exactly who means by the Membership (the Direct Members of the ECF or chess players as a whole). However, for the benefit of John (and I’m sure Sir Alan would agree) your Customer is whoever is paying you money. Today, that's primarily the leagues, congresses, counties and unions, of which the leagues are the largest payer. I don't know how John's local league operates but mine makes decisions at meetings to which every club is entitled to send a representative. So John, are you saying that the leagues are out of touch with what the chess players in clubs want? Perhaps there are two different groups of chess player. The first (and largest group) plays only in the local league and so doesn’t want a grade more than once a year. The second (and more vocal) plays 4NCL and congresses and so wants grades several times a year.

[Greg Breed]

The first (and largest group) plays only in the local league and so doesn’t want a grade more than once a year. The second (and more vocal) plays 4NCL and congresses and so wants grades several times a year.

I can concur there. In my local league I would say nearly half of the players' only rated games come from the league and therefore only get about 10 per season. They play for 7 months (October to April) and average a little over 1 game per month. Bear in mind these are rated games. The rest usually play in congresses and also other leagues and probably average over 30 rated games per season. Can the ELO system (or whichever is supposed to be the best one) provide accurate grades based on 3-4 games? At least the ECF formula uses results from previous seasons (if under 30 games) to increase the 'correctness' of the grade.

Again, I'm not saying that the ECF way is the best way or the even the the FIDE way is best, but that if we were to change the grading formula, etc. then a substantial revamping of the way in which the grading system works would be necessary. By this I mean what happens to get the results to the official grader.

[John Upham]

Via the League Management web application (that I am currently developing) I hope to demonstrate that it is technically feasible to drip feed results from ganes / matches / tournaments / congresses /whatever as soon as they are confirmed by both captains as being correct. After all, the rating officer for that League will probably believe them also in normal, once a year, circumstances.

I'm hoping to develop an XML standard for results presentation although upon discussion the usual response is "What is XML?"

Ignoring these little barriers to communication, I hope to demonstrate that it is possible at least.

I may need an API into the ECF database but where there is a will there may be a way. Maybe Blue Apricot could / would assist?

Those who know what I'm babbling on about may be able to put 2 & 2 together and make the leap forward and those don't won't be any the wiser!

We can only try in the face of stern resistance : when is the AGM?

John

[Carl Hibbard]

Via the League Management web application (that I am currently developing) I hope to demonstrate that it is technically feasible to drip feed results from ganes / matches / tournaments / congresses /whatever as soon as they are confirmed by both captains as being correct. After all, the rating officer for that League will probably believe them also in normal, once a year, circumstances.

I'm hoping to develop an XML standard for results presentation although upon discussion the usual response is "What is XML?"

Ignoring these little barriers to communication, I hope to demonstrate that it is possible at least.

I may need an API into the ECF database but where there is a will there may be a way. Maybe Blue Apricot could / would assist?

Those who know what I'm babbling on about may be able to put 2 & 2 together and make the leap forward and those don't won't be any the wiser!

We can only try in the face of stern resistance : when is the AGM?

John

I might be able to assist at some point

Is this a FREE piece of web management software then John?

[John Upham]

FREE is a dirty word!

I've spent at least 100 hrs of development time so far and its been running in anger for two years for two different leagues.

I'm doing a major rewrite of the underlying code to make it marketable.

Not sure if to obfuscate the code yet : hopefully its been written such that won't be needed!

John

[Roger de Coverly]

Via the League Management web application (that I am currently developing) I hope to demonstrate that it is technically feasible to drip feed results from ganes / matches / tournaments / congresses /whatever as soon as they are confirmed by both captains as being correct.

As the league controller and a match captain in one of the leagues where this software has been installed, I can confirm that it works as John says. The key point to getting speedy and accurate result reporting and grading is to use a library of names and codes that is the same as used in the ECF grading database. Many Congresses seem able to post their crosstables on the web within hours of the event completion. So I don't think a quarterly system fails because you can't get the results.

The practical problem is there's not one ELO system but a multitude. For example the international system (FIDE) is not the same as the ICC system. All work on the basis of adjusting an initial rating by something which is a function of your result - you stay the same if you draw with someone of equal rating and go up if you beat them and down if you lose. Also if you play someone much better you are assigned an expected score - perhaps - 0.25. So if you win or draw you gain points. Whole papers have been written about how to calculate the initial rating, how to assign the expected score and how to compute the gain or loss of points.

A national ELO system doesn't even have to use 4 digit numbers.

Every player that played international chess would still have two ratings for the simple reason that different games would be included and different calculation methods used.

Not withstanding the above I would be mildly in favour of a quarterly domestic ELO provided the parameters were sensible. I wasn't in favour of the 6 monthly ECF because as someone who plays more than 30 games in a calendar year I know that the annual grade measures my performance 1st June to 31st May. I didn't see the point of publishing another list to measure the performance 1st December to 30th November overlapping the earlier grade.

[Greg Breed]

Via the League Management web application (that I am currently developing) I hope to demonstrate that it is technically feasible to drip feed results from ganes / matches / tournaments / congresses /whatever as soon as they are confirmed by both captains as being correct. After all, the rating officer for that League will probably believe them also in normal, once a year, circumstances.

As the league controller and a match captain in one of the leagues where this software has been installed, I can confirm that it works as John says. The key point to getting speedy and accurate result reporting and grading is to use a library of names and codes that is the same as used in the ECF grading database. Many Congresses seem able to post their crosstables on the web within hours of the event completion. So I don't think a quarterly system fails because you can't get the results.

That would be great if it could be done and if it does work then the ECF should stump up whatever is necessary to bring chess into the new millenium. People tend to want their results, league standings and grades yesterday!

A national ELO system doesn't even have to use 4 digit numbers.

Although it would help if it matched the FIDE system otherwise what's the point?

Every player that played international chess would still have two ratings for the simple reason that different games would be included and different calculation methods used.

Surely the point is to amalgamate as much as possible to reduce differences between one country and another and the international system!

Not withstanding the above I would be mildly in favour of a quarterly domestic ELO provided the parameters were sensible.

Ditto, so long as an efficient system was put in place (See John Upham's post).

I wasn't in favour of the 6 monthly ECF because as someone who plays more than 30 games in a calendar year I know that the annual grade measures my performance 1st June to 31st May. I didn't see the point of publishing another list to measure the performance 1st December to 30th November overlapping the earlier grade.

The six monthly list is used for rapidplay only and it does have it's uses as people can do a lot more rapidplay games than standard.

[Roger de Coverly]

Although it would help if it matched the FIDE system otherwise what's the point?

You will have a national system which includes games which are not eligible to be rated by the international system - therefore the answers are different. If you have a different numeric range it's very clear that they aren't the same thing. That said, I believe every national system uses the same 0-3000 range as the FIDE one. FIDE did once publish a 3 digit rapid play list. You had entries like Adams Michael 287, Speelman Jonathan 263, Short Nigel 262 which made it look very British!

The six monthly list is used for rapidplay only

Last year's defeated proposal was to publish a 6 monthly standard play list.

[Greg Breed]

So your point is (and I imagine everyone elses is too) that even though the grades will look similar, they aren't the same as the FIDE grades as they are not internationally based. It is merely for comparison purposes. I suppose it would be easier to then compare yourself to the rest of the world. It would make life easier for tournament controllers who have Brits (English) entering foreign events.

I suppose the final question is: What are we going to do about it? We can't even decide if we want to discuss it let alone how we go about it!

I do like John Upham's idea of a universally accepted administrative system for submitting results.

[Roger de Coverly]

I suppose the final question is: What are we going to do about it?

The minimum standards for inclusion in the international system are higher than the ECF.

From memory

minimum rating 1400 (or is it 1200?) - equates to around 100 ECF.
minimum playing session 4 hours
maximum games per day 2
you have to play 9 existing rated players to get an initial rating.

On the assumption that you aren't going to lobby FIDE to weaken these and you aren't going to close down evening leagues and 3 rounds on Saturday weekend tournaments, then you are going to have a domestic rating system running alongside the international one.

In France where they don't do rated evening leagues and weekend tournaments, I believe they are intending to phase out their domestic system.

It would make life easier for tournament controllers who have Brits (English) entering foreign events.

If you have an international rating then it's used and the ECF rating ignored. Otherwise you just tell them the traditional conversion formula

175 = 2000
150 = 1800
125 = 1600
100 = 1400

which is usually accepted. If they want to validate it, point them at the ECF online grading page. At the sub 2000 level many european players still only have a domestic ELO rating in any event.

It's only a relative handful of adults that play outside Britain anyway.

[Sean Hewitt]

Yes, you're fired because you've ignnored what your Customer has already told you! A year ago, ECF Council decided that the ECF didn't want standard play lists more than once a year.

Except (of course) that council is not the customer - the players are. The old foggies in council may not want more regular grading lists (they probably want a return to petrol sold in gallons, and paid for in pounds shillings and pence) but the vast majority of players most certainly do want more frequent publication of grades.

And the players pay the game fee - via either congress entry fees or league entry fees paid for from their club subscriptions!

[Chris Majer]

Sean wrote:

the vast majority of players most certainly do want more frequent publication of grades.

Do you have any recent tangible evidence for this assertion? If you do I would be very interested because it would provide a justification for reopening the debate on more frequent grades.

Sean also wrote:

Except (of course) that council is not the customer - the players are.

With the current ECF structure I don't think that this is realistic. The events are intermediaries for the players and are supposed to provide a challenge of communication between the ECF and the players. It can be asserted that this communication doesn't work well and that the ECF really needs to be a membership organisation of the players as opposed to a federation of chess organisations. But that is of course the very debate that the ECF Board has kicked off.

[Greg Breed]

Assuming that the grading system gets improved and the formula tidied up, couldn't we switch to using the four digit system without much trouble?
Starting with 600 ECF-ELO (which equates to zero ECF)
200 ELO = 25 ECF
The 40-point rule becomes the 320-point rule.

It could still work couldn't it? without using this K-factor business. It may not be as accurate (another discussion) but it will at least look similar.

[Edward Tandi]

I quite like the idea of moving to ELO. One system for the world.

I reckon we should have a national vote. Post a poll on the ECF web site with the following options:

1) Leave the current rating system alone (point out problems)
2) Modify the ECF rating system to cope with deflation (list details)
3) Move to ELO, with quarterly updates (might take time to implement)

It would be interesting to see what happens.