Chess Grading Destruction

[Roger de Coverly]

I think that there would be merit in giving greater weight to more recent games.

If you play less than thirty games in a year, that already happens. If you play 20 games every year, your grade is based on two thirds of this year's result and one third of last year's.

[Stewart Reuben]

>Do typical club players play sufficient games each season to justify going over to monthly grading? Six monthly list seem to work well, perhaps the next step is quarterly?<

Initially FIDE had an annual rating system. It proceeded to 6 monthly, 3 monthly and then monthly over a long period. A daily unofficial list is published for 2700+ players.
Initially FIDE went down to 2200, which was very similar to the BCF. I suggested it go down to 1000 and that took about 10 years from 2000 in 200 points steps. The ECF seems to have no minimum.
Both ECF and FIDE could be daily, if that is what people wanted and the administrators were prepared to cooperate.


Some people seem to think the FIDE regulations regarding rate of play, number of rounds a day, etc. are immutable. If I thought people wanted 90 minute games to be FIDE Rated for players 2200+, I would have recommended a change in the regulations.

[David Pardoe]

The notion that players grade should change on a daily, weekly, monthly, or even six monthly basis, doesn't make a great deal of sense for most players, I don't believe.
But it might depend on what you want from a grade....
Grades simply reflect results, nothing to do with playing strength except perhaps as a broad measure.
Yes, for players rated FIDE 2300+ playing international tournament chess, there may well be merit in more frequent updates.
Maybe too, for fast improving juniors..
For most players, I`d prefer a system that give players a grading code, say a letter, representing a 20 or 25 point grading band. This would be used as a broader guide to `playing strength`, and enable team captains greater freedom to pick players for teams, or for qualifying entry to Congresses, etc..
There are any number of players whose playing strength may not be reflected by grade, who can pull `surprise results` or `shock results`, just as often as they play `to there grade`, or so called strength....
All sorts of factors can come into play....

[NickFaulks]

In fact what I'm saying is that the ECF system over the last few years, whilst of course not perfect, has actually been a much more reliable indicator of recent playing strength than has been FIDE.

Actually, it has been a reliable indicator of recent playing success, which is a very easy thing to measure if that's what you want. The FIDE system attacks the different and much harder problem of producing a best estimate of current playing strength.

The FIDE system is particularly poor for club strength players - it reacts far too slowly because of the too low K factor (despite the recent change) and misses too many games actually played.

No, extensive statistical analysis has shown the current k factors to be close to optimal, as a predictor of future performance. The point about "missing" games is of course important in England, but I don't see how the system is at fault for the local tradition that many results which are perfectly good for FIDE rating are never sent in.

[Brian Valentine]

In fact what I'm saying is that the ECF system over the last few years, whilst of course not perfect, has actually been a much more reliable indicator of recent playing strength than has been FIDE.

Actually, it has been a reliable indicator of recent playing success, which is a very easy thing to measure if that's what you want. The FIDE system attacks the different and much harder problem of producing a best estimate of current playing strength.

This seems to be the place that gets the footfall to discuss things!

Nick, You are going to have to explain this one. I'd have said both systems are trying to do the same thing, but have evolved from different traditions. My starting position with Maurice's posts is that my role is to maintain the best estimate of current playing strength, where best isn't describing the best model, but within the ECF model.

[Roger de Coverly]

I'd have said both systems are trying to do the same thing, but have evolved from different traditions.

The BCF/ECF system uses a performance rating directly to establish an estimate of strength suitable for use as a ranking order. The Elo approach is slightly more nuanced in that it merges the previous estimate of strength to recent performance with weightings (K factors) which vary as to how much the recent results are believed to reflect a change of strength rather than fluctuations.

[MartinCarpenter]

I guess easiest to see how a hypothetical system would react to someone playing at say 170 for 20 games/year for 3 years then produces a ~200 performance in a single 5 game tournament.

How far up do you move their grade? Presuming an 'instant' calculation for clarity.

I suspect that FIDE is the most conservative system here. ECF would be 5*200 and 25*170. So 175, which seems very reasonable to me too. Yorkshire's live grades are mildly biased towards recent games, so you'd likely push up to ~180.

The current ECF set up does of course then 'remember' that boost for a reasonable time period. The 30 game window would have kept it in at full strength for about 15 months then dropped it entirely, producing a big boost up and a corresponding drop on the other side.

As for Michael's questions on the previous page. Well, Yorkshire manages the organisation quite nicely and we're quite a decent scale, so it definitely isn't impossible. You basically need a centralised way of electronically reporting results and a few responsible people for each league. It would obviously be harder to do it country wide though.

As to whether there's any real incentive, I don't know. You have to bear in mind that you're not trying to accurately predict future performance here - that involves being quite cautious. Just to intrigue people. You could logically view http://www.chessnuts.org.uk/ny5/ as some sort of grand experiment in doing it Worth having a look if intrigued.

[Maurice Lawson]

Apologies, I had just mistakenly posted this on to the previous page, by editing a post. So just making the point here....

However, it is inevitable that in time, with further integration and development, we will indeed end up with greater frequency of publication. It would be rather luddite to think any differently. So for example at some stage quarterly publication might be an interim stage before going to monthly - but it will inevitably happen at some stage, and that will happen at the point where the software allows it to happen without the leagues of people having to intervene.

At some point the ECF needs to stop focussing excessively on grading stability and start using the grading system to promote competitive chess. Requiring the system to track back over a full year, regardless of the number of games played, and regardless of the recency of games played, cannot possibly endure as a system and it would be better to recognise that now and start planning for something contemporary and competent which takes chess forwards instead of backwards.

[Michael Farthing]

There is already a perfect grading system that is very up-to-date and responsive; an accurate indicator of current playing ability and well able to generate motivation. The grading stretches back just one game. 1 for a win; 1/2 for a draw; 0 for a loss. I recommend it.

[Roger de Coverly]

Requiring the system to track back over a full year, regardless of the number of games played, and regardless of the recency of games played, cannot possibly endure as a system

There's no technical reason why it doesn't work, provided games are accurately dated to the relevant month. The estimation process for new players and juniors would need a revamp though. It is however a problem as to how you tweak a system designed around annual measurement period to cope with more frequent publication. My suggestion would be to stick with six monthly lists, or even go back to yearly ones, but to supplement this with "year to date". That's similar to what the Welsh and Scots do, namely it's what your grade or rating would be if you played no more games until the next formal rating cut-off. So everyone had two figures.

[MartinCarpenter]

Yearly/year to date makes sense to me (if practical.). January is a slightly odd compromise at the moment, but organisation not easy! You would of course have to explain to people how they've ended up slightly different.

Going back over a whole 'year' will absolutely endure for the July grades because it makes so much sense.

The vast bulk of UK chess is played from mid-late September through to April/early May. So the July grade works to summarise a whole seasons performance/to accurately predict performance in the next one and absolutely should use every game played in that time period.

[Maurice Lawson]

Going back over a whole 'year' will absolutely endure for the July grades because it makes so much sense.
The vast bulk of UK chess is played from mid-late September through to April/early May. So the July grade works to summarise a whole seasons performance/to accurately predict performance in the next one and absolutely should use every game played in that time period.

Martin, although I can understand why you think this, I do not agree that it makes as much sense as you say. Firstly the timing you state is by no means the case these days. Chess has become a year round activity, available to anyone who wants it that way. The way people use it differs, with many playing tournaments whenever these happen to fit in with their diaries. Even if the larger portion of games fits more heavily into a particular part of the year, it doesn't mean that the grading system should be bound by that kind of thing. It is an anachronistic approach, going back some three of four decades and beyond.

Each chess player is an individual, playing for himself, and choosing when to play. And this applies even when he plays in a team match. There are very few players who are truly more concerned about the team result than about their own result - and I see nothing strange about that! The grading system needs to start becoming an up to date moving tool, serving people's diverse interest in the modern game. It needs to keep up with the world as it is, not as it used to be.

[MartinCarpenter]

Some old things are there because they work you know?

All the major leagues and the bulk of the congresses - forming the considerable majority of graded chess I'd think - run outside summer and need sensibly stable yearly summary grades to operate, so people can go on the right boards and the right congress sections etc. That won't ever change.

Teams chess? Some of it is like you say, some of it is rather more serious and then it really is about playing for the team not yourself. One of the teams I play for you just don't get on unless you do that. They're rather different experiences and all different to congresses of course.

[Stewart Reuben]

The ECF system provides an historic record of how a player performed over a certain number of games, or a 6 month period.
FIDE have a different objective. It is to provide an idea of the current playing strength of a player, based on previous data.
I have sometime said, the ECF provides a perfectly accurate number which is perfectly useless.
FIDE provides an extremely inaccurate number which is very useful. I make no pretence to claim accuracy for either comment!

A FIDE Rating turns over in a certain number of games, depending on the k factor.
Turn over = 750/k approximately. Thus
k=10 for players once 2400+ 75 games
k=20 for adults <2400 38 games
k=40 for juniors <2400 19 games
k=20 for all players in blitz or rapidplay.
The attempt has been made to change to k=15 for players 2400+, but that met with considerable resistance.
Work has been done on changing to a Glicko System. There the k factor depends on the level of activity. It makes the current system look simple.

An underrated junior could play more than 20 games in one month and his new rating be higher than any of his performances. When FIDE was annual, Neil McDonald managed this and went up to 2395, without ever having had a performance as high as that.

It is partly about marketing. Imagine explaining to a child that, after a year, you will get a grade. Even 6 months must be a real turn-off. Leonard Barden used to publish junior international lists weekly. Where did that get us? Well, England became the second strongest chess-playing federation and that was one of the reasons.

There is no problem with a league deciding to make the grade at the beginning of the season apply thoughout that season for team eligibility, if that is what is wanted. If you have a rule that a player can only play below another player if <10 grading points higher, then you could use the grade in a month-old list to ensure no problem with communication. For congresses which start shortly after the new grading list appears, the eligibility has to be spelled out with care. e.g. the British U180. A player enters that in good faith at 175. He then becomes 182 before the event starts. 182 is the number that will be used for grading his opponents' games.

[Paul McKeown]

There are very few players who are truly more concerned about the team result than about their own result - and I see nothing strange about that!

I'm glad I haven't been a captain or player of any time you are involved in, in that case. In my experience, players are very aware of what the team needs in any given context and will strive towards that regardless of the consequences for their ECF grade or FIDE rating.