Resetting your rating?

[kishanpattni]

I have been playing occasional FIDE rated chess for around 2 years now. I learnt only recently that your FIDE ratings can not be reset (please correct me if I a mistaken). This is unlike ecf standard or rapidplay grading I believe, where a player can return after taking a break and find a noticeable difference in their grade. Obviously one of the main appreciations of having ratings and grades is to give an idea of the strength of a player. Surely I must not be the only one to notice a high number of unrealistic differences between FIDE ratings and national grades. I'm wondering if changes to the current system would be desirable if not already in the horizon?

[Roger de Coverly]

Surely I must not be the only one to notice a high number of unrealistic differences between FIDE ratings and national grades. I'm wondering if changes to the current system would be desirable if not already in the horizon?

FIDE organises meetings from time to time. Here's a report by ChessBase on one of them
http://www.chessbase.com/newsdetail.asp?newsid=6401

As commented in that report, you have to get worldwide acceptance of any changes. You wouldn't for example want to make a change that improved the rating reliability of an amateur playing a handful of internationally rated games if it damaged the ratings of a Grandmaster playing a hundred a year.

As with every statistic, reliability can improve with measurement. A national rating based on thirty games should be more reliable than an international rating based on ten. Tinkering with the international system would not change this.

FIDE apply minimum time standards for international rating. Much of the British league and Congress chess is played at too fast a rate to be eligible.

[Stewart Reuben]

Mikko Markkula, Chairman of the Qualification Commission, has suggested that a player's FIDE Rating should be deleted after 10 years. I disagree and it will no doubt be discussed in Turkey at the QC Meeting. Anybody can attend and anybody can speak. It is very democratic.
I disgree because o my own experience in the US although that was not FIDE Rating. I left the US in 1965 with a rating of about 2150. I didn't play there again until 1991. I told the arbiter of my old rating and recommended he inform the USCF of this. My Rating rose over a period to 2240 where it again stabilised. I think having me start again from scratch would have been less statistically sound. I was not a stronger player in 1991, they had had i nflation in their system.

The difference with the ECF System is that the grade is an average of your last year's results. Some games may be imported from earlir times to bring the number up to 30. The FIDE system is a continuous calculation. For somebody of my strength and experience my rating turns over in 50 games. For an inexperienced player 25. The ECF could go over to a rolling grade which would enable them to have much more frequent list.

The whole matter of the revised FIDE Rating Regulations will be discussed in Turkey. As I wrote on another thread, anyboy can attend and anybody can speak. It is very democratic.

I hope by now most people realise blitz and rapidplay Ratings will start being collated from 1 January and the first two lists will appear 1 July. There will be no charge of any type for these ratings in 2012.

[Andrew Stone]

I played someone in Belgium last weekend who knew Professor Elo. Apparently the rating system only works well for the top players. It does seem important what your first rating is. If your first rating is well below your real strength (ie a bad tournament or two) then you could take years to correct it. For juniors of course by the time they have done this they are much stronger anyway so need to catch up again.

[Christopher Kreuzer]

As I wrote on another thread, anyboy can attend and anybody can speak. It is very democratic.

Actually, you wrote that earlier in the same post...

I hope by now most people realise blitz and rapidplay Ratings will start being collated from 1 January and the first two lists will appear 1 July. There will be no charge of any type for these ratings in 2012.

But charges could be brought in later, presumably if the service is popular? How many amateur blitz tournaments are likely to get rated anyway? Isn't this more for the GM blitz tournaments (are they not already rated?).

[IM Jack Rudd]

How many amateur blitz tournaments are likely to get rated anyway? Isn't this more for the GM blitz tournaments (are they not already rated?).

You can look up the current criteria for rating at the moment. Do GM blitz tournaments satisfy those criteria?

[Stewart Reuben]

I have just realised that I idiotically repeated myself in my previous posting! But Christopher got there first.

The Rating system Arpad Elo introduced is not the same as the current FIDE Rating System. It only went down to 2200, whereas from 1 July it will go to 1000. It appeared just once a year and from July will appear 12 times a year. He took the average of the opponents' ratings and decided the expected score from that. Now each game is rated separately. It was nonsense to average numbers which were not related linearly.
If you are active enough, it dosn't matter all that much what your first rating is. In 25 games it will turn over completely because your K=30. But infrequent activity provides problems and these will be exacerbated when the lists become monthly.
Take a player new to the list who is rated 1600 from 10 games. In his next several events he plays at 1710. He plays 10 games in each rating period.
Start 1600.
Period 1 1.5 x30 = 45 gain.
Period 2 he is 1645 27 gain, but his K now becomes 15.
Period 3 he is 1659
Period 4 he is 1670
Period 5 he is 1679 5 gain.
Period 6 he is 1685 and so on. That is 60 games and he still hasn't reached his actual playing strength.

But had he played 25 games in the first period he would have got to the magic 1713, without ever having played as well as 1713.

No system is perfect. If you want to get a high rating, first play 9 games at a rating much lower than your playing strength. Then wait until the next period and play 60 games (the maximum). You will overshoot your rating big time. Or get to 2399 and wait for the next period. Then play at least 50 games at your true higher strength. The new blitz and rapidly ratings will have safeguards to avoid this at my suggestion.

I think Jeff Sonas at one time recommended ALL standardplay, rapidplay and blitz be rated in the same list.

For the coming year I suggest that all club-night blitz tournaments be submitted for FIDE Rating. But, of course to get on the FIDE List as ENG, people will have to be members. I have suggested there be a lower fee for those players who will be FIDE Rated solely on the blitz or rapidplay list. But I have received no response. Well, it's breaking news.

[Alex Holowczak]

For the coming year I suggest that all club-night blitz tournaments be submitted for FIDE Rating. But, of course to get on the FIDE List as ENG, people will have to be members. I have suggested there be a lower fee for those players who will be FIDE Rated solely on the blitz or rapidplay list. But I have received no response. Well, it's breaking news.

We're likely to be holding the British Blitz in Birmingham next March, and I've been speaking to Andrew Farthing about this. It'll be FIDE-rated, and there's no requirement for anyone to be an ECF Member to play in it.

[E Michael White]

In 25 games it will turn over completely because your K=30.

For somebody of my strength and experience my rating turns over in 50 games.

Presumably you are using the approximation
games=800/k.

Except in unusual patterns of opponents grades the above formula relies on the approximation which you described here as nonsense.

He took the average of the opponents' ratings and decided the expected score from that. Now each game is rated separately. It was nonsense to average numbers which were not related linearly.

The ECF could go over to a rolling grade which would enable them to have much more frequent list.

It would also cause a faster rate of decay by deflation and overall spread.

[Stewart Reuben]

No I used the FIDE figures as in B.02.12.3. It was pointed out to me quite a good approximation is 750/K = the number of games to turn the rating over. Of course it only works even approximately if all the opponents have the same rating, but it does provide a fair estimate. Certainly once differences exceed about 350 everything becomes very iffy. I don't know where 12.3 came from (it is in my 1997 book so it probably was my own calculation. At that time ratings were indeed averaged.), but it seems a closer approximation than either 700 or 800.

The statement 'It would also cause a faster rate of decay by deflation and overall spread.' is an assertion. There are all sorts of factors affecting possible inflation or deflation in a rating system and correcting factors can be introduced. One never commented on, if a player rated 2400 loses a game against somebody 2328, then he loses 6 points. his opponent gains 9. I have never understood why that doesn't lead to an unstable situation. But then I'm not a statistician.
Moreover one can hardly hold the ECF system up as a paragon of the virtues.

Alex >We're likely to be holding the British Blitz in Birmingham next March, and I've been speaking to Andrew Farthing about this. It'll be FIDE-rated, and there's no requirement for anyone to be an ECF Member to play in it.<

But will they be required to be members to have their FIDE Blitz rating published? Adam said yes. Of course all those who already have FIDE Ratings who are ENG are required to be members of the ECF.

[Alex Holowczak]

Alex >We're likely to be holding the British Blitz in Birmingham next March, and I've been speaking to Andrew Farthing about this. It'll be FIDE-rated, and there's no requirement for anyone to be an ECF Member to play in it.<

But will they be required to be members to have their FIDE Blitz rating published? Adam said yes. Of course all those who already have FIDE Ratings who are ENG are required to be members of the ECF.

Yes, I believe that's true.

[Roger de Coverly]

Yes, I believe that's true.

So how does that work then? You run a ten player all play all blitz. This can be FIDE blitz rated at no cost if you have four players with existing ratings. If none of of the other six players are ECF members (at £ 27 a head), what do you submit to FIDE? Or are you going to give players a rating, then partially take it away again?

Lest there be any doubt, membership is an ECF rule, not a FIDE one.

[Gareth Harley-Yeo]

I'd love to reset my falling rating - my only worry is it will come out even worse than my current one!

[Alex Holowczak]

Oh dear. I seem to have said a buzzword...

Yes, I believe that's true.

So how does that work then? You run a ten player all play all blitz. This can be FIDE blitz rated at no cost if you have four players with existing ratings. If none of of the other six players are ECF members (at £ 27 a head), what do you submit to FIDE? Or are you going to give players a rating, then partially take it away again?

I'm not running an All-Play-All blitz. I'm running a Swiss. I don't know what the policy on an APA would be because I'm not running one, so didn't bother to ask.

Lest there be any doubt, membership is an ECF rule, not a FIDE one.

In other news, water is wet.

[Roger de Coverly]

I'm not running an All-Play-All blitz. I'm running a Swiss. I don't know what the policy on an APA would be because I'm not running one, so didn't bother to ask.
.

So how does it work for a Swiss? Players without FIDE ratings should get a part-rating, but are you going to not submit their results if they aren't members? If so, is this option available for a standard play Swiss?

Stewart is trying to promote the idea that a club evening blitz could be rated. Perhaps so, but what conditions are the ECF applying?

The point about membership being an ECF requirement is worth repeating, particularly as the ECF website implies the contrary, that it's FIDE insisting on individual membership. There's a ten year history of players with ENG international ratings being told they had to be individual members of the ECF.

http://www.englishchess.org.uk/?page_id=2

- It is a FIDE requirement that “to be included in the FRL or FIDE Rapidplay Rating list, a player must be a member of a national chess federation which is a member of FIDE” (FIDE Handbook B.02, paragraph 16.1).