FIDE grading..curiosities

[David Pardoe]

Having just completed 9 games in the 4NCL Northern League, I`ve discovered that 5 were against players with no FIDE rating (although they do ALL have officially recognised national UK ratings).
....So they get excluded for FIDE grading purposes.
#And because I didnt score 1 point from the games that were against FIDE rated players, I dont even get a part rating.
This seems very odd, when I`ve scored 50% in the tournament, yet I end up with nothing..
Does anyone else think this is rather strange, given that all 9 games will actually count for ECF rating purposes....
Should the FIDE rating system not take account of other officially recognised grades in calculating ratings. Otherwise you could play endless games in these events against unrated (FIDE) players, and never get a FIDE rating....regardless of whether its just `another number`.....
And yet great science and debate has gone into equivalence between ECF & FIDE ratings. If such equivalence is not recognised then it would appear to be meaningless.
Incidentally, I`ve always been somewhat doubtful about the ECF rating system...and this increases with the move to 6-monthly ratings. Its the `plus or minus 50` factor that I think is suspect, because it gives too much weighting to individual games....particularly where few games are actually played. I reckon that `plus or minus 25` might give a truer picture for most players.

[Roger de Coverly]

Does anyone else think this is rather strange, given that all 9 games will actually count for ECF rating purposes....

As the rules for qualifying for an Elo style rating are at least fifty years old, most of us are used to it. The problem addressed is that if you score 0% , then the only information available to estimate your rating is that you are below x where x is 400 points below your lowest opponent.

National Elo systems such as the Scots, Welsh and Irish may do ECF style estimates but the International Elo doesn't. It has the merit of simplicity and elitism. So you don't get a rating just for turning up, you have to put points on the scoreboard as well.

[David Pardoe]

Hi Roger,
Maybe if the FIDE rating system hasnt changed for 50 years, it is time to take a fresh look at it.
If you cant get a rating (or even a part rating), from a good quality 9 round event where you score 50%, I think its very strange, particularly when you`re dealing with established players who in many cases have good recognisable history. Its not like we`re taking here about people who just jumped off the number 9 bus and decided they fancied a game of chess...

[Roger de Coverly]

Hi Roger,
Maybe if the FIDE rating system hasnt changed for 50 years, it is time to take a fresh look at it.

The rule is straightforward and fundamental to how the International rating system operates. In order to obtain an international rating, you have to score points against players who already have an international rating. The only exceptions are events like the Olympiads which are allowed to award ratings as prizes for meritorious percentage performances. Arguably there are too many of these. There are players out there with CM titles ( rating of 2200) who have probably never exceeded BCF/ECF 150 in their entire career.

There's been enough ratings fraud or attempted ratings fraud over the years for FIDE to be quite correct in not accepting national ratings as a starting point.

[Alex Holowczak]

And because I didnt score 1 point from the games that were against FIDE rated players, I dont even get a part rating.

If you want a half-decent rating, then this is a good thing!

Does anyone else think this is rather strange, given that all 9 games will actually count for ECF rating purposes....

I can't speak for anyone else, but I don't.

And yet great science and debate has gone into equivalence between ECF & FIDE ratings. If such equivalence is not recognised then it would appear to be meaningless.

It's meaningless for tittle-tattle. It's useful in things like the 4NCL where you need to put unrated players in the board order somewhere. An estimate based on their grades is a good idea.

[IM Jack Rudd]

Hi Roger,
Maybe if the FIDE rating system hasnt changed for 50 years, it is time to take a fresh look at it.

The rule is straightforward and fundamental to how the International rating system operates. In order to obtain an international rating, you have to score points against players who already have an international rating. The only exceptions are events like the Olympiads which are allowed to award ratings as prizes for meritorious percentage performances.

There are also all-play-alls; you can get a rating from one of those having scored no points against rated players. (I have constructed a way to make a marvellous rating scam that uses the all-play-alls' method of creating ratings, should you wish to know about it...)

[David Sedgwick]

I have constructed a way to make a marvellous rating scam that uses the all-play-alls' method of creating ratings, should you wish to know about it...

Is it similar to the one allegedly used by Myanmar 12 to 15 years ago, or something brand new?

[David Pardoe]

Jack...No we dont want any tricks or jingoism....
Yes, I can see that the international nature of this creates some potential issues....
Yes, for the top GMs & IMs you need to ensure good levels of credibility....
But for most players, certainly those under say 2100, the accuracy and variability of grades will be such that the precise number does not actually carry much real precise meaning/value...although these things have token/symbolic/prestigious value....and `street cred...` !?!
By this I mean that a grade probably means `a range` in practice. ie, a club player with grade of say ECF 150, in practice probably plays in a range of say 130 - 170 (and maybe more specifically, a range of 140 - 160), for the majority of the time.
This is why I think that gradings might better be expressed in `bands`... ie I`d probably go for 10 point band ranges of `A` to `Z` where `A` was say 260 - 251, `B` was 250 - 241, and so on..... Now, I`d attach more precision to the top end grades...so players graded from 260 - 200 I`d give & publish precise grades to the actual point. Then, for players graded 199 - 170 I `d publish in say 5 - point band ranges. Then, for players graded below this, I`d publish in 10 point band groups. And maybe for players graded U100 I`d probably be tempted to publish as 15 point band ranges.
Its for this sort of reason that I dont think that publishing grades every 6-months really makes a great deal of sence for the vast majority of club players. Its going to cause confusion for league programmes, where grades will change part way through the season. ie, I can imagine issues arising regarding board order, partly caused by freak sequences of short term results. ie, a player who plays few games could string together 4 or 5 wins or loses & his grade suddenly bounces....
Let me give an example to illustrate `typical` variability...
One player in our team had the following results.
He beat a player graded 201, he lost to two players graded U150. He drew with 3 or 4 players graded 180+..
Do you see the huge level of inconsistancy/variability in this performance... This is not an untypical senario for many club players.
So, this brings us to the question... `What does a grade really mean`.....
Further...what does a FIDE grade mean (for most players graded say U 2100..).
And consequently, for most practical purposes, I think we should be able to extrapolate from National grades into equivalent FIDE grades, without necessarily being too concerned about minor variations..... for the purposes of determining gradings that are reasonable & usable.

[David Robertson]

So, this brings us to the question... `What does a grade really mean`.....
Further...what does a FIDE grade mean?

Such questions are beyond the scope of this forum

[Roger de Coverly]

So, this brings us to the question... `What does a grade really mean`.....
Further...what does a FIDE grade mean (for most players graded say U 2100..)

You will probably find any number of different interpretations. This is how I see it.

Both ECF grades and FIDE ratings attempt to measure the relative strength of players. This is expressed by the assertion that if the grade or rating of A is higher than that of B, then against the same set of players there is a confidence that A will score more points than B with the difference in grade or rating expressing just how much better. This is useful for board orders, seeded pairings and eligibility restricted events. You can also attempt to predict results between two players. Both ECF and Elo measure performance over a period of time or number of games but they differ in that for many practical purposes the ECF approach ignores the grade of the player at the start of the period, whereas Elo approachs usually reflect it.

[Paul Dupré]

Let me give an example to illustrate `typical` variability...
One player in our team had the following results.
He beat a player graded 201, he lost to two players graded U150. He drew with 3 or 4 players graded 180+..
Do you see the huge level of inconsistancy/variability in this performance... This is not an untypical senario for many club players.
So, this brings us to the question... `What does a grade really mean`.....

I'm afraid you have this completely the wrong way around.

It's not that grades are a waste of time, they aren't. The problem is in leagues. Where a game of chess should give the better player a greater chance to win, in leagues in doesn't always. Whereas, as you pointed out the weaker player has a subtle advantage, in that the better player has to take unreasonable amount of risks to win.

That is why grades are good and local leagues are bad. At least the local leagues I use to play in are not fair towards the better player.

I was unemployed and could not afford to play adjournments, however many richer and I must say weaker opponents insisted that I play a further season. I did have the benefit of playing at my home venue though, for a minimal cost of only £15.00 - that's why I gave up playing in the Surrey/Hampshire Border league anyway.

[John Upham]

- that's why I gave up playing in the Surrey/Hampshire Border league anyway.

Paul,

You can no longer be bullied into an adjournment in the Surrey Border League : both players must agree.

As far as I am aware zero games were adjourned this season although a non-zero number might have had an adjournment time control agreed at the start of the game.

In fact, we have now an incremental Fischer time control of http://www.borderleague.org.uk/local/timecontrols.php for those who prefer to playing chess rather winning or losing on time.

[IM Jack Rudd]

I have constructed a way to make a marvellous rating scam that uses the all-play-alls' method of creating ratings, should you wish to know about it...

Is it similar to the one allegedly used by Myanmar 12 to 15 years ago, or something brand new?

It starts with a tournament containing the four highest-rated players you can find and the six weakest players you can find. Run an all-play-all tournament between them. Provided the six weak players all score one point against each other, they'll all come on the list with low but nevertheless inflated ratings.

Then you run a tournament with four of those six weak players and six players who are also weak and unrated, but who would probably score close to 100% against your first set of six.

...and so on. It's similar to the alleged Myanmar scam, but it doesn't require any fiddled results - just cleverly constructed fields.

[David Pardoe]

Let me give an example to illustrate `typical` variability...
One player in our team had the following results.
He beat a player graded 201, he lost to two players graded U150. He drew with 3 or 4 players graded 180+..
Do you see the huge level of inconsistancy/variability in this performance... This is not an untypical senario for many club players.
So, this brings us to the question... `What does a grade really mean`.....

I'm afraid you have this completely the wrong way around.

It's not that grades are a waste of time, they aren't. The problem is in leagues. Where a game of chess should give the better player a greater chance to win, in leagues in doesn't always. Whereas, as you pointed out the weaker player has a subtle advantage, in that the better player has to take unreasonable amount of risks to win.

That is why grades are good and local leagues are bad. At least the local leagues I use to play in are not fair towards the better player.

I was unemployed and could not afford to play adjournments, however many richer and I must say weaker opponents insisted that I play a further season. I did have the benefit of playing at my home venue though, for a minimal cost of only £15.00 - that's why I gave up playing in the Surrey/Hampshire Border league anyway.

Paul, the example I quoted was from the 4NCL league.... not your standard `rapid play` local league stuff.
There can be all sorts of factors effecting `performance`, and, as my example illustrates, its probably quite common to get huge variations in performance over the course of several games...particularly for the `typical` good standard club player.
Thats why I think the rating system might be better if it was based on `plus or minus 25 points` rather than the 50 at present, which I think will tend to distort ratings, particularly for those who play say 10 - 15 rated games per season, which is probably not untypical of many club players.
But that brings me back to a broader point about the meaning of grades. To me they are simply broad indicators of playing strength for the vast majority of players. Hence, on this basis, to have these guidelines updated every 12 months is probably more sensible than every 6 months.
But I also think they should be able to read across the various national ratings to incorporate more games when forming a FIDE rating. And certainly in events like 4NCL, which are good quality, they should be able to include most games actually played, in my view. You see FIDE ratings bandied around all over these events, on results sheets, tables, etc...which make it look like these are official numbers, but in fact are full of `estimates`. Maybe they could put an `E` in front of such figures. But my view is that they could probably safely incorporate these (officially recognised National grades) to at least produce partial ratings that are reasonable in most cases. In the longer term, any distortions would tend to self correct as more results data was gathered.
David.. As for this being beyond the scope of this Forum...I feel there is scope for interpretation to reach a balanced view...but I agree we`re not talking about `rocket science` ..although I do reckon you can get fairly high confidence levels that numbers would be reasonable.

[Alan Walton]

Personally, I think there is nothing wrong with the current system of getting a rating

The problem is that people think the 3rd Division (including the Northern League) is a sure fire event to get a full FIDE rating, in reality in all depends how well the team you play for is doing, if they are performing badly you are likely to only get a handful of games against already rated players, and then it is down to yourself to get the necessary score for those games to count towards a possible part rating.

It was different when I got my first rating in the 4NCL over 11 years ago, I played 9 rated players in 11 games in the 1999-2000 second division (18 team swiss), it was very rare that you play many player unrated, and that was when ratings had to be over 2000

Personally, I would say enter 9 round Swiss opens to achieve part/full ratings you are more likely to be guaranteed to get the neccesary volume of rated players