ECF grades compared with FIDE ratings

General discussions about ratings.
Roger de Coverly
Posts: 21291
Joined: Tue Apr 15, 2008 2:51 pm

Re: ECF grades compared with FIDE ratings

Post by Roger de Coverly » Wed Feb 26, 2014 6:12 pm

David Shepherd wrote: This data shows that the juniors are significantly under-rated compared with adults (assuming the ECF rating is a correct reflection of their playing strength which it may not be).
It's an absolute certainty that the published ECF grade isn't a true reflection of playing strength as measured by results, since the ECF add an age related supplement to it. They then compound the issue by not using the grade as published for the next calculation process, preferring to treat juniors as new players. With neither cook present in the parallel FIDE ratings, these lag behind.

The performance of English juniors in International events against their peers from other countries doesn't particularly suggest they are under-rated against fellow juniors by world standards.

Matt Fletcher
Posts: 271
Joined: Wed Mar 09, 2011 9:42 pm

Re: ECF grades compared with FIDE ratings

Post by Matt Fletcher » Wed Feb 26, 2014 6:16 pm

David Shepherd wrote: If the juniors were removed the fit for the remaining players would be different, so the difference between the two is likely to be more than 150 FIDE points.
Actually the fit for adults doesn't change much because there are about 1000 adults and about 200 juniors (in my calculations) so there was far more weighting to the adults in the original fit anyway.

User avatar
David Shepherd
Posts: 912
Joined: Fri Nov 23, 2007 3:46 pm

Re: ECF grades compared with FIDE ratings

Post by David Shepherd » Wed Feb 26, 2014 6:31 pm

Matt, yes I noticed that but I suspect it will impact a bit as those 200 players are nearly all above the mean conversion, my instinct is that it would make about 20 points difference, but that's not based on any calculation so may be way wide of the mark.

Roger I think there is significant evidence that the junior under rating (compared to adults) is not restricted to this country. I also believe that between countries there is a variation in the actual strength of juniors compared to their rating (possibly related to the quantity of FIDE games a junior typically plays in each country).

Matt Fletcher
Posts: 271
Joined: Wed Mar 09, 2011 9:42 pm

Re: ECF grades compared with FIDE ratings

Post by Matt Fletcher » Wed Feb 26, 2014 7:53 pm

David - you're quite right, it does change by about 20 Elo points, was just that I didn't consider that to be a massive change.

It's worth noting the original change between the calculation using the full dataset and just the juniors isn't actually 150, it's about 125. This is because the broad-brush 700 point addition "should be" (based on the data set I'm using) more like 675 across the whole population so on average everyone comes out "under-rated" by about 25 points.

So when you split adults and juniors, the best fit comes out at about 150 points difference.

MartinCarpenter
Posts: 3041
Joined: Tue May 24, 2011 10:58 am

Re: ECF grades compared with FIDE ratings

Post by MartinCarpenter » Thu Feb 27, 2014 10:06 am

Look at the juniors at the last british - iirc considerable net over performance vs their Elo ratings. Suppose you could check the 4NCL/e2e4 events etc too. The juniors in the british will of course have had a decent chance to correct their ratings upwards a bit in previous events, but its a slow process.

User avatar
Robert Jurjevic
Posts: 207
Joined: Wed May 16, 2007 1:31 pm
Location: Surrey

Re: ECF grades compared with FIDE ratings

Post by Robert Jurjevic » Wed Sep 24, 2014 4:07 pm

Please allow me to remark that the formula must be of the form '8*ECF + const = FIDE', i.e., the factor 8 is exactly 8, as that is by a definition (please see the details at http://www.ecforum.org.uk/viewtopic.php ... &start=547 and at http://www.jurjevic.org.uk/chess/grade/ ... malies.htm in 'ECF grade vs FIDE rating scale...' section). Thanks.
Robert Jurjevic
Vafra

Roger de Coverly
Posts: 21291
Joined: Tue Apr 15, 2008 2:51 pm

Re: ECF grades compared with FIDE ratings

Post by Roger de Coverly » Wed Sep 24, 2014 4:29 pm

Robert Jurjevic wrote: Please allow me to remark that the formula must be of the form '8*ECF + const = FIDE'

I'm afraid we've moved on from that. If you compare the ECF grading performance to a FIDE rating performance in the 0 to 200 rating point range against the 0 to 25 ECF point range, which is where a lot of the double rated chess is played, it supports a 7.something conversion. That also comes up when a least squares test is done against players with ratings and grades in both systems. The FIDE Elo uses tables rather than formula.



The ECF decided on a bit of simplicity and used 7.5

http://www.ecforum.org.uk/viewtopic.php ... =15#p84447
Note the comment by Richard Bates.

User avatar
Robert Jurjevic
Posts: 207
Joined: Wed May 16, 2007 1:31 pm
Location: Surrey

Re: ECF grades compared with FIDE ratings

Post by Robert Jurjevic » Wed Sep 24, 2014 5:07 pm

If '8*ECF + const = FIDE' does not work then perhaps '8*ECF + f(ECF) = FIDE' could be used where 'f(ECF)' is some function of ECF, as, as far as I am aware of, the factor 8 is exactly how FIDE and ELO scales differ, by definition.

"Élo (originally) suggested scaling ratings so that a difference of 200 rating points in chess would mean that the stronger player has an expected score of approximately 0.75."

"(In order to keep present ECF grade scale) one should suggest scaling grades so that a difference of 25 (not 200) grading points in chess would mean that the stronger player has an expected score of approximately 0.75 (i.e. 75%)."

May I ask which is a currently adopted formula?

Thanks.
Robert Jurjevic
Vafra

Roger de Coverly
Posts: 21291
Joined: Tue Apr 15, 2008 2:51 pm

Re: ECF grades compared with FIDE ratings

Post by Roger de Coverly » Wed Sep 24, 2014 5:20 pm

Robert Jurjevic wrote: May I ask which is a currently adopted formula? .
The ECF grading manager announced the use of FIDE = 7.5 * ECF + 700 earlier this year.

A very traditional benchmark, that 200 ECF = 2200 FIDE has been reestablished.

If you read the thread indicated, you will see that the reasoning behind this had general support. It only affects calculations in the limited extent where English players play abroad and ask for their grade to include the foreign games. Otherwise it's at the discretion of tournament and league organisers how they convert for eligibility, seeding and board order purposes.

User avatar
Robert Jurjevic
Posts: 207
Joined: Wed May 16, 2007 1:31 pm
Location: Surrey

Re: ECF grades compared with FIDE ratings

Post by Robert Jurjevic » Wed Sep 24, 2014 5:53 pm

Thank you very much for letting me know the formula. I will use it. :)

You need not to worry, I think, about 'FIDE = 7.5*ECF + 700' formula not having factor 8, as mathematically you can write it as 'FIDE = 8*ECF + f(ELO)', where 'f(ELO) = 700 - 0.5*ECF', so you may claim that you have kept the factor 8, nevertheless, some other 'f(ECF)' may give even a better fit.

The only objection to 'FIDE = 7.5*ECF + 700', when comparing it to 'FIDE = 8*ECF + 650', is that it gives for me a bit lower FIDE rating, but that is my fault, as I was, am, and always will be a patzer.


P.S.

Could you please check my calculation of 'f(ELO)'? From '7.5*ECF + 700 = 8*ECF + f(ELO)' it follows 'f(ELO) = 7.5*ECF + 700 - 8*ECF = 700 + ECF*(7.5-8) = 700 - 0.5*ECF'.
Robert Jurjevic
Vafra

User avatar
Michael Farthing
Posts: 2069
Joined: Fri Apr 04, 2014 1:28 pm
Location: Morecambe, Europe

Re: ECF grades compared with FIDE ratings

Post by Michael Farthing » Wed Sep 24, 2014 6:10 pm

Robert Jurjevic wrote:
"(In order to keep present ECF grade scale) one should suggest scaling grades so that a difference of 25 (not 200) grading points in chess would mean that the stronger player has an expected score of approximately 0.75 (i.e. 75%)."

May I ask which is a currently adopted formula?
This is still the notional formula [ie 25 point difference has an expected score of 0.75 (exactly)].
This is easily seen by taking 4 games where the stronger player wins three and loses one where the players already have a gradev difference of 25. The stronger player scores 4x+150 (for the wins) -50 (for the loss) = 4x+100. 4x+100 over 4 games is x+25 leaving him (correctly) with an unchanged grade.

However, this assumes that players with other grade differences will generate results in an intuitively similar fashion (ie if A is twice as good as B and B twice as good as C then A is four times as good as C). In practice this is not the case and a further distortion is also caused by the ECF 40 point rule. The 7.5 multiple that Roger talks about is the figure deduced by looking at the empirical evidence.

User avatar
Robert Jurjevic
Posts: 207
Joined: Wed May 16, 2007 1:31 pm
Location: Surrey

Re: ECF grades compared with FIDE ratings

Post by Robert Jurjevic » Wed Sep 24, 2014 6:14 pm

P.P.S.

Please note that 'f(ECF) = 700 - 0.5*ECF' in 'FIDE = 8*ECF + f(ECF)' may indicate that ECF grades are stretched in comparison to FIDE ratings, as 'f(ECF) = 700 - 0.5*ECF' decreases with ECF, i.e., as one artificially shrinks the FIDE ratings with '- 0.5*ECF', in order to achieve a better fit, which may be in accordance with my finding at http://www.jurjevic.org.uk/chess/grade/ ... malies.htm. Apologises if I am a nuisance.
Robert Jurjevic
Vafra

Roger de Coverly
Posts: 21291
Joined: Tue Apr 15, 2008 2:51 pm

Re: ECF grades compared with FIDE ratings

Post by Roger de Coverly » Wed Sep 24, 2014 6:41 pm

Michael Farthing wrote: The 7.5 multiple that Roger talks about is the figure deduced by looking at the empirical evidence.
It isn't though. Refer back to the post by Richard Bates earlier in the thread where he points out that if you do a line by line comparison of what the FIDE table suggests against the ECF line, you go under 8 for the multiplier at most conversions up to the 75% mark.

The least squares fit and similar methods show this in practice.

Roger de Coverly
Posts: 21291
Joined: Tue Apr 15, 2008 2:51 pm

Re: ECF grades compared with FIDE ratings

Post by Roger de Coverly » Wed Sep 24, 2014 6:53 pm

Michael Farthing wrote: This is still the notional formula [ie 25 point difference has an expected score of 0.75 (exactly)].
The FIDE rating table is at http://www.fide.com/fide/handbook.html? ... ew=article

Reading off for a 75% result, it's at a rating difference of 193 implying a conversion factor of 7.7

At 60%, which is 10 ECF points, the rating difference is 72, implying a factor of 7.2

It accelerates for the bigger differences, so that a 90% result (40 ECF points) corresponds to a rating difference of 366, thus a factor of 9.15.

The other part of the observation is that there are more games where the players are within 25/200 than ones outside that.

User avatar
Robert Jurjevic
Posts: 207
Joined: Wed May 16, 2007 1:31 pm
Location: Surrey

Re: ECF grades compared with FIDE ratings

Post by Robert Jurjevic » Wed Sep 24, 2014 7:08 pm

Might I remark that 'FIDE2 - FIDE1 = 8 * (ECF2 - ECF1)' by definition, and that 8 is 8 and cannot be anything else but 8.

"Elo suggested scaling ratings so that a difference of 200 rating points in chess would mean that the stronger player has an expected score (which basically is an expected average score) of approximately 0.75."

http://en.wikipedia.org/wiki/Elo_rating_system

"Somebody in ECF suggested scaling grades (ratings) so that a difference of 25 grading (not 200 rating) points in chess would mean that the stronger player has an expected score (which basically is an expected average score) of approximately 0.75 (75%)."
Robert Jurjevic
Vafra

Post Reply