FIDE rating differences between countries

Discuss anything you like about chess related matters in this forum.
NickFaulks
Posts: 8452
Joined: Sat Jan 02, 2010 1:28 pm

Re: FIDE rating differences between countries

Post by NickFaulks » Sat Jun 12, 2021 9:03 am

Alex McFarlane wrote:
Sat Jun 12, 2021 7:29 am
(All Play Alls have come down in price for some categories but can be expensive to rate.)
Only if they invlolve highly rated players, norm round robins and above, which isn't what we're bothered about.
If you want a picture of the future, imagine a QR code stamped on a human face — forever.

Paul Cooksey
Posts: 1519
Joined: Fri Oct 21, 2016 4:15 pm

Re: FIDE rating differences between countries

Post by Paul Cooksey » Sat Jun 12, 2021 9:12 am

This is a lazy question, since I imagine I could work it out myself. But on the basis someone else might know and save me some trouble...

If a junior improves by 600 points over a five year period at a steady rate of 10 points per month, how many games do we need to have rated during that period for their rating to be reasonably accurate throughout?

Or am I framing this incorrectly, and the issue is about interaction with the pool of players with accurate ratings not the volume of games required?

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

Re: FIDE rating differences between countries

Post by Roger de Coverly » Sat Jun 12, 2021 9:36 am

Paul Cooksey wrote:
Sat Jun 12, 2021 9:12 am
Or am I framing this incorrectly, and the issue is about interaction with the pool of players with accurate ratings not the volume of games required?
It's both. If you have a closed pool of players who are about the same strength, their ratings don't increase even if their strength does, In practice a totally closed pool doesn't arise, so there's always external players to measure against. Statistical theory points to 30 games being needed for a reliable estimate of strength, but may be making the assumption that the strength being measured is fixed over time.

It can be a weakness of the Elo approach that if you take two identical performances, one by a player unknown to the rating system and the other by one with an established if out of date rating, that they aren't given equal ratings on the subsequent rating list.

Brian Valentine
Posts: 574
Joined: Fri Apr 03, 2009 1:30 pm

Re: FIDE rating differences between countries

Post by Brian Valentine » Sat Jun 12, 2021 9:47 am

Paul Cooksey wrote:
Sat Jun 12, 2021 9:12 am
This is a lazy question, since I imagine I could work it out myself. But on the basis someone else might know and save me some trouble...

If a junior improves by 600 points over a five year period at a steady rate of 10 points per month, how many games do we need to have rated during that period for their rating to be reasonably accurate throughout?

Or am I framing this incorrectly, and the issue is about interaction with the pool of players with accurate ratings not the volume of games required?
I suspect that you need to frame the question differently. I don't know if it will help but this is my basic premise:

As a first order approximation Elo can be written

R1 = R0(1-.058n) + .058nE
Where R0 and R1 are the starting and ending Ratings, n the number of results in the rating period and E the average expected performance. The .058 is linked to k=40 in the usual formula.

One other obvious input into the "600 points" is how wrong the intial rating is.

Paul Cooksey
Posts: 1519
Joined: Fri Oct 21, 2016 4:15 pm

Re: FIDE rating differences between countries

Post by Paul Cooksey » Sat Jun 12, 2021 10:01 am

I am assuming that the initial rating is correct. Maybe I should simplify by assuming all games are played against players with accurate ratings, and then add that complexity later.

I am trying to understand what behaviour I see if the Junior is reasonably active but also improving consistently over a longish period. An average of 5 games a month would be reasonably active in my opinion.

Do I see a reasonable approximation of strength with some lag throughout, a pattern of deviations then corrections to real strength, or a growing gap?

NickFaulks
Posts: 8452
Joined: Sat Jan 02, 2010 1:28 pm

Re: FIDE rating differences between countries

Post by NickFaulks » Sat Jun 12, 2021 10:06 am

Paul Cooksey wrote:
Sat Jun 12, 2021 9:12 am
Or am I framing this incorrectly, and the issue is about interaction with the pool of players with accurate ratings not the volume of games required?
That is certainly the case. Underrated juniors playing rated games against each other does nothing to rebalance the rating system ( though that's not a reason why they shouldn't do it, since those results still provide valuable information ).

Taking into account only games against, on average, reliably rated opposition, I use a simple rule of thumb. For k=40, if a player has nine such games in a rating period, their resulting rating will be roughly the average of their opening rating and their performance. There are similar rules for other numbers of games.

A quick calculation suggests that if your hypothetical player has such a bunch of nine "good" games each quarter, then if they start rated 1400 and worth 1800, they should after six quarters on average be rated 1910 and worth 1980 - so seven months behind. I would regard that as satisfactory.

FIDE's monthly rating lists, which were presented as improving accuracy, are actually a culprit here. They were introduced entirely for PR reasons and I warned that there would be unwelcome side effects, but when the politicians get the bit between their teeth little can be done.
If you want a picture of the future, imagine a QR code stamped on a human face — forever.

Paul Cooksey
Posts: 1519
Joined: Fri Oct 21, 2016 4:15 pm

Re: FIDE rating differences between countries

Post by Paul Cooksey » Sat Jun 12, 2021 10:49 am

Thanks Nick. That does give me some faith in the rating system, that number of games we need to keep up with normal improvement is not excessive.

Walid - I am interested in your experience. How hard is it for a junior to get 9 rated games a quarter against players with accurate ratings?

Wadih Khoury
Posts: 604
Joined: Sun Jul 12, 2020 8:14 pm

Re: FIDE rating differences between countries

Post by Wadih Khoury » Sat Jun 12, 2021 10:57 am

Paul Cooksey wrote:
Sat Jun 12, 2021 9:12 am
This is a lazy question, since I imagine I could work it out myself. But on the basis someone else might know and save me some trouble...

If a junior improves by 600 points over a five year period at a steady rate of 10 points per month, how many games do we need to have rated during that period for their rating to be reasonably accurate throughout?

Or am I framing this incorrectly, and the issue is about interaction with the pool of players with accurate ratings not the volume of games required?
It depends on who they play.
I am away from my computer, so I'll try to remember an analysis I did last week.

If a junior is rated 225 points lower pre pandemic (let's say rated 1600 but true strength of 1825 on Jan 2019), he will win 11 points per game by playing a 1825 and achieving an average of 0.5. It will therefore take about 20 games to catch-up.
However this analysis assumes he plays all games in a single rating period (one month) otherwise it will take longer. Also, it assumes he only plays correctly rated players. Practically, a junior will play 2-3 junior in an average open weekend congress.
This probably pushes the number of games required to 50.
Now if said junior has since progressed to 2000, you can increase this number further. That probably means the junior would need 80-100 games to reach his real rating. That's up to 20 weekend congresses, which would probably take 18 months. During which the junior would have improved further.
Last edited by Wadih Khoury on Sun Jun 13, 2021 8:44 am, edited 1 time in total.

User avatar
Christopher Kreuzer
Posts: 8781
Joined: Fri Aug 06, 2010 2:34 am
Location: London

Re: FIDE rating differences between countries

Post by Christopher Kreuzer » Sat Jun 12, 2021 12:18 pm

Wadih Khoury wrote:
Sat Jun 12, 2021 10:57 am
Now if.said junior has since progressed to 2000, you can increase this number further. That probably means the junior would need 80-100 games to reach his real rating. That's up to 20 weekend congresses, which would probably take 18 months. During which the junior would have improved further.
This is an important point, one I had not fully realised before. Juniors continue to improve as they try and catch up in terms of rating, so that introduces additional lag into the catch up.

My question is this. Obviously juniors that reach IM or GM levels of strength will quickly catch up. But if a junior is just a "normal" player (i.e. one that reaches 150-210 strength, let's say), does it take longer for a lower-rated player to catch up, simply because they are not as strong a player? If so, that is a serious impediment and discouragement for weaker players trying to get an accurate rating.

NickFaulks
Posts: 8452
Joined: Sat Jan 02, 2010 1:28 pm

Re: FIDE rating differences between countries

Post by NickFaulks » Sat Jun 12, 2021 12:30 pm

Christopher Kreuzer wrote:
Sat Jun 12, 2021 12:18 pm
But if a junior is just a "normal" player (i.e. one that reaches 150-210 strength, let's say), does it take longer for a lower-rated player to catch up, simply because they are not as strong a player?
My rule of thumb above still applies. Yes, the less underrated they are, the slower the asymptotic approach to their true level ( in terms of absolute discrepancy ) will be.
If you want a picture of the future, imagine a QR code stamped on a human face — forever.

Brian Valentine
Posts: 574
Joined: Fri Apr 03, 2009 1:30 pm

Re: FIDE rating differences between countries

Post by Brian Valentine » Sat Jun 12, 2021 4:12 pm

NickFaulks wrote:
Sat Jun 12, 2021 12:30 pm
My rule of thumb above still applies. Yes, the less underrated they are, the slower the asymptotic approach to their true level ( in terms of absolute discrepancy ) will be.
I think this statement needs reframing. A more underrated player would at some point become less underrated with the same asymptotic approach thereafter. So a player less underrated should take less time which would appear to be faster.

NickFaulks
Posts: 8452
Joined: Sat Jan 02, 2010 1:28 pm

Re: FIDE rating differences between countries

Post by NickFaulks » Sat Jun 12, 2021 4:28 pm

Brian Valentine wrote:
Sat Jun 12, 2021 4:12 pm
NickFaulks wrote:
Sat Jun 12, 2021 12:30 pm
My rule of thumb above still applies. Yes, the less underrated they are, the slower the asymptotic approach to their true level ( in terms of absolute discrepancy ) will be.
I think this statement needs reframing. A more underrated player would at some point become less underrated with the same asymptotic approach thereafter. So a player less underrated should take less time which would appear to be faster.
I'll stay with my version.
If you want a picture of the future, imagine a QR code stamped on a human face — forever.

Wadih Khoury
Posts: 604
Joined: Sun Jul 12, 2020 8:14 pm

Re: FIDE rating differences between countries

Post by Wadih Khoury » Sun Jun 13, 2021 9:02 am

Paul Cooksey wrote:
Sat Jun 12, 2021 10:49 am

Walid - I am interested in your experience. How hard is it for a junior to get 9 rated games a quarter against players with accurate ratings?
Sorry I missed that question yesterday.
It's not too hard in the south of England for sub 1900-2000 players. Recurring congresses like Hampstead would normally provide you with 15 games a quarter, of which you can expect 6-9 adults. Add the odd 4NCL or congress and it should be fine. The thing is, as per the comments on the thread, it would still take a very long time to get near equilibrium.

NickFaulks
Posts: 8452
Joined: Sat Jan 02, 2010 1:28 pm

Re: FIDE rating differences between countries

Post by NickFaulks » Sun Jun 13, 2021 9:10 am

Wadih Khoury wrote:
Sun Jun 13, 2021 9:02 am
The thing is, as per the comments on the thread, it would still take a very long time to get near equilibrium.
Is a year a very long time? I can remember annual rating lists.
If you want a picture of the future, imagine a QR code stamped on a human face — forever.

Paul Cooksey
Posts: 1519
Joined: Fri Oct 21, 2016 4:15 pm

Re: FIDE rating differences between countries

Post by Paul Cooksey » Sun Jun 13, 2021 9:17 am

We are in danger of slipping into 4 Yorkshire men territory, but I recall when I was a 140 teenager I lost to Simon Ansell who was about 96. I was 180, and he was much stronger than me. Young people today don't really know what underrated is!

But I'd think it would be ok if an improving junior who played actively stayed within 50 points of their rating while they are improving. Do we think we are within that sort of margin if they don't start off with a terrible initial rating? Or is it irrelevant because they always start with a terrible initial rating?

Post Reply