FIDE ratings: major changes on the way

zeitnot

See https://www.fide.com/news/2538 and https://www.fide.com/docs/presentations/Sonas%20Proposal%20-%20Repairing%20the%20FIDE%20Standard%20Elo%20Rating%20System.pdf: a FIDE call for comments (first link) on major changes to the FIDE rating system (second link), proposed to take effect in January 2024.

The main proposal is to boost all ratings below 2000 by 40% of the gap to 2000, and to make 1400 the new lower limit for FIDE ratings.

I assume the ICU will probably have to follow suit.

macelligott

It's hard to know what the ICU should do. The amount of work needed to properly audit the ICU ratings properly is well beyond what we could expect of a part time ICU rating officer.

But it seems obvious that there has been deflation in the lower ratings.

About eight months ago I played someone rated 1180 who was able to tell me, after the game, that my 7th move (in an obscure variation) was not the main line!

That player is now rated around 1450, but I would guess he is more a 1600 level player of years ago.

zeitnot

It might not be so hard to apply the main change: a one-time (hopefully, anyway) boost to ratings below 2000.

Since players can have results of overseas tournaments counted for ICU purposes, the assumption that ICU and FIDE ratings are roughly consistent is already baked in. I'd have thought there was a good case for simply making the same adjustment that FIDE makes.

macelligott

Results of Galway Blitz

https://www.chessspectator.com/GalwayBlitz2023/index.html?fbclid=IwAR26MWzahqWZMAwiZrECmis1QYelOu2Mhy_wzuOpV-IM8itDyQyK-jh6YxI

RooksPawn

https://www.boards.ie/discussion/comment/120901599#Comment_120901599

Definitely ICU should adopt the Sonas proposals in full and as soon as possible.

Implementation of a 1400 minimum rating should ideally be brought in before the new season starts in September. with the other changes implemented at the same time as FIDE makes them

I remember FM John Delaney (who is a professional statistician) saying several years ago that ICU made a big mistake in allowing very low initial ratings for juniors.

Pete Morriss

Would ICU members rated under 1000/1400 be happy to be denied any rating at all? For many years the ICU has been able to say that one of the few things that it does for ALL members is provide them with a rating, so changing that would have huge implications. Maybe FIDE and the ICU are catering to substantially different audiences, and so should have rating systems that do different things?

zeitnot

https://www.boards.ie/discussion/comment/120936354#Comment_120936354

I said above that the ICU will probably have to follow FIDE if these changes are implemented. What I meant is that the main change (compression of 1000-2000 ratings to 1400-2000) would have to be implemented. I didn't mean to suggest that players below 1400 should be dropped from the lists. The ICU has had measures that handle low-rated players for a long time (see https://ratings.icu.ie/articles/12, for example).

The more I think about Sonas's proposed changes, the less I like them. His explanations seem at best incomplete, and at worst quite muddled. FIDE would be better off delaying the proposed changes for a year and come up with something more structured. Fiddling around with the system every year or two doesn't seem to be the best approach.

Pete Morriss

https://www.boards.ie/discussion/comment/120936405#Comment_120936405

Sorry, I was replying to Rookspawn.

I agree that Sonas's document is not very satisfactory. FIDE announced it by saying that it "seeks public discussion" of the proposals, but its announcement ends by saying that it will make a decision in October to become effective in January 2024, so it obviously doesn't want too much discussion. My (uninformed) guess is that FIDE will go ahead unless it receives a substantial number of dissenting responses from sources it would regard as well-informed, or possibly substantial opposition from Federations. The ICU will then have the difficult task of working out how to rejig its own rating system, which, as you said, does have mitigating features built in that the FIDE one doesn't.

macelligott

In regard to the Irish ratings: I think we are bottom heavy. There are too many players rated in the 600-900 bracket.

many of these players are, in my view, young, vastly underrated and are continuing to improve.

Despite the bonus system, implemented by the ICU, I suspect that as these 600 to 900 rated players continue to play, they will take pointed off established players and lead to deflation.

I would argue that something needs to be done. What that might be, can be debated.

Pete Morriss

"In regard to the Irish ratings: I think we are bottom heavy. There are too many players rated in the 600-900 bracket."

Too many for what?

"Despite the bonus system, implemented by the ICU, I suspect that as these 600 to 900 rated players continue to play, they will take pointed off established players and lead to deflation."

Do you have data to support this? Even without the bonus system, young and new players have a higher K factor than us established players, so if they beat us (as they eventually do, alas) they will gain more points than we lose. This should lead to inflation, not deflation, other things being equal.

I think that there is not much point discussing this without data. The situation over the last couple of years is probably exceptional because of the hiatus caused by covid, and the higher than normal number of new players at present, some of whom have quite high online ratings. The system should settle down in another year or two, so now is probably not a particularly good time to introduce changes. But if FIDE does something drastic in January then presumably the ICU would have to consider how to respond.

sodacat11

The ratings have somehow become ridiculously out of kilter, probably largely due to the Covid break. I am not a weaker player than I was ten years ago but my rating has plummeted because I played in a number of opens where I lost to players rated as low as 1200 yet who had 2100+ rating performances. I have noticed that in under 2000 tournaments I really struggle to maintain my rating but it is easy enough to enter an over 1800 tournament and gain rating points. There are many Irish players grimly hanging on to their over inflated ratings by mostly avoding playing low rated players and there are just as many 16 and 1700 rated players who would more than hold their own in over 1800 rated tournaments. Probably the best way to redress things would be to have no rating restrictions in tournaments for a year or two and just expose everyone to playing players of all strengths and ratings in Opens. A couple of years of Opens only would soon sort out the discrepencies.

zeitnot

https://www.boards.ie/discussion/comment/120951829#Comment_120951829

Would you be in favour of the changes being proposed by FIDE? In particular, the one-time bonus of 40% of the points between current rating and 2000? (I assume the ICU will have to implement the same change.)

sodacat11

https://www.boards.ie/discussion/comment/120953663#Comment_120953663

I can see its merits but my fear is that it would eventually lead to the already over inflated ratings from Carlsen down to 2000 becoming even more inflated. I don't know what the best solution is but at least the problem is being addressed.

zeitnot

https://www.boards.ie/discussion/comment/120953767#Comment_120953767

Thanks. I'm not sure what to think myself. The claimed (simulated) results seem good enough, as far as they go, but the report seems muddled in various ways, which makes me wonder what will happen next.

cdeb

https://www.boards.ie/discussion/comment/120900588#Comment_120900588

Covid has definitely played havoc with ratings. How could it not? But it'll even itself out soon enough (I think it's nearly done). It's not necessarily a reason to rejig the ratings of itself

Eugene Donohoe

https://www.boards.ie/discussion/comment/120955137#Comment_120955137

I don't agree with you. Deflation in the lower ratings has been a steadily increasing issue before covid ever arrived on the scene. I think that starting juniors (age wise) at such a low rating has played havoc with the lower rated group 1000-1500 as fast improvers move up, despite the ratings adjustment that the icu has put in place. A review of that process is long overdue. This is not a criticism of the ratings officer whom I think does an excellent job given he is part time.

cdeb

Has that been shown though? I thought John Delaney (was it?) did something up a few years back which showed ratings inflation wasn't significant. There seems to be plenty of stable-rated players too (again ignoring the post covid blip)

ICU ratings are higher than FIDE ratings (1700 ICU is about 1800 FIDE) but that's a different matter of course

And would have no problems with a review of ratings to see if anything definite can be shown

Eugene Donohoe

https://www.boards.ie/discussion/comment/120971838#Comment_120971838

That's why I think we need a review. Having a review is I think good governance. I don't want to quote/misquote John Delaney so I'll refrain.

macelligott

1158 wins Open Weekender

macelligott

Strange pairings in the open weekender. In R4, the eventual who was on 3 points played someone on 1 point.

I guess accelerated ratings were used as the entry was so big and there were only 5 rounds. I’d have thought accelerated pairings would only be used in the first 2 rounds. But seems to have been used in all rounds. Very strange.

https://chess-results.com/tnr804476.aspx?lan=1&art=0&fed=IRL

Retd.LoyolaCpt

https://www.boards.ie/discussion/comment/120975734#Comment_120975734

A mistake was made (clearly); In the swiss pairings setup, acceleration ran "until r4" meant including round 4 which obviously wasn't the intention. Prizes were adjusted accordingly by band I think. Lesson learned.

Some crazy entry numbers to stay on top of:

Open Weekender: 115 (record I think)

Weekday open: 46 (record I think, double most recent years)

Blitz: 96 (record I think)

Rapid: 113 (record I think)

Challenger: 85 (record I think)

Major: 54 - hmm, possibly a record.. not too sure.

Most of those are double some recent entries.

Pete Morriss

Retd.LoyolaCpt said "A mistake was made (clearly); In the swiss pairings setup, acceleration ran "until r4" meant including round 4 which obviously wasn't the intention."

The problem goes deeper than this. I am not aware of any rationale for having a 2-point pairing bonus in round 2 of a 5-round tournament.

BTW, Isn't this in the wrong thread?

macelligott

In round 3 of the Open Weekender, eight players on two points were each paired against players on zero points. How does that reduce the number on full points? It is beyond my understanding. Anyway, it seems a lesson was learned. Mistakes happen, and once they are not repeated, we make progress.

Pete Morriss

That's what happens when you have a pairing bonus of 2 points: players in the bottom half with two points play people in the top half with zero. The idea is that the higher-rated players will win, thus rapidly reducing the number on two points. That didn't happen in this case: most of these games were won by the lower-rated player (maybe suggesting that ratings are still unreliable after the covid hiatus).

But a much better way to do it is to have a pairing bonus of 1 point from the start, so that these games occur in round 2, leaving more time to ensure that only those in the bottom half who are really good get a tilt at winning the tournament. That's the FIDE-recommended method. (The pairing bonus should also be removed gradually, rather than in one go.)

sodacat11

https://www.boards.ie/discussion/comment/120978383#Comment_120978383

It beggars belief that the mistakes were not noticed during the tournament by the Controller. I am also puzzled as to why contestants weren't aware of what was happening or if they were then why didn't they protest?