FIDE Rating Consultation

[Ian Jamieson]

"Also a rating system that is suitable for professional players may not be suitable for amateur players and vice versa. Maybe the solution is to have two separate rating systems."

Why not? Didn't the Clarke system had a national list which started at roughly 190 (I think) and then separate "Union" lists. So I propose a FIDE list starting at 2200, and then a separate list for amateurs going up to 2200, and if you get there you get promoted. So on the amateur list, you could do as much fiddling of ratings as you liked, maybe use different K etc.

Thank you for your support and yes, your proposal is along the lines of what I had in mind.

Edit

With the benefit of hindsight it’s a pity FIDE didn’t do this instead of keeping one system but reducing the minimum rating to 1000 or whatever it is now.

[Ian Jamieson]

So I propose a FIDE list starting at 2200, and then a separate list for amateurs going up to 2200, and if you get there you get promoted.

When players from the two different lists play each other, will the game count?

I don’t know what Kevin’s view is but personally, to start with, until both systems settle down, the game wouldn’t count. This isn’t so much different from some games counting for national lists but not FIDE rating and vice versa. In time arrangements could possibly be made for the games to count.

[Ian Jamieson]

Qualification for the professional list could also be by other criteria than rating. I seem to remember reading somewhere that Jacob Aagaard’s definition of a professional player is a player who is capable of taking points off a titled player. You could therefore make such results the qualifying criteria for the professional list.

[Ian Jamieson]

While I am at it, and playing the devil’s advocate to a certain extent, why do we still need to have separate titles for women? I accept that they were probably a good idea / worthwhile in the past but I would argue that their time is past and they should be abolished.

[Paul Cooksey]

He was himself incapable of understanding that players simply didn't want k=11.6.

I might be with him. I think players care more about the results than the methodology.

[Ian Thompson]

So I propose a FIDE list starting at 2200, and then a separate list for amateurs going up to 2200, and if you get there you get promoted.

And what happens when someone over 2200 drops below that level? Demotion to the amateur list?

[Ian Jamieson]

So I propose a FIDE list starting at 2200, and then a separate list for amateurs going up to 2200, and if you get there you get promoted.

And what happens when someone over 2200 drops below that level? Demotion to the amateur list?

Again I don’t know what Kevin’s view is but I don’t see why someone wouldn’t stay on the professional list. Equally well just because someone qualified for the professional list I don’t see why they couldn’t stay on the amateur list as well. These are just my immediate thoughts however and even if these aren’t the right solutions I am more confident that acceptable solutions can be found to these sort of issues than to sorting out one rating list spanning from 2800 to 1000 or whatever the span is.

[Ian Jamieson]

In Portugal and maybe in other countries they use a national list until someone ‘qualifies’ for the FIDE list. These sort of questions must therefore already have been thought about and solutions found.

[Brian Valentine]

Also a rating system that is suitable for professional players may not be suitable for amateur players and vice versa. Maybe the solution is to have two separate rating systems.

Don't we need to first demonstrate that FIDE rating is not suitable over the range? Clearly something needs to be done, but the tables on page 8 of the report show amazing consistency along the top left-botton right diagonals. This may suggest the problems are identical everywhere.

[Roger de Coverly]

In Portugal and maybe in other countries they use a national list until someone ‘qualifies’ for the FIDE list.

For that matter so does the ECF and many other Federations although for UK players the national ratings continue in parallel. It's not a particularly high qualification standard now the minimum FIDE rating is 1000 although when chess for under-10s is rated it might seem that way.

For many years the minimum standard for FIDE ratings was 2000 later reduced to 1800. It was the case in those days that many players under 2200 had higher FIDE ratings than converted ECF ones. The reasons were to my mind quite obvious. When you had players who could only qualify with a decent performance by their standards, the international rating reflected their better results, whilst their national rating would reflect all results, Equally if they had poor results against lower rated opposition, these only counted for the national list if the lower rated players hadn't yet qualified. Going back even further in time, when the minimum rating was 2205, players would claw their way onto the list, only to be thrown off with a poor result at the next tournament.

[Ian Jamieson]

Also a rating system that is suitable for professional players may not be suitable for amateur players and vice versa. Maybe the solution is to have two separate rating systems.

Don't we need to first demonstrate that FIDE rating is not suitable over the range? Clearly something needs to be done, but the tables on page 8 of the report show amazing consistency along the top left-botton right diagonals. This may suggest the problems are identical everywhere.

I confess I haven’t read all of this thread let alone the report. I might read the thread if I have time although I suspect I have more pressing things to do. I’m even less likely to read the paper - maybe I would read the paper if the paper was by someone who I have a higher regard for when it comes to ratings than Sonas. I’m not saying that he doesn’t know what he’s talking about just that I find myself often disagreeing with his conclusions.

[Ian Jamieson]

One thing that does occur to me is that with the advances in IT, or whatever you want to call it, it is easier and easier to acquire the knowledge that you need to become a better player than it was in the past. As a result it’s possible that one or more of the basic assumptions underlying the Elo method no longer hold - don’t ask me which - I’m just saying it’s a possibility - in which case it’s hardly surprising some people think the method no longer works. Similar things have happened e.g in financial circles where the assumptions underlying models have ceased to hold and as a result the models themselves have ceased to work. If it is the case that the basic assumptions no longer hold then I doubt any amount of tinkering will solve the problems unless by accident / serendipity.

Edit

Systems can also work for unexpected or unconsidered reasons and even considered changes to systems can throw systems off balance. Hence if it ain’t broke don’t fix it.

[Roger de Coverly]

. As a result it’s possible that one or more of the basic assumptions underlying the Elo method no longer hold

There's an underlying assumption that "chess strength" is a stable random variable and you use results to attempt to estimate what it is. It's obvious that when applied to inexperienced players, particularly juniors, that assumption is suspect, however the method is still widely used. One of the simplest hacks to deal with improving players is just to throw away previous data and recalculate as if they are new entrants.

[Roger de Coverly]

This was done in immense depth in the Kaggle contest, an amazing resource which in my opinion left little to be said.

Didn't that fail because it used real tournament data? Entrants were able to exploit knowledge of swiss pairing rules to infer who were the stronger players and predict results accordingly. It's not that uncommon for the typical round 1 top half v bottom half to give a clean sweep or near clean sweep to the top guys.

[Ian Jamieson]

. As a result it’s possible that one or more of the basic assumptions underlying the Elo method no longer hold

There's an underlying assumption that "chess strength" is a stable random variable and you use results to attempt to estimate what it is. It's obvious that when applied to inexperienced players, particularly juniors, that assumption is suspect, however the method is still widely used. One of the simplest hacks to deal with improving players is just to throw away previous data and recalculate as if they are new entrants.

Yes, but that would require FIDE to have a system that might actually work. It’s probably one of the reasons why the original FIDE ratings worked reasonably well because how many inexperienced players, particularly juniors, would there have been on the early lists when the rating floor was 2200?Not many I suspect.