CFC vs FIDE Rating Calculations

[Fred McKim]

The general formula is the same for the two systems. The main difference is the memory at which old events vs the recent event are weighed in.

FIDE uses a K factor. This is the number that you multiply your "differential" by. For example FIDE > 2400 have a K of 10. I play a tournament with one game and beat a player with my rating. My expected score is 0.5, but I get 1, so my differential is 0.5 Multiply by the K and I gain 5 rating points.

CFC uses a memory of 25 games. If I again play one game and beat a player of the same rating of me, I gain 400/25 = 16 points.

If you want to work in the same system - the CFC K factor is 32, while the FIDE memory is 80.

Also the memory for CFC players > 2200 is 50 (K of 16), while the K factor for FIDE < 2400 is 15 (memory of 53.3).

All this means that FIDE ratings are less volatile than CFC ratings.

What I wonder about is how all of the thousands of FIDE junior rated players are going to get their points to improve from 1400 to 2000 or wherever they peak at. FIDE only recently lowered the floor below 2000, and I don't see how deflation won't become rampant in a few years. There probably isn't a lot of adult vs junior interaction in a lot of countries.

The only contravening factor is that when 2400+ play 2400- the lower rated players will be more volatile, and thus will over time will put a few points into the system that way, assuming they are more likely to be improving.

It may also be that players quit when there ratings are at an ebb, and thus they leave rating points in the pool that they may have been able to win back in the normal course of a player's +/- a certain mark.

I think we should be all interested to see how FIDE handles the eventual drain on professional player's ratings by the amateurs rise through the ranks.

[Roger Patterson]

Re: CFC vs FIDE Rating Calculations

there are possible inflationary factors which counter the 'improvement deflation' to some degree. For example, if you systematically give new players a higher rating than their strength. In the FIDE system, the simple existence of a rating floor will do that. Some people who are below the rating floor will have a good tournament once and get an inflated rating [which will eventually bleed into the rating pool providing inflation]. The corresponding case of someone who gets a bad tournament below the rating floor doesn't enter the system so it doesn't average out. This effect was discussed in a ChessBase article [you'll have to google it yourself] in the last year or so.

The CFC system rates all comers so it doesn't have this particular issue although there is some evidence to suggest that it does overrate new provisional ratings. See http://victoriachess.com/cfc/extremes_summary.php

[Fred McKim]

Re: CFC vs FIDE Rating Calculations

there are possible inflationary factors which counter the 'improvement deflation' to some degree. For example, if you systematically give new players a higher rating than their strength. In the FIDE system, the simple existence of a rating floor will do that. Some people who are below the rating floor will have a good tournament once and get an inflated rating [which will eventually bleed into the rating pool providing inflation]. The corresponding case of someone who gets a bad tournament below the rating floor doesn't enter the system so it doesn't average out. This effect was discussed in a ChessBase article [you'll have to google it yourself] in the last year or so.

The CFC system rates all comers so it doesn't have this particular issue although there is some evidence to suggest that it does overrate new provisional ratings. See http://victoriachess.com/cfc/extremes_summary.php

Of course another oddity in the FIDE system is that, while an unrated player in CFC would get an initial rating = to their performance rating in all cases, in the FIDE system if the player's score is above 50% they only get 12.5 points for every 1/2 pt above 50%.

This would have lowered Jason Cao's initial FIDE rating. Instead of getting an initial rating 240 points above his opponent's average rating (the normal result of 8-2), he only got 6*12.5 points = 75 points.

[Roger Patterson]

Re: CFC vs FIDE Rating Calculations

Of course another oddity in the FIDE system is that, while an unrated player in CFC would get an initial rating = to their performance rating in all cases, in the FIDE system if the player's score is above 50% they only get 12.5 points for every 1/2 pt above 50%.

This would have lowered Jason Cao's initial FIDE rating. Instead of getting an initial rating 240 points above his opponent's average rating (the normal result of 8-2), he only got 6*12.5 points = 75 points.

well maybe that's a technique motivated to try and mitigate the inflationary aspect I mentioned. My personal opinion is that Jason's actual strength (although this is a moving target) is about ~1900 and is consistent with his Victoria CC rating and performance this weekend at the Jack Taylor Memorial. So, the FIDE rating they gave him turns not to be so far offbase although it seems to be by accident not by mathematical design. [his CFC rating however is lagging badly - it will be years before it catches up.]

The rated players in the U10 group were pretty much all 1700-1900. It makes me suspicious to see such a homogeneous group of ratings for what should be a widely varying group of people.