# Why doesn't FIDE recalculate ratings match after match?

In this question I brought up two rating systems and wondered which is used by FIDE.

Briefly, they are:

• Method 1 Ratings are updated (recalculated in other words) after every match.

• Method 2 Ratings are recalculated after a specific period of time, say after a tournament or once a month.

I found out that FIDE uses method 2 (recalculates ratings once a month).

### My question

Why doesn't FIDE recalculate ratings after every match?

Is there some kind of statistical problem with that? I don't mean to be partial to either method, but there must have been a compelling reason for FIDE to choose method 2 over method 1, and I'd like to know what that is.

I always thought recalculating the ratings match after match would make the system more accurate, but I'm not an expert, so I can't be sure.

• Suppose I first lose to play A and then beat player B, in subsequent rounds. Should I really have a different rating if I beat A and lost to B instead? Commented Dec 20, 2017 at 8:46
• @RemcoGerlich Maybe it would matter if the strengths of A and B are different. If A is rated 2200, B is 1900, and you are 2000, it would matter, I think. Just my opinion, though, I'm not an expert! Commented Dec 20, 2017 at 9:00
• If ratings are updated after every game, they would even be different if A and B had the same rating. Commented Dec 20, 2017 at 9:04
• @RemcoGerlich True, but is that a disadvantage? I should have made my example clearer: if A is rated 2400, B is 1950, and you are 2000, beating A and losing to B would be different from losing to A and beating B. Commented Dec 20, 2017 at 9:11
• @IanRingrose Not unless the same person was playing in both.
– D M
Commented Dec 20, 2017 at 18:05

I do not know what the true answer is but here are a couple of points.

The FIDE ratings started to be calculated sometimes in the 80's and published twice a year. This on its own explains why they were not calculated in real time (it just was not feasible with the technology at the time, well, it was feasible but too costly).

It is quite natural that FIDE did not want to change the system in place, they started publishing the ratings every month but otherwise kept the original system in place. One good reason for this is to keep what works well. Another good reason is the comparability of the ratings across time. If you changed the system to update the ratings in real time then the ratings before and after the change would not be exactly comparable (although the differences would very likely be negligible).

Recalculating the rating after every game is not necessarily more accurate. This depends on what your objective is. A player's performance in every single game depends on several factors like his skill (you can call it strength), his fitness perhaps, luck, etc. It depends which of the factors you actually want to measure. If it is the player's skill then that is a rather long term thing and it makes more sense to average the player's results over some period of time (or a number of games) to eliminate fluctuations that are purely due to luck. It is a non-trivial problem and does not have an obvious solution. I am almost certain that a perfect solution does not exist.