Some funny behavior of USCF ratings

Running Chess Tournaments
1

I can understand the mathematical reasoning and need for re-rates. In the long run, I think that the concept is solid and beneficial. The re-rates serve to provide a more accurate chess rating. However, there are some quirks as well. Let me give two examples.

Isn’t it a bit strange for a player whose last tournament was in December 2003 to see his rating suddenly change by 1 point ? I was looking at the rating history of a college teammate and observed just this behavior. Look at USCF ID 12558771.

Along the same lines, what effect do re-rates have on titles earned? One of my students (USCF ID 12682225) jumped from 2166 to 2200 in January. He did not play again for several months and the April rating list showed him as 2200. However, some time in late March, a re-rate caused his rating change to drop from 2166 to 2199. Can he now claim that he became a USCF master at that tournament in January?

Hence my suggestion for you all is this: Many players see their ratings shift by 1 or 2 points due to a re-rate of an old tournament. It is a bit perplexing to see a tournament from many months (years) ago affect your current rating. Therefore I propose a tolerance factor E so that if the net change for a re-rate of one tournament is less than E, then the system will ignore the change for that player completely. I would recommend an E = 3. Note that E only applies to old events–more than one (maybe two) rating supplements in the past.

Fire away!

Michael Aigner
2

I am forwarding your questions to the chair of the ratings committee.

A tolerance factor was considered at one point, but we did not have sufficient experience with rerates at the time to make a decision about it. Also, we have had very few complaints from players because their rating went up or down by a few points because of rerating. (I think we get 2-3 of them a month.)

I think the explanation for this is that inactive players don’t care much about what their rating is, and for active players their rating changes frequently enough that rerates aren’t noticeable.

I think the office’s procedure on ratings-based titles is that if the player’s rating would qualify him for a title at any point after the event has been rerated for the first time, they award it. I have not searched to see how many cases there have been of someone going over a title threshhold (such as 2200) and back as a result of rerates, I’m not aware of anyone having complained about that.

This procedure could probably stand refinement and clarification.

The most recent complaint I can recall about the awarding of the title of Master came from someone (not the player) who was concerned that a player made Master in an event in which he did not play anyone rated over 2100. I think that there may have been a provision in the rating system back in the early or mid 1990’s that a player must face some minimum level of players (such as another Master) in an event in order to make Master from that event, but there is no such provision at this time.

3

I know that people talk about this at local tournaments. They’re probably more confused than upset. As you mention, it is only a point or two. I know a lot of people that check their USCF rating daily, so they DO notice this effect.

Anyways, I think this is more a public relations nuisance. Thanks for forwarding my email to the appropriate committee chair.

Michael Aigner