I was at a tournament this past weekend and I saw a player who was rated 100 win an under 1200 section with a perfect score of 5-0. They had not played a tournament since 2005. I asked the tournament director about this and they said once rated always rated. This player played only 5 games in 2005 but now it is 2022. I think players that have left for an extended amount of time in such a situation should have a rating reset so that the section they play in ratings are not inaccurate because of their performance and that they have an accurate rating instead of them having a much lower or higher rating in which they can take advantage of the system. I was wondering what people may think of this. Feel free to mention some ideas.
If a player with an old 100 rating wins an under-1200 section with a perfect 5-0 score, his post-event rating won’t be anywhere near 100 anymore. Tell me, what is this player’s rating now, after the tournament? I’m guessing it would be at least 1000 or so.
Bill Smythe
I once lost a game to a player with a 3-digit rating (900 or thereabouts). He went on to beat an A-player in the next round, and scored 4 out of 5 in the tournament (this was an open tournament). Like the player in Mr. Antonucci’s example, he hadn’t played rated chess for quite a few years and had a stale rating. But, as Mr. Smythe suggests, his rating corrected itself in short order, and within a few years he was a Master.
If the suggestion is that the rating be reset (reset to what?) before the tournament, how would you do that? I have also seen players return to rated chess after a multi-year absence, and be markedly weaker than their old rating would suggest – because they haven’t played at all in the intervening years, and in some cases, were old enough to have lost some strength in any event. A stale rating is likely to be inaccurate, but it could be off in either direction. If the player has an online rating, I suppose that could be used, but I don’t necessarily trust online ratings, and even if there hasn’t been any hanky-panky and it’s an honest rating, online chess is a different game, and I’m not sure how well that rating translates. I would prefer to let the rating correct itself in the manner that Mr. Smythe suggests.
This player is an example of the phenomenon to which Mr. Kosterman refers. http://www.uschess.org/msa/MbrDtlTnmtHst.php?12681247 Note that his rating went up 300 points after his first event in eleven years. Long gone are the days where a player could gain a maximum of 32 points per game. These days ratings stabilize much more quickly.
The CFC handles the ratings of stale U.S. players in a different way. If you haven’t played in Canada for two years, your rating is re-initialized to your (current) U.S. Chess rating not only for pairing and prize purposes, but also for your “pre-rating” in the rating system. I think we have, or have discussed, something similar using FIDE ratings for foreign players, if only because it would be unsavory to have a GM who played one event in the U.S. when he was seven be listed on the wallchart with a 1350 rating.
Alex Relyea
The TD always has the right to assign a rating to any player. However, unless you have pretty good evidence that the player is far better or far worse than the player’s rating would lead you to believe you would be best advised not to tinker with it. As others have stated, the stale rating will correct itself pretty quickly once the player becomes active again. The OP doesn’t state in his example whether or not the TD knew anything about the player’s current ability level before the tournament started.
The event has not been rated yet. Interesting all the opinions and ideas that you all have. Thank you.
100P5 to 1085P10 is pretty impressive.
Alex Relyea
Just as I predicted, remember?
Also, if the pre-event rating was 100, doesn’t that mean it was based on all losses? I thought every player below 150 gets his 100 floor raised by 1 point for every draw and 2 points for every win, or something like that.
Bill Smythe
The player’s first tournament was 17 years ago and all loses. Like someone said before it is impossible to tell if someone is good or better even after so many years of not playing.
Variable minimum floors weren’t implemented until 2008 and were not retroactive, so players who haven’t played in 15 or more years could have a 100 rating.
Also, someone with a XXX/5 rating is still under the special ratings formula, which means that player is not yet eligible for bonus points, which probably limits how high that player could go even after a 5-0 performance.
A suggestion that was made some time ago (I don’t remember by whom) was to look at any player with a rating more than X years stale and see what that player’s rating would be if the player was treated as unrated, then take the higher of the rating starting from the stale rating or the rating as if unrated. This would complicate the ratings program quite a bit, and it might only be used 5 or 10 times a year depending upon the value of X. (And the potential for interaction effects if more than one such player was in a section was not considered.)
I’m pretty sure that idea was not fully analyzed to make sure it was implementable, there are a number of edge cases to consider. Also, it isn’t clear how the blended initialization procedure that was implemented last year (which weights more recent ratings higher than older ones) would be affected by it.
The Executive Board has tasked the Ratings Committee to look into ratings variability, but I’m not sure what they’re studying, so I don’t know how staleness is being taken into account.