Chess Stack Exchange is a question and answer site for serious players and enthusiasts of chess. It only takes a minute to sign up.
Sign up to join this community
I recently went to a USCF tournament. Although it was quick/blitz rated, regular ratings were used for the purposes of pairings and prizes.
This led me to wonder: are regular ratings actually better predictors of results in quick/blitz games than the quick/blitz ratings themselves are?
Probably not technically if everyone played a lot of rapid games, however, I have seen organizers do this before since the regular ratings usually have a more established basis. In other words, because people have usually played a lot more standard rated games, in that respect, they are more accurate. It helps prevent sandbagging.
In reality, in rapid and blitz, a better predictor is probably age. Kids often perform better than their standard rating the faster the game. I once won an under-1800 blitz tournament when I was an under-rated 1100-player (I went up to 1254 after the main tournament). I was about 18, and the mind was very quick back then (there also were not a lot of kids like there are today).
There's a tradeoff involved here. On the one hand, blitz ratings tend to reflect a player's strength in blitz (by definition). But on the other hand, most players play many more standard rated games, meaning that rating could be considered more accurate.
In general though, standard ratings are a reasonable enough indicator of blitz results. Most players who are good tend to also be good in blitz, although there are of course exceptions (older players, for example).
Of course, if you're referring to online blitz ratings vs USCF standard ratings, the online blitz ratings are clearly a better indicator. People play far more online blitz games than OTB games.
Classical ratings aren't designed to predict blitz play in any way. Yes, they might be predictive but you are backwards inferring by association rather than using any actual proof. I would prefer actual numbers that mean something over guesses regardless of how accurate the guesses might be.
No. They are used because too few people have faster ratings in many tournaments.
They are generally as accurate for pairing as other rating would be although some players do play better at fast speeds so it would not be as good as truly VALID speed ratings for ALL players would be.
They might actually be better as the speed rating bracket covers way too wide a time to be valid. As I recall it varies from, which for me would be virtually postal to way too fast to play well. My sweet spot is in the bracket but at one extreme I would do a little worse and at the other terrible bad.
Speculation is all well and good, but I wanted data. I selected an OTB quick tournament from February and compared final scores with initial quick and regular ratings. After removing any previously unrated players, (surprisingly, everyone who previously had a regular rating also previously had a quick rating) the scatterplots looked like this:
The regular ratings did appear to be slightly better predictors in this case than the quick ratings.
Of course, this is a rather small sample size (27 players in one 4-round tournament) and they aren't independent since the players were playing each other. By using score as a measure I'm ignoring strength of opponents. Also, some players received one half or full point bye, and this of course affected their final scores.
Then I took a blitz tournament at the same day and location. (Mostly chosen so I wouldn't have to re-look up the regular ratings of most of the players; there was a lot of overlap in the player pool between these two events.) After removing players with no previous blitz rating, and also removing a withdrawal and a late entry, I got this:
It's a lot harder to say by looking whether the correlation is better between the blitz and the regular. Again, it's just one small tournament (10 rounds, and only 16 players after removing the ones without a previous blitz rating.)
