Where can I find a description of the "old" USCF rating system - not the Harkness system, but the one after that, that was in use up until about 2001 when the "new rating system" was adopted? I was sure I had a copy of this at one point, but I can no longer locate it. The US Chess Federation Official Rules of Chess, 3rd Edition, has a high-level overview of the old system, but not all of the specifics.

I am particularly interested in a refresher on how the computation of provisional ratings was done. Based on my recollection, I think that for a player's first 20 games, their provisional rating was the average of their performance ratings for every game they played, calculated as "opponent's rating +400" for a win, "opponent's rating -400" for a loss, simply "opponent's rating" for a draw. Then after 20 games, a more Elo-like formula was used, with a fixed K factor of 24 for players rated under 2000, or 18 for players rated 2000 or over. But, I can't find the actual specifications of this anywhere. The 3rd Edition rulebook does talk about calculation of provisional ratings in the way I described, but it's not clear to me from reading the high-level description if the "average of performance ratings" based provisional rating was carried through the entire 20 game provisional period, or if it was done for a smaller number of games and then something else was done for the rest of the provisional period.

In any case, I'm sure that finding a copy of the actual description of the rating system (which the 3rd Edition rulebook says can be had "free for the asking from the USCF" - though I am not sure they'd still have any copies available of the specifications for a rating system that hasn't been used in about 20 years. ;-) ) would answer all of my questions - but I just can't locate it anywhere....

  • I managed to get hold of a 4th Edition rulebook as well (off of archive.org), and it has a chapter with some more extensive information about the rating system (though still not the complete system with all the formulas and everything), which does confirm that the "average performance" provisional rating formula did persist for all of a player's first 20 games. Would still like to see the complete document on the system, though....
    – patbarron
    Commented May 18, 2021 at 2:19
  • I also learned that the K factor for established ratings was 32 for players rated under 2100, 24 for players rated 2100 to 2300, and 16 for players rated 2400 and above.
    – patbarron
    Commented May 18, 2021 at 2:23

1 Answer 1


I found this page from Glickman which contains the old provisional formula. That formula is:

R[post] = (N R[pre] + mR[avg] + (W - L)400) / (N+m)

where R[pre] is the player's pre-tournament rating, N is the number of games upon which R[pre] is based, m is the number of games the player completes in the tournament, R[avg] is the average of the opponents' ratings, W is the number of wins, and L is the number of losses. If the player is unrated, set N=0 and R[pre]=0. The final result is rounded to an integer.

The August 1994 report of the USCF Ratings Committee discusses the old provisional system a little. There are some details in there, like how unrated players were initially treated as if they had a rating of 1000 when calculating the ratings of other unrated players.

  • Awesome, thanks! They discuss a formula like this in the 3rd Edition rulebook too. It turns out that taking the average the performance ratings of all "provisional" games (opponent rating +400 for a win, opponent rating -400 for a loss, and just opponent rating for a draw) and averaging them produces the same result (though you have to have the complete history of the player's provisionally rated play, which may not be ideal). The bit about assuming unrated players to have a 1000 rating (I thought this was 1200...) for computation of provisional ratings is the missing piece.
    – patbarron
    Commented May 18, 2021 at 3:30
  • 1
    According to that Ratings Committee report, "The USCF Rating System sheet does not provide a complete description of the actual implemented algorithm." That might explain why you're not seeing the entire algorithm in other places, like the rulebook.
    – D M
    Commented May 18, 2021 at 3:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.