The First Chess Rating Lists


by Sam Sloan
People are always making preposterous and ridiculous statements about the rating system. Most of those who make such statements know almost nothing about the rating system or how it works. Frequently, the claim is made that Professor Arpad E. Elo invented the chess rating system. This is not true. The chess rating system had been existence for fully ten years before Professor Elo first got involved. Professor Elo made significant modifications and improvements to the existing rating system, but what he did was mathematically trivial, something which any average graduate student of mathematics or statistics could have done with the data already available from the existing rating system. Professor Elo is one of the most overrated persons in chess history.

I happened to be a delegate at the USCF meeting in St. Louis in 1960 when Professor Elo's proposed modifications to the established rating system were first introduced. I was the only person who voted against Professor Elo's proposed changes to the rating system. The reason I voted against was that, under the changes proposed by Professor Elo, the rating system would move slower. As I was still a kid (I was 15 at the time) I believed that my rating would go up quickly and therefore I wanted the older, faster moving system to remain.

The first National Chess Rating list was published in the December 1950 issue of Chess Review magazine, page 354. The first list covered 2306 players and 582 tournaments covering a 30 year period ending on July 31, 1950. The system was devised by Kenneth Harkness. The first rating statistician was William Byland of Texas. He did all of his calculations by hand. The starting point of the rating system was that every player who had an even score in the US Open started with a rating of 2000. The ratings were then calculated backwards and forwards until there was a rating for every player.

The basics of the rating system was that if you achieved an even score in a tournament, your performance rating was the rating of your average opponent. If you got a plus score, your performance rating was the average rating of your opponents plus 10 points for each percentage point of your score above 50%. Thus, if you scored 75%, your performance rating would be 250 points more than the average rating of your opponents. If you scored 100%, your performance rating would be 500 points more than the average rating of your opponents. If you achieved a minus score, the opposite would be true.

Note that this is almost exactly the same as the present system we have today. The only difference is that nowadays, if you score 100% your performance rating is 400 points above the average of your opponents (not 500 points). Similarly, if you score 75%, your performance rating is 200 points above the average of your opponents (not 250 points). In this respect, Elo's modifications to the rating system did not change the system much.

The big change which Elo introduced was that a player's rating went up or down game by game. Under the old system, if you played only one tournament of at least five games within a six month rating period, your performance rating in that one tournament was averaged with the old rating to create your new rating. For example, if you had a rating of 1800 and played only one tournament and had a performance rating of 2000 in that tournament, then your new rating would be 1900. If, however, during that rating period, you had two performances of 2000, then your new rating would be 1933.

The weaknesses of this system are readily apparent. Only round robin or Swiss system tournaments of at least 5 games counted. If a player played four games and then dropped out, presumably his rating would not be affected.

Under the Elo system, each game, rather than each tournament, was rated, except that all the games in a single tournament were rated as a group. If you played an opponent with the same rating as you and won, you gained 16 points and he lost 16 points. If your opponent was 200 points higher than you, and you won, you gained 24 points and he lost 24 points. If you lost, you lost 8 points and he gained 8 points. If you drew, you gained 8 points and he lost 8 points. Note that the sum of what you stood to gain and what you stood to lose always equaled 32. This is now known as the "K" factor. Also note that if you played in a tournament of exactly 12 1/2 games, then your new rating would be exactly the average of your existing rating and your performance rating in the 12.5 game tournament. (Please work this out, if you cannot see this right away.) That is the system which we have today, with only minor modifications, such as a lower "K" factor for higher rated players.

Here are the highest rated players on the first USCF rating list. This list was as of July 31, 1950:

Reuben Fine 2817
Samuel Reshevsky 2770
Alexander Kevitz 2610
Arthur W. Dake 2598
A. C. Simonson 2596
Fred Reinfeld 2593
Arnold S. Denker 2575
Isaac Kashdan 2574
I. A. Horowitz 2558
Abraham Kupchik 2538
David S. Polland 2521
George N. Treysman 2521
Larry Evans 2484
Herbert Seidman 2451
Max Pavey 2442
George Shainswit 2442
Albert S. Pincus 2422
Arthur S. Bisguier 2394
George Kramer 2394
Herman Steiner 2394
Donald Byrne 2392
Weaver W. Adams 2383
H. Hahlbohm 2376
A. E. Santasiere 2368
Robert Byrne 2352
Hermann V. Hesse 2352

Six months later, the second USCF rating list came out. This was published in the April, 1951 issue of Chess Review, page 103. There were few changes from the first list, because there were not many tournaments in those days. Anybody who had not played since January 1, 1948 was not on the new list. Here is the new list of top players:

Reuben Fine 2817
Samuel Reshevsky 2734
Arthur W. Dake 2598
I. A. Horowitz 2558
Isaac Kashdan 2494
Larry Evans 2484
Herbert Seidman 2451
Max Pavey 2442
George Shainswit 2442
Arnold S. Denker 2431
Albert S. Pincus 2422
Arthur S. Bisguier 2394
George Kramer 2394
Donald Byrne 2392
Weaver W. Adams 2383
Herman Steiner 2355
Robert Byrne 2352
Hermann V. Hesse 2352
.......
A. E. Santasiere 2333

Note that something important is happening here. The rating of every player either stayed the same or went down. Not a single player gained rating points. Kevitz, Simonson, Reinfeld, Kupchik, Polland, Treysman and Hahlbohm dropped off of the list due to inactivity. Fine, Dake, Horowitz, Kashdan, Evans, Seidman, Pavey, Shainswit, Pincus, Bisguier, Kramer, Donald Byrne, Adams, Robert Byrne and Hesse stayed the same. Denker lost 144 rating points!!! Steiner lost 39 points. Reshevsky lost 36 points. Santasiere lost 35 points.

The third USCF rating list was published in Chess Review for November, 1951, page 327. This list was as of July 31, 1951. Here are the top players:

Samuel Reshevsky 2747
Reuben Fine 2711
I. A. Horowitz 2565
Larry Evans 2554
Arthur W. Dake 2539
Arnold S. Denker 2504
Robert Byrne 2465
George Shainswit 2444
Isaac Kashdan 2441
Max Pavey 2441
Arthur S. Bisguier 2421
Albert S. Pincus 2421
George Kramer 2396
Donald Byrne 2391
Weaver W. Adams 2390
Edward Lasker 2378
D. H. Mugridge 2359
Edward Schwartz 2358
Albert N. Sandrin 2356
.........
Herman Steiner 2340
........
Hermann V. Hesse 2322
.......
A. E. Santasiere 2304

Now, we see some big changes. The biggest change is the drop of Fine of 106 points!!! Dake loses 59 points. Denker recovers 73 points. Evans, Bisguier and Robert Byrne, all still just kids, move up fast. Evans, who wins the 1951 US Championship, gains 70 points. Robert Byrne gains 113 points. Bisguier gains 27 points. Established player Hesse loses 30 points and Santasiere loses 29 points.

The fourth USCF rating list was published in the April 1952 Chess Review, page 103. It is as of December 31, 1951. Here are the top players:

Samuel Reshevsky 2734
Reuben Fine 2676
Larry Evans 2660
I. A. Horowitz 2545
Arthur W. Dake 2510
Arnold S. Denker 2504
Max Pavey 2502
Robert Byrne 2462
Isaac Kashdan 2455
Alexander Kevitz 2450
Herbert Seidman 2447
George Shainswit 2444
Arthur S. Bisguier 2428
Herman Steiner 2427
George Kramer 2413
Albert N. Sandrin 2363
Donald Byrne 2359
D. H. Mugridge 2359
Sidney Bernstein 2358
Edward Schwartz 2358
Milton Hanauer 2352
A. C. Simonson 2352
.......
Edward Lasker 2342
A. E. Santasiere 2342
Albert S. Pincus 2341
........
Hermann V. Hesse 2322
.......
Weaver W. Adams 2305

Now, we see more big changes and swings. Fine drops another 35 points. Larry Evans, aged 19, gains 106 points!!! Dake drops 29 points. Pavey gains 61 points. Kevitz, Seidman and Simonson return to the list after periods of inactivity.

Throughout this period, a player rated over 2700 was called a grandmaster, as player rated over 2500 was a senior master, a player over 2300 was a master and a player over 2100 was an expert.

As can be seen from the above figures, the general trend of the rating system was down. Established players such as Fine and Dake lost rating points to rapidly improving teenagers such as Evans, Bisguier and Robert Byrne. This process continued and, by 1956, all of the top players had lost points. As a result, the standards were dropped, so that over 2600 was grandmaster, over 2400 was senior master and over 2200 was master. It was also understood by 1956 that points had to be injected into the rating system to compensate for the points being taken out of the system by these rapidly improving young players.

The first actual Elo rating list was published in Chess Review for September 1969, page 260. This was an international list, not a USCF list. Here are the top players:

Fischer 2720
Spassky 2690
Korchnoi 2680
Botvinnik 2660
Petrosian 2650
Larsen 2630
Smyslov 2620
Portisch 2620
Geller 2620
Polugaevsky 2610
Stein 2610
Keres 2610
Tal 2610

Note that Fischer, the highest rated player in the world, was rated 97 points lower than Fine had been on the July, 1950 USCF rating list, 19 years earlier! Also, it seems startling to see players like Keres and Tal with low ratings like 2610 and World Champion Spassky with only 2690. However, the ratings have inflated considerably since that time.



Here are links:
Sam Sloan's chess page

Contact address - please send e-mail to the following address:

Sloan@ishipress.com