Great Chess Quotes
September 8, 2019
International Rating List
In the early 1960s, Arpad Elo was working on his rating system for the United States Chess Federation.
It could also be used to rate present and past masters. I can remember reading this in the August 1962 issue of Chess Life and the mention of three standard tournaments remained forever in my memory:
It might be of some interest to describe the procedure that was used to obtain the present ratings. Actually this rating study began with three tournaments, namely, Wiesbaden 1880; Berlin 1881 and Vienna 1882. These three tournaments were selected since the over-all strength of the competition was just about the same in all three; (L. Paulsen for example scored 7.5-7.5; 8-8 and 16.5-15.5 in the respective events) and 15 of the leading players of the day participated in at least two of the tournaments. The samples on these 15 players, who might be designated now as the primary standards, comprised anywhere from 31 to 63 games. As it turns out a 31 game sample of the play of an individual is sufficient to determine the class of the player with what is termed in statistics as “reasonable certainty,” i.e. with a 95% probability. Essentially the methods described in the June 1961 CHESS LIFE article by the writer were used to establish the ratings of the 15 individuals with respect to the norm of the group and this norm in turn was arbitrarily designated as zero rating. The members of the initial group then served as standards of comparison in other events and so eventually as subsequent tournaments were rated more and more players were brought into the rating list. This procedure was carried through an eighty year period right down to 1963 and during the process there were rated: (a) every large international tournament (b) many smaller tournaments of international scope and (c) a great number of tournaments of national scope which, however, include international masters.
Actually it is not essential to rate every tournament in which a player participated to obtain a trustworthy rating for him. It is more important to have samples of his performances which are statistically adequate and which are of recent origin.
Throughout the 80 year period the calculations of the ratings were made with respect to the arbitrary norm mentioned above and only at the end was the conversion made to the USCF scale.
This conversion presented no particular problem wince the many US masters who competed in international events also competed in strictly domestic events with other US players whose ratings were well established on the USCF scale.
Thus essentially only a splicing of the two scales were required since the meaning of the class interval or of any given difference in rating is identical on the two scales. It might be noted, however, that the ratings of some of the US senior masters have been revised as a consequence of their performances in recent international tournaments.
Arpad Elo, International Rating List in Chess Life , August 1962, p. 165
_________
It is of interest to look at that first international rating list. These from that:
Botvinnik 2736
Euwe 2553
Fischer 2713
Geller 2655
Korchnoi 2667
Petrosian 2674
Reshevsky 2597
Smyslov 2601
Spassky 2663
Tal 2706
Yanofsky 2460
September 8, 2019
International Rating List
In the early 1960s, Arpad Elo was working on his rating system for the United States Chess Federation.
It could also be used to rate present and past masters. I can remember reading this in the August 1962 issue of Chess Life and the mention of three standard tournaments remained forever in my memory:
It might be of some interest to describe the procedure that was used to obtain the present ratings. Actually this rating study began with three tournaments, namely, Wiesbaden 1880; Berlin 1881 and Vienna 1882. These three tournaments were selected since the over-all strength of the competition was just about the same in all three; (L. Paulsen for example scored 7.5-7.5; 8-8 and 16.5-15.5 in the respective events) and 15 of the leading players of the day participated in at least two of the tournaments. The samples on these 15 players, who might be designated now as the primary standards, comprised anywhere from 31 to 63 games. As it turns out a 31 game sample of the play of an individual is sufficient to determine the class of the player with what is termed in statistics as “reasonable certainty,” i.e. with a 95% probability. Essentially the methods described in the June 1961 CHESS LIFE article by the writer were used to establish the ratings of the 15 individuals with respect to the norm of the group and this norm in turn was arbitrarily designated as zero rating. The members of the initial group then served as standards of comparison in other events and so eventually as subsequent tournaments were rated more and more players were brought into the rating list. This procedure was carried through an eighty year period right down to 1963 and during the process there were rated: (a) every large international tournament (b) many smaller tournaments of international scope and (c) a great number of tournaments of national scope which, however, include international masters.
Actually it is not essential to rate every tournament in which a player participated to obtain a trustworthy rating for him. It is more important to have samples of his performances which are statistically adequate and which are of recent origin.
Throughout the 80 year period the calculations of the ratings were made with respect to the arbitrary norm mentioned above and only at the end was the conversion made to the USCF scale.
This conversion presented no particular problem wince the many US masters who competed in international events also competed in strictly domestic events with other US players whose ratings were well established on the USCF scale.
Thus essentially only a splicing of the two scales were required since the meaning of the class interval or of any given difference in rating is identical on the two scales. It might be noted, however, that the ratings of some of the US senior masters have been revised as a consequence of their performances in recent international tournaments.
Arpad Elo, International Rating List in Chess Life , August 1962, p. 165
_________
It is of interest to look at that first international rating list. These from that:
Botvinnik 2736
Euwe 2553
Fischer 2713
Geller 2655
Korchnoi 2667
Petrosian 2674
Reshevsky 2597
Smyslov 2601
Spassky 2663
Tal 2706
Yanofsky 2460
Comment