Thats fascinating regarding the age gap between Fischer and his rivals at the time!!
Was Sammy Reshevsky Underated
Sort:
Uhohspaghettio1 wrote:
Botvinnik_the_6th wrote:
Uhohspaghettio1 wrote:
I guess everyone likes to tell the dramatic tale of Fischer standing up to the Soviet machine and beating them despite not having the huge advantages of living there around the old greats. Reshevsky is sort of neither here nor there because he was from that part of the world but came to the US when he was 9 and proceeded to become among the world's elite. He's sort of inconvenient to the epic tale of Fischer being the only one to do anything like it.
So yeah, I guess he is sort of underrated. Keep in mind also though that he was the generation before Fischer, and he never became world champion. He died in 1992, meaning most people today have no experience of him being up there among the elite.
Fischer was better of course, but Reshevsky could hold his own to say the least and actually drew their match.
Everyone loves the epic tale of Bobby Fischer. Often forget to mention most of the players he was beating were more than a decade older than himself. Bobby got destroyed when they were on equal footing and made a return in the late 60s. Then he dominated the old competition (apart from Spassky who was only 6 years older than Fischer) and when the next generation emerged in Karpov, we know how that went.
The players Fischer was beating in the early 70s, which eventually culminated in his huge rating gap (1970-72) were Bent Larson (more than a decade older than Fischer), Petrosian (15 years older than Fischer), Korchnoi (more than a decade older than Fischer), Efim Geller (20 years older than Fischer), Lev Polugaevsky (more than a decade older than Fischer), Botvinnik retired in 1970 but he was still in the top 10 at age 60 (33 years older than Fischer).
Reshevsky played the best his entire career but no, he never became world champion, I think the most interesting question is who is greater, Fine or Reshevsky?
Also 10 years or 20 years in chess was a much longer time back then than it is today. Chess evolved very rapidly during those years, chess material and opening developments and theory exploded, which greatly favoured the young, upcoming Fischer above the older generation.
Despite his suspicious-sounding name Reshevsky was also Polish not Soviet, so unless he also spoke Russian wouldn't have had access to all the Soviet books written in Russian, and as stated moved to the US when just 9 years old.
I completely agree with all of that, the lack of access to Russian material is a really good point. The vast majority of chess literature was being published in Russian from the 50s onwards. You're also right, theory was moving extremely fast. Partly why it was so hard for Botvinnik to keep his title, because he was focused on his doctorate, his family and his house to study and he fell well behind theory.
@81
"The "calculating backwards" makes unsupported assumptions of what players without elos would have been rated based on results vs players who later had elo ratings."
++ You still do not get it. Normal elo calculation is forward. Results from chess games are processed as they come in. What chessmetrics did was performing the same calculation backwards. So Alekhine got his rating from games against e.g. Botvinnik. Capablanca got his rating from his games against e.g. Alekhine. Lasker got his rating from his games against e.g. Capablanca. Pillsbury got his rating from his games against e.g. Lasker and Tarrasch. Steinitz got his rating from his games against e.g. Lasker. Zukertort got his rating e.g. from his games against Steinitz. Anderssen got his rating e.g. from his games against Zukertort and Steinitz. Morphy got his rating e.g. from his games against Anderssen.
It is just the same calculation we know, except backwards in time instead of forward in time.
A player's rating is purely objective, but how does one measure whether someone is "underrated" or "overrated"?
The question, I believe, comes down to whether we "like" a player, more or less relative everyone else...It is a subjective call...
Jan 30, 2024
0
#86
tygxc wrote:
@81
"The "calculating backwards" makes unsupported assumptions of what players without elos would have been rated based on results vs players who later had elo ratings."
++ You still do not get it. Normal elo calculation is forward. Results from chess games are processed as they come in. What chessmetrics did was performing the same calculation backwards. So Alekhine got his rating from games against e.g. Botvinnik. Capablanca got his rating from his games against e.g. Alekhine. Lasker got his rating from his games against e.g. Capablanca. Pillsbury got his rating from his games against e.g. Lasker and Tarrasch. Steinitz got his rating from his games against e.g. Lasker. Zukertort got his rating e.g. from his games against Steinitz. Anderssen got his rating e.g. from his games against Zukertort and Steinitz. Morphy got his rating e.g. from his games against Anderssen.
It is just the same calculation we know, except backwards in time instead of forward in time.
You still do not get it. "Alekhine got his rating from games eg against Botvinnik", but Botvinnik did NOT have a rating when he played Alekhine. The rating used in the calculation came from at least 15 years later. Then you take Alekhine's dubious rating and use it to rate games he played against Capablanca 10 years before that. Every step back in time just increases the margin of error.
@86
You still do not get it.
1971 ratings were calculated from 1970 ratings and 1970 results.
1972 ratings were calculated from 1971 ratings and 1971 results.
1973 ratings were calculated from 1972 ratings and 1972 results.
....
2024 ratings were calculated from 2023 ratings and 2023 results.
The statistician Jeff Sonas on the chessmetrics site extended this calculation backwards in time.
1969 ratings were calculated from 1970 ratings and 1969 results.
1968 ratings were calculated from 1969 ratings and 1968 results.
1967 ratings were calculated from 1968 ratings and 1968 results.
...
1884 ratings were calculated from 1885 ratings and 1885 results.
mpaetz wrote:
tygxc wrote:
@81
"The "calculating backwards" makes unsupported assumptions of what players without elos would have been rated based on results vs players who later had elo ratings."
++ You still do not get it. Normal elo calculation is forward. Results from chess games are processed as they come in. What chessmetrics did was performing the same calculation backwards. So Alekhine got his rating from games against e.g. Botvinnik. Capablanca got his rating from his games against e.g. Alekhine. Lasker got his rating from his games against e.g. Capablanca. Pillsbury got his rating from his games against e.g. Lasker and Tarrasch. Steinitz got his rating from his games against e.g. Lasker. Zukertort got his rating e.g. from his games against Steinitz. Anderssen got his rating e.g. from his games against Zukertort and Steinitz. Morphy got his rating e.g. from his games against Anderssen.
It is just the same calculation we know, except backwards in time instead of forward in time.
You still do not get it. "Alekhine got his rating from games eg against Botvinnik", but Botvinnik did NOT have a rating when he played Alekhine. The rating used in the calculation came from at least 15 years later. Then you take Alekhine's dubious rating and use it to rate games he played against Capablanca 10 years before that. Every step back in time just increases the margin of error.
You are thinking about this in a very wrong way.
Suppose there was never any rating system up to now and we had to come up with one. We decide to use a elo formula which happen to have the exact same parameters as FIDE have used. We start with the earliest results we could where there was a decent amount of recorded games among top players and all the elo ratings are calculated from that right up until today.
Now let's say we find out about the modern rating system and we notice Carlsen has a different rating compared to our one, say we have him at 4000. All we have to do to callibrate ratings is literally substract 1,100 or whatever it is from Carlsen and every other player. As we are using the exact same formula with the same value parameter for the spread of players (eg. 300 rating point difference gives the lower rated player a 10% chance of winning"), everything will work out.
That is literally all these calculations are. There is literally no, nada, ZERO difference between these ratings and if FIDE had begun rating players earlier - how coiuld there be? FIDE is the only one allowed to use mathematics and it can only be done at a particular moment in time?
You say that "the further you go back in time the more error there is" - that's only true in the same way it is for all rating systems as you go back in time. For example the current rating system is strongly believed to have inflated over time, though it can't be proven. I don't like this "calculating backwards" idea, nothing is really calculated backwards, that's just confusing people, especially considering that Elos are updated more than once a year so you should be doing your calculations using the updated values.
Jan 31, 2024
0
#89
You are still using ratings from a later point in some players' careers as a measure of their strength to determine the rating of those who played them earlier. 1930s ratings are being calculated on the basis of 1950s results of one of the players.
Here is the full rating profile of Reshevsky from 1934 to 1988.
http://www.chessmetrics.com/cm/CM2/PlayerProfile.asp?Params=199510SSSSS3S108065000000111000000000000010100
Here is the full rating profile of Fine from 1932 to 1954.
http://www.chessmetrics.com/cm/CM2/PlayerProfile.asp?Params=199510SSSSS3S037943000000111000000000000010100
Reshevsky and Fine played 20 games: Reshevsky won 5, lost 1 and drew 14.
https://www.chessgames.com/perl/chess.pl?yearcomp=exactly&year=&playercomp=either&pid=20102&player=&pid2=11209&player2=&movescomp=exactly&moves=&opening=&eco=&result=
@78
"The elos you're referring to are simply estimated elos." ++ No they are calculated backwards.
"The elo system wasn't used until way later. FIDE adopted elo officially in 1970." ++ Yes, but based on those the calculation was performed backwards with the historical results.
The "calculating backwards" makes unsupported assumptions of what players without elos would have been rated based on results vs players who later had elo ratings. The possibility of inaccuracy is high.