CHESS STRENGTH : A Comparison Between Correspondence Chess and OTB Chess

CHESS STRENGTH : A Comparison Between Correspondence Chess and OTB Chess 

              Before we begin to evaluate the comparison of various chess rating values, we have to make an important point. Rating is obtained as a result of games played by chess players. The point I am trying to make is that it doesn't say anything explicitly about the strength of the player or the played game. For example if you take a thousand players with a rating of 2000 and make them play amongst themselves some of the players' ratings can reach 2700.
                We can make two types of comparisons. We can either compare the ratings or the games of the players at the time they were playing OTB or correspondence chess. We won't do the rating comparison because it is quantitative and unnecessary. The reason behind this is that although the FIDE and ICCF ratings of players may be different this is something that may vary according to the character of the player.  While some players are good at fast games the others may close the gap as the thinking time increases. This is something decided in great proportion by the character specialties of a given player. Players with the highest rating in Correspondence chess may only reach 2200-2400 level in OTB chess. However the rating score of Rafael D. Leitao ( a Grand Master in both ICCF and FIDE)  is very  close to each other.  From this we can derive that a player will get a higher rating in the area which he is more inclined to. In fact the correspondence performance of OTB players has always been a matter of curiosity.  Oosterom who is a very strong GM acquired a very strong performance for a while.
                On the other hand rating doesn't say anything about the game that has been played. It is only interested with the score of the game. For example the national rating and the international rating of a player is usually different.
                There is an experiment where a player plays correspondence chess without interfering with the moves of the computer. ( http://chessok.com/?p=24526 ) . Here we see a 2386 rating performance. Considering the technological possibilities   that are available today we can surmise that this performance rating can be increased up to the 2400s. From this point on the rating can only increase by the interference of a human with exceptional positional evaluation ability. We are still in an era where human intervention can improve or change the course of the game. One of the best games that demonstrates the power of correspondence chess and human intervention is the Chytilek-Langeveld game. ( http://chessactive.blogspot.com/2013/01/the-mysterious-chytilek-langeveld-game.html ). In this game there are some moves that the computer could not find. And this is in spite of the PC finding Langeveld's moves after a certain depth. For example Chytilek's move 13. Ng5! couldn't be found even at 31st depth.
                We can say that the rating level of OTB games has increased by 100 points relative to the time when Karpov-Kasparov World Title match was being played. This rating increase is related to the increase of players with high rating.
                We have to add that in some OTB games we encounter 'home preparation' moves which are equal in strength with strong correspondence moves. Home preparation moves are similar to human intervention moves in correspondence chess. The  Nf7 move in the game Topalov – Kramnik game ( http://www.chessgames.com/perl/chessgame?gid=1482320   ) is a very good example to this. The second example is (   http://www.tatasteelchess.com/tournament/livegames )
 the  Aronian –Anand game in the Tata Steel Chess Tournament. These two parties include moves that can only be found in strong correspondence games.  Instead of trying to figure out a rating we can say that the game comprises moves that are above 3000 rating level.
  Another thing is that a correspondence player rarely completes the opening stage of a game poorly because he has access to very many databases that are available.  When you look at the games at the Tata Steel Chess Tournament “Tata Steel Chess 2013” (http://www.tatasteelchess.com/tournament/livegames) in a 2700 rating average game we can find at least one big mistake with analysis by Houdini x64. But when the rating average increases to 2800 the mistake rate decreases more that the rating difference indicates.
The rating that CEGT gives to the strongest engine Houdini 3 in 40/120 games is 3042 http://www.husvankempen.de/nunn//40120new/40_120_ratinglist/40_120_AllVersion/rangliste.html . Under the light of all this information we can come up with a table like the one below.

In this chart red bars represent OTB chess and the blue bars represent correspondence chess. As you can see clearly a 3000 rating level  game of OTB chess has the same strength with rating of a 2400 level game in correspondence chess. Aronian - Anand game and Topalov - Kramnik game reach around 3000 rating level of OTB chess.

              Order Link for  Opening Choices from Chess Active's 16.000.000 game PGN Chess Database



  1. Difficult to understand the diagram. Where exactly are the underlying figures that this diagram is based on...? (Not only me asking.)

    1. OTB players have reached around 2850 with their play and around 3 or 4 CORR players have managed to pass the 2700 level. It is clear that a 'Home Preparation Game' (OTB) has a higher rating as it has more depth in calculation - maybe 2900-3000 level rating. This is considered as an equivalent of a very well played CORR game. And the article draws conclusions from these cornerstones.