Dedicated Chess Machine Elo vs Human Elo, a least squares analysis

Discussion of chess software programming and technical issues.

Moderators: hgm, Rebel, chrisw

Uri Blass
Posts: 10280
Joined: Thu Mar 09, 2006 12:37 am
Location: Tel-Aviv Israel

Re: Dedicated Chess Machine Elo vs Human Elo, a least squares analysis

Post by Uri Blass »

pedrox wrote: Sun Sep 01, 2019 6:31 am Uri,

The article tries to establish a relationship between the Aktiv list of dedicated chess machines and the FIDE list. The fact that the Elo FIDE starts at 1000 points or the Elo USCF at 100 is not something critical for what the article tries to explain. The article is not recent, maybe 4 years old and as you know it is possible that at this time the minimum EIo FIDE has been going down from 1400 to 1200 to 1000 and who knows if in the future there will be Elo FIDE minors.

The article tries to explain that if the official USCF conversion formulas between the Elo FIDE and Elo USCF are used, then the Elo of the dedicated machines of the Aktiv list seems to have a good relationship when dealing with Elo (FIDE) above 2000. And if the official formula for lower Elo is used then the relationship between the Aktiv list and the FIDE list does not seem to fit reality, for that reason a different formula is proposed in the article. But all this has to be seen from the point of view of dedicated chess machines.

You are right that there may be USCF players with Elo of 100 and these appear to be class J (100-199). But even so I believe that it is possible to start in some cases with USCF of 750 or with different ones according to other criteria. It is indicated in the document in section 2.1 point 5. But I am unaware of the USCF system.
Fide reduced the minimal rating of players to 1000 elo more than 4 years ago.
You can see from table 4 that there were players with fide rating 1000-1299 already in 2010.
https://en.chessbase.com/post/problems- ... or-players

Note that based on my experience the fide rating for low rated players is not something that we can trust and I guess that the difference in playing strength between fide rating of 1100 and fide rating of 1400 is clearly smaller than the difference between fide rating of 2300 and fide rating of 2600.

I doubt if the formula that translate 1050 fide rating to 300 CCRL rating is correct.

Note that the lowest CCRL rating at 40/40 is 1377 and if we talk about 40/4 CCRL rating list the random mover has a rating of 260 CCRL elo because they test buggy engines so I do not know how you can compare CCRL elo of numbers like 0 100 or 200 in the table

I do not believe that 1691 of TSCP is translated to above 2000 fide rating at 40/40 time control.
User avatar
pedrox
Posts: 1056
Joined: Fri Mar 10, 2006 6:07 am
Location: Basque Country (Spain)

Re: Dedicated Chess Machine Elo vs Human Elo, a least squares analysis

Post by pedrox »

There is a thread in which Larry Kaufman indicates that he estimated that it could be passed from Elo CCRL to Elo FIDE with:
Elo FIDE = 0.75 x Elo CCRL + constant.

Kai seems to have done a study and gave the formula:
Elo FIDE = 0.7 x Elo CCRL + 840

I guess the Kai formula is for CCRL 40/40 and I don't know if the formula for CCRL 40/4 could be different, but seeing the Elo of engines in a list and another one does not observe a difference greater than 50 points, so I think the formula could be worth it.

This is the formula used in the table that I put and I also indicated that in my opinion Kai's formula does not work well for Elo below 1400 FIDE.

I agree with you that you cannot compare a random move engine with a player with an Elo FIDE of 1000 points. I have checked games of players with 1000 points of Elo FIDE and in general in most moves do not make serious tactical errors and the strategy is clearly superior.

For these low Elo, for me is better a formula such as:
(1) Elo FIDE (1300-1500) = 0.75 x Elo CCRL + 700
(2) Elo FIDE (1000-1300) = 0.8 x Elo CCRL + 560
(3) Elo FIDE (less than 1000) = 0.85 x Elo CCRL + 420

In this way an Elo FIDE of about 1000 points would be equivalent to an Elo of CCRL of about 682, an engine with this Elo is EasyPeasy, maybe the human still wins but it seems tighter. But this is all theoretical since here you do not have data to compare.

If TSCP has an Elo CCRL of 1691, with kai formula:

Elo FIDE = 0.7 x 1691+ 840 = 2023

I do not know if this is true, but I am sure that TSCP will easily win players with Elo FIDE of 1700 and also easily win a dedicated machine like the Mephisto Roma 68020 in which many experts consider it has a value greater than 2000 Elo FIDE points. I don't have other data to compare.