- Blind mode tutorial
lichess.org
Donate

Computer eval of Fischer-Random (960) starting positions?

#1

I've seen plenty of discussions about 960. I won't even get into my thoughts on that but, has anybody ever actually seen evaluations of the starting positions in 960? It almost strikes me that there are positions which are nearly lost on move one. I don't know that as a fact, but I'm sure somebody has looked into it. I just know that when I watch things like the freestyle chess coverage, I see a lot of pretty good players who are not so sure what to do with move one. That tells me there are a lot of alligators in the water that you need to get across. Is there anybody who's ever done an eval of all the positions?

I've seen plenty of discussions about 960. I won't even get into my thoughts on that but, has anybody ever actually seen evaluations of the starting positions in 960? It almost strikes me that there are positions which are nearly lost on move one. I don't know that as a fact, but I'm sure somebody has looked into it. I just know that when I watch things like the freestyle chess coverage, I see a lot of pretty good players who are not so sure what to do with move one. That tells me there are a lot of alligators in the water that you need to get across. Is there anybody who's ever done an eval of all the positions?
#2

Sesse's supercomputer did an analysis of all 960 positions back in 2018, and the highest Stockfish evaluation was 0.57 for position 80, https://lichess.org/analysis/chess960?position=80.

Santiago Beltrán (@justiniano565) wrote an article about critical Chess960 positions in the 2025 edition of Stefan Löffler's Chess Calendar where he examines position 80 and several others. The conclusion is that Black's defensive task is difficult, but that it is possible to survive.

Sesse's supercomputer did an analysis of all 960 positions back in 2018, and the highest Stockfish evaluation was 0.57 for position 80, https://lichess.org/analysis/chess960?position=80. Santiago Beltrán (@justiniano565) wrote an article about critical Chess960 positions in the 2025 edition of Stefan Löffler's Chess Calendar where he examines position 80 and several others. The conclusion is that Black's defensive task is difficult, but that it is possible to survive.
#3

Thanks for read my article. I based it in Sesse as first approximation and using with my experience as 960 player for including position that I felt specially uncomfortable . Sesse's Analysis was made in 2016 or so and needs an update, but I don't think that it will change very much. What was unbalanced in 2016 is still today. I tried to demonstrate that black doen't lose the most difficult positions and find a defense.
Position 80 is hard for black only starting with 1.g4 any other first move loses that starting adventatge. I think that I found a cunning defense for black. At least for do not lose.
I rewrote the article for Yearbook edited by Arno Nickel trying to improve it
Of course I am not happy with result, how much deep is needed without trying to create an 'opening theory '?

Thanks for read my article. I based it in Sesse as first approximation and using with my experience as 960 player for including position that I felt specially uncomfortable . Sesse's Analysis was made in 2016 or so and needs an update, but I don't think that it will change very much. What was unbalanced in 2016 is still today. I tried to demonstrate that black doen't lose the most difficult positions and find a defense. Position 80 is hard for black only starting with 1.g4 any other first move loses that starting adventatge. I think that I found a cunning defense for black. At least for do not lose. I rewrote the article for Yearbook edited by Arno Nickel trying to improve it Of course I am not happy with result, how much deep is needed without trying to create an 'opening theory '?
#4

Any actual study of opening moves from any starting position results in the development of opening theory for that position. Look at how deep opening theory is for 518 and you can imagine that, given enough effort and time (= "study") that theory will be developed for all positions. It's just overwhelming to the human mind and that was part of Fischer's insight.

Any actual *study* of opening moves from *any* starting position results in the development of opening theory for that position. Look at how deep opening theory is for 518 and you can imagine that, given enough effort and time (= "study") that theory will be developed for all positions. It's just overwhelming to the human mind and that was part of Fischer's insight.
#5

@zwenna
Thanks! I figured somebody would know. As I opened that analysis window on my tablet here, at depth 42 it's saying +0.8. I don't know if it needs a little more time to arrive at .57 or not , but wow...

I didn't really know what to expect, but I would not have been surprised if there had been positions greater than +1. Some positions just seem really imbalanced at times when I've watched the freestyle chess events. I've never looked at stats on it but as I watch top level 960 being played, it seems to me it's far more decisive. And at least one thing seems very unintuitive to me, it seems like there are an awful lot of wins from the black side.

And thanks also to @justiniano565, I'll look up that article.

@zwenna Thanks! I figured somebody would know. As I opened that analysis window on my tablet here, at depth 42 it's saying +0.8. I don't know if it needs a little more time to arrive at .57 or not , but wow... I didn't really know what to expect, but I would not have been surprised if there had been positions greater than +1. Some positions just seem really imbalanced at times when I've watched the freestyle chess events. I've never looked at stats on it but as I watch top level 960 being played, it seems to me it's far more decisive. And at least one thing seems very unintuitive to me, it seems like there are an awful lot of wins from the black side. And thanks also to @justiniano565, I'll look up that article.
#6

At the other side of Sesse most unbalanced positions are the most balanced of them, mainly those with one bishop in each corner that forces to play a kind of double fianchetto and there are 60 of them if I do not miscalculate.
And how to play for white to get some adventatge . Maybe it is for a new article

At the other side of Sesse most unbalanced positions are the most balanced of them, mainly those with one bishop in each corner that forces to play a kind of double fianchetto and there are 60 of them if I do not miscalculate. And how to play for white to get some adventatge . Maybe it is for a new article
#8

I wonder if the evaluation would change if we wouldn't use the "off-the-shelf" stockfish, but train it instead on a very specific opening position.

I wonder if the evaluation would change if we wouldn't use the "off-the-shelf" stockfish, but train it instead on a very specific opening position.
#9

@nadjarostowa said in #8:

I wonder if the evaluation would change if we wouldn't use the "off-the-shelf" stockfish, but train it instead on a very specific opening position.
I don't understand what you mean by train it to a position?

@nadjarostowa said in #8: > I wonder if the evaluation would change if we wouldn't use the "off-the-shelf" stockfish, but train it instead on a very specific opening position. I don't understand what you mean by train it to a position?
#10

@V1g1yy said in #9:

I don't understand what you mean by train it to a position?

Modern engines train their networks by playing themselves. I think typically you play games not only from the start position, but from several opening positions - but they probably are all from standard chess.

One might have the hope / expectation that if you start those games from one of the 960 starting positions, this might give a little boost - as the learned patterns for standard chess (especially in the opening with many pieces on the board) might not be relevant, but some others might apply.

@V1g1yy said in #9: > I don't understand what you mean by train it to a position? Modern engines train their networks by playing themselves. I think typically you play games not only from the start position, but from several opening positions - but they probably are all from standard chess. One might have the hope / expectation that if you start those games from one of the 960 starting positions, this might give a little boost - as the learned patterns for standard chess (especially in the opening with many pieces on the board) might not be relevant, but some others might apply.

This topic has been archived and can no longer be replied to.