conclusion I can get at reasonable depth 1.e4 1.d4 1.Nf3 with more than 2*10^9 nodes and less than 1.2*10^10 nodes so they are theory moves that stockfish choose.
I may get more moves if I repeat the analysis.
Now it is possible to analyze everyone of the positions after 1.e4 1.d4 1.Nf3 to continue theory and continue in this way.
questions 1)suppose you give stockfish analyze from clear hash every position in this way 10 times how many lines you are going to get at for the first n plies(for the first ply I get at least 3)? 2)Are we going to get 100% draws or maybe one line is going to be not draw? 3)Is there a tool to do it automatically because I do not intend to do it manually inspite of my interest in the results?
mwyoung
Posts : 880 Join date : 2020-11-25 Location : USA
Subject: Re: opening theory without opening book by stockfish Mon Nov 30, 2020 3:41 am
Uri Blass wrote:
part of the analysis of the opening position both 1.e4 and 1.d4 are reasonable moves.
Stockfish_20112916_x64_bmi2: Found 145 tablebases NNUE evaluation using nn-62ef826d1a6d.nnue enabled
conclusion I can get at reasonable depth 1.e4 1.d4 1.Nf3 with more than 2*10^9 nodes and less than 1.2*10^10 nodes so they are theory moves that stockfish choose.
I may get more moves if I repeat the analysis.
Now it is possible to analyze everyone of the positions after 1.e4 1.d4 1.Nf3 to continue theory and continue in this way.
questions 1)suppose you give stockfish analyze from clear hash every position in this way 10 times how many lines you are going to get at for the first n plies(for the first ply I get at least 3)? 2)Are we going to get 100% draws or maybe one line is going to be not draw? 3)Is there a tool to do it automatically because I do not intend to do it manually inspite of my interest in the results?
Is it possible no. You can not analyze everyone of the positions after any first move. Using Stockfish, it is a type B searcher. So you can not assume it evaluations are correct. No matter the results, even if you could do this automatically with unlimited time.
kai_laskos
Posts : 37 Join date : 2020-11-17
Subject: Re: opening theory without opening book by stockfish Mon Nov 30, 2020 2:17 pm
Uri Blass wrote:
part of the analysis of the opening position both 1.e4 and 1.d4 are reasonable moves.
Stockfish_20112916_x64_bmi2: Found 145 tablebases NNUE evaluation using nn-62ef826d1a6d.nnue enabled
conclusion I can get at reasonable depth 1.e4 1.d4 1.Nf3 with more than 2*10^9 nodes and less than 1.2*10^10 nodes so they are theory moves that stockfish choose.
I may get more moves if I repeat the analysis.
Now it is possible to analyze everyone of the positions after 1.e4 1.d4 1.Nf3 to continue theory and continue in this way.
questions 1)suppose you give stockfish analyze from clear hash every position in this way 10 times how many lines you are going to get at for the first n plies(for the first ply I get at least 3)? 2)Are we going to get 100% draws or maybe one line is going to be not draw? 3)Is there a tool to do it automatically because I do not intend to do it manually inspite of my interest in the results?
There is no such tool AFAIK. For single threaded SF12 I got data that it changes mind in early quiet openings on average once every 5-6 plies (close to that). With this one can estimate how many different lines one gets in several tries to the first n plies for every position. You have to consider transpositions and changing mind for an earlier move.
matejst likes this post
Sponsored content
Subject: Re: opening theory without opening book by stockfish
Home 
