Why is white leading by 0.3 pawns at the begining of the game? if both players (engines) play perfect, will it be a draw or what happens to the 0.3? does it go down or up the longer you calculate? if you would calculate for infinity would white get 1 pawn or 0 pawns because 0.3 is not possible.
Nobody knows the result of a perfect game. Engines don't know how to play perfectly as little as humans do.
Values like 0.3 are an approximation used by engines instead of the unknown truth.
Values like 0.3 are an approximation used by engines instead of the unknown truth.
Well, if we could hypothetically calculate to the infinity, it would eventually say mate in 100 or whatever or it would say 0, meaning it can be a forced draw. I’m not exactly sure what stuff like 1.5 or 2.8 means, but I’m guessing it’s the average advantage of all the best moves for a player.
"if you would calculate for infinity would white get 1 pawn or 0 pawns because 0.3 is not possible."
After 2 hours the arbiter would come and call it a loss for running out of time.
After 2 hours the arbiter would come and call it a loss for running out of time.
Check this page out: hxim.github.io/Stockfish-Evaluation-Guide/
Using this page you can click on the categories on the left hand side and find out the evals for those categories for the given position *without look-ahead*. The code to combine these is the "main evaluation".
You can enter an FEN at the top right.
A javascript implementation of the chosen evaluation term is shown.
The factional points come from converting the internal values to the normal value system; pawn=1 etc.
When Stockfish is calculating its move it looks-ahead until it reaches what it considers "terminal nodes" and then uses code analogous to what is on this page to get a value for that terminal position. It then backs up the tree, basically using min/max of the values and the alpha/beta algorithm, to get a final evaluation for the positions off the original position. This is what you see when you are looking at the moves and scores Stockfish is considering. This is a simplified account of what is happening. The actual Stockfish code is more complex.
Using this page you can click on the categories on the left hand side and find out the evals for those categories for the given position *without look-ahead*. The code to combine these is the "main evaluation".
You can enter an FEN at the top right.
A javascript implementation of the chosen evaluation term is shown.
The factional points come from converting the internal values to the normal value system; pawn=1 etc.
When Stockfish is calculating its move it looks-ahead until it reaches what it considers "terminal nodes" and then uses code analogous to what is on this page to get a value for that terminal position. It then backs up the tree, basically using min/max of the values and the alpha/beta algorithm, to get a final evaluation for the positions off the original position. This is what you see when you are looking at the moves and scores Stockfish is considering. This is a simplified account of what is happening. The actual Stockfish code is more complex.
1. e4
33 % white victory
42 % draw
25 % black victory
1. d4
34 % white victory
43 % draw
23 % black victory
So white has an advantage and stockfish sees this advantage too.
33 % white victory
42 % draw
25 % black victory
1. d4
34 % white victory
43 % draw
23 % black victory
So white has an advantage and stockfish sees this advantage too.
This topic has been archived and can no longer be replied to.