- Blind mode tutorial
lichess.org
Donate

Calculating Piece Activity

#2

I love this. Maybe we can find numeric values for K-MAPS:
King safety - we need this
Material advantage - this we know
piece Activity - this post
Pawn structure - we need this
Space advantage - I think we can compute this easily.

I always wanted to implement a tool that shows the KMAPS letters and colors them red/yellow/green during the analysis.

I love this. Maybe we can find numeric values for K-MAPS: King safety - we need this Material advantage - this we know piece Activity - this post Pawn structure - we need this Space advantage - I think we can compute this easily. I always wanted to implement a tool that shows the KMAPS letters and colors them red/yellow/green during the analysis.
#3

Seems to me that piece activity and space advantage are calculated the same way. You add squares attacked, giving one point for them being in the opponent's side and another for being near the king. You add something for being in the central 4 or central 16 and you get space advantage.

Seems to me that piece activity and space advantage are calculated the same way. You add squares attacked, giving one point for them being in the opponent's side and another for being near the king. You add something for being in the central 4 or central 16 and you get space advantage.
#4

I also see an issue that usually pops up when you are trying to calculate piece sacs. There is no way to differentiate them from hanging pieces. For example you may centralize the queen: great mobility, but it may be in an attacked square. And I think tension should matter here are well: which pieces are on the attacked squares.

I also see an issue that usually pops up when you are trying to calculate piece sacs. There is no way to differentiate them from hanging pieces. For example you may centralize the queen: great mobility, but it may be in an attacked square. And I think tension should matter here are well: which pieces are on the attacked squares.
#5

@TotalNoob69 said in #3:

Seems to me that piece activity and space advantage are calculated the same way. You add squares attacked, giving one point for them being in the opponent's side and another for being near the king. You add something for being in the central 4 or central 16 and you get space advantage.

I would guess that space advantage depends also a lot on the position of the pawns. Maybe space can be calculated by the squares behind the pawns for each side?

@TotalNoob69 said in #3: > Seems to me that piece activity and space advantage are calculated the same way. You add squares attacked, giving one point for them being in the opponent's side and another for being near the king. You add something for being in the central 4 or central 16 and you get space advantage. I would guess that space advantage depends also a lot on the position of the pawns. Maybe space can be calculated by the squares behind the pawns for each side?
#6

As always, it's difficult to say which formula would be more correct, but I certainly think that there is room for improvement. I'm interested to hear some suggestions from you about the formula.

One could look to older Stockfish versions' static evaluation functions for inspiration. Pawns controlling squares can be more valuable than queens controlling squares.

> As always, it's difficult to say which formula would be more correct, but I certainly think that there is room for improvement. I'm interested to hear some suggestions from you about the formula. One could look to older Stockfish versions' static evaluation functions for inspiration. Pawns controlling squares can be more valuable than queens controlling squares.
#7

which site do u use for checking piece activity?

which site do u use for checking piece activity?
#8

What a great idea!
I think the data is pointing us in some very interesting directions:
Can we plug in a large sample size and conclude that if you can achieve a certain value you are strategically winning?
Or if you are under a certain value you are lost?
Can we correlate this to a computer analysis to learn if activity is better than space or over-protection or other hard to quantify compensation?
Can we program a bot to only consider activity and mate threats? What would that bot's rating look like?
Lot's of great questions can arrive from this approach. I am the most interested to learn how much more "activity value" do I need over my opponent before they are essentially in zugzwang.

What a great idea! I think the data is pointing us in some very interesting directions: Can we plug in a large sample size and conclude that if you can achieve a certain value you are strategically winning? Or if you are under a certain value you are lost? Can we correlate this to a computer analysis to learn if activity is better than space or over-protection or other hard to quantify compensation? Can we program a bot to only consider activity and mate threats? What would that bot's rating look like? Lot's of great questions can arrive from this approach. I am the most interested to learn how much more "activity value" do I need over my opponent before they are essentially in zugzwang.
#9

This is truly well done. Both in conception and in execution, it is top notch.

It has long seemed to me that numerically -- quantifiably -- there is a ton that could be done, but that has not been done, to help in the study and with the understanding of chess.

I haven't gotten off fanny to do anything about that. But it's both inspiring and a relief to see others not just try to forge new paths but actually succeed.

This is truly well done. Both in conception and in execution, it is top notch. It has long seemed to me that numerically -- quantifiably -- there is a ton that could be done, but that has not been done, to help in the study and with the understanding of chess. I haven't gotten off fanny to do anything about that. But it's both inspiring and a relief to see others not just try to forge new paths but actually succeed.
#10

Good points! A suggested addition: if I got it right, your method counts control of squares next to the king with triple weight - but not the square of the king itself. This could be controlled, too, in case of a check, and we might want to count it even a bit higher than three.

Good points! A suggested addition: if I got it right, your method counts control of squares next to the king with triple weight - but not the square of the king itself. This could be controlled, too, in case of a check, and we might want to count it even a bit higher than three.