lichess.org
Donate

Finding the sweet spot move!

I really like this, sometimes while studying I want to see more clearly what the other options / less strong options/ that are also good would be, and sometimes that's a little hard to determine. I think that this would be ideal maybe as an advanced setting that one could set up. It looks very nice.

I really like this, sometimes while studying I want to see more clearly what the other options / less strong options/ that are also good would be, and sometimes that's a little hard to determine. I think that this would be ideal maybe as an advanced setting that one could set up. It looks very nice.

"What do you guys think?"

"moves are like a product you are trying to buy"

  • It is more like paths to choose, some have pitfalls.

"Consider a gambit. That move has high chances to bring you an advantage with most adversaries, even if some will crush you."

  • A gambit helps you to quickly defeat weaker players, that you would defeat anyway, but puts you at a disadvantage against stronger players, where you would need most help.

"It's the price that is hard to define." * You get a certain time control, time is your currency. You are free to spend your time as you see fit.

"Not winning, mind you, just not totally losing." * Chess is a draw. The initial position is equal. White has the initiative: the right to move first, which translates to +0.3, not enough to win. So the task is to avoid losing moves and to exploit any losing move from your opponent.

"let's say 100-200 centipawns from optimal" * That is not right. If you go from +0.3 to +0.1 it has no consequences: it is from draw to draw. If you move from -0.9 to -1.1 it is bad: from drawing to losing. If you go from +1.1 to +0.9 it is bad: you let the win slip.

"Qc5 is the best move, but it's a bit high level. Qc3 and Qd2 are OK" * That is right.
"c4 is just as hard to find as the best move, but much worse, while Qe3 is bad and hard to find" * Both are bad and thus to avoid.

"which move to use in your repertoire" * A repertoire is a set of moves that do not lose.

"if the move that wins" * If you have one or more moves that win, you must play one of these.

"risk assuming your opponent won't find it" * No, do not do that. If you can see it, then he probably can too.

"What do you guys think?" "moves are like a product you are trying to buy" * It is more like paths to choose, some have pitfalls. "Consider a gambit. That move has high chances to bring you an advantage with most adversaries, even if some will crush you." * A gambit helps you to quickly defeat weaker players, that you would defeat anyway, but puts you at a disadvantage against stronger players, where you would need most help. "It's the price that is hard to define." * You get a certain time control, time is your currency. You are free to spend your time as you see fit. "Not winning, mind you, just not totally losing." * Chess is a draw. The initial position is equal. White has the initiative: the right to move first, which translates to +0.3, not enough to win. So the task is to avoid losing moves and to exploit any losing move from your opponent. "let's say 100-200 centipawns from optimal" * That is not right. If you go from +0.3 to +0.1 it has no consequences: it is from draw to draw. If you move from -0.9 to -1.1 it is bad: from drawing to losing. If you go from +1.1 to +0.9 it is bad: you let the win slip. "Qc5 is the best move, but it's a bit high level. Qc3 and Qd2 are OK" * That is right. "c4 is just as hard to find as the best move, but much worse, while Qe3 is bad and hard to find" * Both are bad and thus to avoid. "which move to use in your repertoire" * A repertoire is a set of moves that do not lose. "if the move that wins" * If you have one or more moves that win, you must play one of these. "risk assuming your opponent won't find it" * No, do not do that. If you can see it, then he probably can too.

@tpr said:

"It's the price that is hard to define." * You get a certain time control, time is your currency. You are free to spend your time as you see fit.

Yes, that was my insight, too. A move's "cost" is the time it takes to find it, regardless of who does the finding.

"Not winning, mind you, just not totally losing." * Chess is a draw. The initial position is equal. White has the initiative: the right to move first, which translates to +0.3, not enough to win. So the task is to avoid losing moves and to exploit any losing move from your opponent.

Chess is more a solution search in a vast space. To simplify it as a draw is like assuming the Earth is a perfect sphere. Also... cows. What I meant was that it's not the best move, but not very bad either.

"let's say 100-200 centipawns from optimal" * That is not right. If you go from +0.3 to +0.1 it has no consequences: it is from draw to draw. If you move from -0.9 to -1.1 it is bad: from drawing to losing. If you go from +1.1 to +0.9 it is bad: you let the win slip.

OK, I was doing the simplification there, but the essence of the idea remains. Computers will always choose the best move, but we can do well with the second or third best most of the time.

"which move to use in your repertoire" * A repertoire is a set of moves that do not lose.

Depends on the level and on what you're trying to get out of it. For example Will Graif's repertoire is made almost entirely of gambits, for example. Chess is not always about winning, it can be about winning in style.

"risk assuming your opponent won't find it" * No, do not do that. If you can see it, then he probably can too.

You are too strict. I understand what you mean, but as I said, different people get different things from chess. Playing the perfect game is not my goal, playing the fun game is.

@tpr said: > "It's the price that is hard to define." * You get a certain time control, time is your currency. You are free to spend your time as you see fit. Yes, that was my insight, too. A move's "cost" is the time it takes to find it, regardless of who does the finding. > "Not winning, mind you, just not totally losing." * Chess is a draw. The initial position is equal. White has the initiative: the right to move first, which translates to +0.3, not enough to win. So the task is to avoid losing moves and to exploit any losing move from your opponent. Chess is more a solution search in a vast space. To simplify it as a draw is like assuming the Earth is a perfect sphere. Also... cows. What I meant was that it's not the best move, but not very bad either. > "let's say 100-200 centipawns from optimal" * That is not right. If you go from +0.3 to +0.1 it has no consequences: it is from draw to draw. If you move from -0.9 to -1.1 it is bad: from drawing to losing. If you go from +1.1 to +0.9 it is bad: you let the win slip. OK, I was doing the simplification there, but the essence of the idea remains. Computers will always choose the best move, but we can do well with the second or third best most of the time. > "which move to use in your repertoire" * A repertoire is a set of moves that do not lose. Depends on the level and on what you're trying to get out of it. For example Will Graif's repertoire is made almost entirely of gambits, for example. Chess is not always about winning, it can be about winning in style. > "risk assuming your opponent won't find it" * No, do not do that. If you can see it, then he probably can too. You are too strict. I understand what you mean, but as I said, different people get different things from chess. Playing the perfect game is not my goal, playing the fun game is.

"it's not the best move, but not very bad either"

  • It depends on the position.
    In some positions like 18 moves are objectively equally good, while in some positions only 1 move is correct.

"we can do well with the second or third best most of the time"

  • Yes, sometimes the fourth, fifth, sixth... are equally good too, but sometimes there is only 1 or 2.

"Will Graif's repertoire is made almost entirely of gambits"

  • 'It is peculiar but a fact nevertheless, that the gamblers in chess have enthusiastic followers.' - Botvinnik
"it's not the best move, but not very bad either" * It depends on the position. In some positions like 18 moves are objectively equally good, while in some positions only 1 move is correct. "we can do well with the second or third best most of the time" * Yes, sometimes the fourth, fifth, sixth... are equally good too, but sometimes there is only 1 or 2. "Will Graif's repertoire is made almost entirely of gambits" * 'It is peculiar but a fact nevertheless, that the gamblers in chess have enthusiastic followers.' - Botvinnik

@TotalNoob69

I like the concept of assigning price through the cost of spending time on a move. I'm just wondering about something.

  • a candidate move is one that had - at any engine depth - been close enough to it; let's say 100-200 centipawns from optimal
  • the price of a move is the lowest depth at which the move was considered a candidate

If a candidate move is 1-2 pawns (100cp = 1p) from the optimal move eval, then wouldn't a massive amount of moves, be considered a candidate at depth 1? I mean it feels like at depth 1 surely a lot of moves would be classified within 1-2 pawns from the optimal move considering how large a margin that is? And if a large amount of moves has the same price, then that wouldn't give any info.

Earlier you said 'You could use the number of nodes that the engine looked at in order to find the move as decent enough.' But what is decent enough? Also how do we know that the number of nodes it takes would be relevant to the way humans think? And what if the move would suddenly be classified as 'not decent' with more search time after an initial claim of being decent enough?

I hope I understood this correctly.

@TotalNoob69 I like the concept of assigning price through the cost of spending time on a move. I'm just wondering about something. >- a candidate move is one that had - at any engine depth - been close enough to it; let's say 100-200 centipawns from optimal >- the price of a move is the lowest depth at which the move was considered a candidate If a candidate move is 1-2 pawns (100cp = 1p) from the optimal move eval, then wouldn't a massive amount of moves, be considered a candidate at depth 1? I mean it feels like at depth 1 surely a lot of moves would be classified within 1-2 pawns from the optimal move considering how large a margin that is? And if a large amount of moves has the same price, then that wouldn't give any info. Earlier you said 'You could use the number of nodes that the engine looked at in order to find the move as decent enough.' But what is decent enough? Also how do we know that the number of nodes it takes would be relevant to the way humans think? And what if the move would suddenly be classified as 'not decent' with more search time after an initial claim of being decent enough? I hope I understood this correctly.

@RuyLopez1000 said:

If a candidate move is 1-2 pawns (100cp = 1p) from the optimal move eval, then wouldn't a massive amount of moves, be considered a candidate at depth 1? I mean it feels like at depth 1 surely a lot of moves would be classified within 1-2 pawns from the optimal move considering how large a margin that is? And if a large amount of moves has the same price, then that wouldn't give any info.

It was a general concept. For example I am just now coding something that considers the first time a move entered the engine PVs the depth at which it "figured it out". This changes significantly, though, if I configure 1pv, 10 pvs or 500. In other places in LiChess Tools I calculate the winning probability based on centipawns. Anything which is 1% closer to the best estimation is considered good, 10% less is a mistake and 20% is a blunder. The definition of the "candidate move" matters.

I don't know what's the best formula. That's why I was waiting for the likes of @hollowleaf and @jk_182 and @noobmasterplayer123 and @matstc to weigh in :)

I am just saying that even if computers find the moves differently than people, the time (nodes, depth, etc.) they spend to consider a move as possibly good can be considered a cost or price of the move. The higher the cost... err... I don't know. That's what I am trying to figure out.

@RuyLopez1000 said: > If a candidate move is 1-2 pawns (100cp = 1p) from the optimal move eval, then wouldn't a massive amount of moves, be considered a candidate at depth 1? I mean it feels like at depth 1 surely a lot of moves would be classified within 1-2 pawns from the optimal move considering how large a margin that is? And if a large amount of moves has the same price, then that wouldn't give any info. It was a general concept. For example I am just now coding something that considers the first time a move entered the engine PVs the depth at which it "figured it out". This changes significantly, though, if I configure 1pv, 10 pvs or 500. In other places in LiChess Tools I calculate the winning probability based on centipawns. Anything which is 1% closer to the best estimation is considered good, 10% less is a mistake and 20% is a blunder. The definition of the "candidate move" matters. I don't know what's the best formula. That's why I was waiting for the likes of @hollowleaf and @jk_182 and @noobmasterplayer123 and @matstc to weigh in :) I am just saying that even if computers find the moves differently than people, the time (nodes, depth, etc.) they spend to consider a move as possibly good can be considered a cost or price of the move. The higher the cost... err... I don't know. That's what I am trying to figure out.

@TotalNoob69

So Price is time spent, approximated by computer search. And Quality is Eval.

So to find the sweet spot we need to find the optimal combination of these two variables.

So Optimum = f(price) + g(eval), whatever the functions will be. And there's the question of how the variables should be defined.

That is the summary so far. Sounds like a fun one!

@TotalNoob69 So Price is time spent, approximated by computer search. And Quality is Eval. So to find the sweet spot we need to find the optimal combination of these two variables. So Optimum = f(price) + g(eval), whatever the functions will be. And there's the question of how the variables should be defined. That is the summary so far. Sounds like a fun one!

P.S. This idea has been implemented in LiChess Tools v2.4.155 and it looks like this:

Only the color changes so as to not take too much space. Colors:

  • neutral - depth = 1
  • green - depth = 2
  • yellow - depth = 5
  • red - depth = 15

Everything in between is a color gradient.

P.S. This idea has been implemented in [LiChess Tools v2.4.155](https://siderite.dev/blog/lichess-tools---user-manual/#cevalLineOptions) and it looks like this: Only the color changes so as to not take too much space. Colors: - neutral - depth = 1 - green - depth = 2 - yellow - depth = 5 - red - depth = 15 Everything in between is a color gradient.

How does this evaluate this specific position: 4k3/8/8/4P3/3K4/8/7R/6r1 b - - 0 1

It's an almost Philidor position. For black (the defending side), there are 17 moves that draw with all equal evaluation. Bus as a human, I would definitely play Rg6, going into the standard Philidor position which is an easy draw.

I think the best move is the highest rated move (according to Stockfish) that you understand.

How does this evaluate this specific position: 4k3/8/8/4P3/3K4/8/7R/6r1 b - - 0 1 It's an almost Philidor position. For black (the defending side), there are 17 moves that draw with all equal evaluation. Bus as a human, I would definitely play Rg6, going into the standard Philidor position which is an easy draw. I think the best move is the highest rated move (according to Stockfish) that you understand.