Do Attacking Players play more Forcing Moves?

Comments on https://lichess.org/@/jk_182/blog/do-attacking-players-play-more-forcing-moves/1maN1mgD

This seems more to say that winning players win more, in that CCTs are all moves that a player plays on their way to victory. I think to better assess the different styles between great players would be to count these moves in the first 10, 20, 30 moves of the game, than in total.

@DanielWinn said in #2:

This seems more to say that winning players win more, in that CCTs are all moves that a player plays on their way to victory. I think to better assess the different styles between great players would be to count these moves in the first 10, 20, 30 moves of the game, than in total.

I don't get your point about winning players, as all players I looked at won a lot of games.
I also think that including the ends of games makes sense, as there are different ways to win a game. One could win by a sacrificial mating attack or by grinding out a long technical endgame and I think that the conversion of a winning position is also part of the style of a player.

I think it would be better to restrict games to opponents that are world champions or world championship candidates' too.
Most of the games you considered are against weaker players, where forcing moves are more abundant.

I like FOSS projects where results are easy to replicate and build further research upon.

I am aware of previous research https://github.com/2014mchidamb/AdversarialChess so I am curious how your "style" research runs deeper, in case a community wish to build great things together...

@tpr said in #4:

I think it would be better to restrict games to opponents that are world champions or world championship candidates' too.
Most of the games you considered are against weaker players, where forcing moves are more abundant.

I'm always worried about restricting the sample size too much or introducing some sort of sampling bias by choosing only some games. For example, I can imagine that many players play differently when they are in a world championship match.
But comparing how players play against different strengths of opposition could also be very interesting.

@Toadofsky said in #5:

I like FOSS projects where results are easy to replicate and build further research upon.

I am aware of previous research https://github.com/2014mchidamb/AdversarialChess so I am curious how your "style" research runs deeper, in case a community wish to build great things together...

I'm not aware of the project you've mentioned, but I'll look into it.
In general, my idea is to try and extract different aspects of games and see how they compare for different players, instead of using a NN.

@jk_182 said in #3:

This seems more to say that winning players win more, in that CCTs are all moves that a player plays on their way to victory. I think to better assess the different styles between great players would be to count these moves in the first 10, 20, 30 moves of the game, than in total.

I don't get your point about winning players, as all players I looked at won a lot of games.
I also think that including the ends of games makes sense, as there are different ways to win a game. One could win by a sacrificial mating attack or by grinding out a long technical endgame and I think that the conversion of a winning position is also part of the style of a player.

My point is that won games share similarities that don't articulate the stylistic differences, which is what you were trying to elucidate. Your response describes my point: your study doesn't differentiate between someone sacing and someone winning with a bunch of a checks at move 60. So your study obfuscated the difference between styles rather than differentiated them. If instead you looked at when the attacking moves happened - early or late - as your comment points out, you would get a better analytical picture of different styles of great players.

I looked at the same data divided by game results:

So players play more forcing moves when they win, but the more attacking players have more forcing moves across all results. Kasparov even players more forcing moves in his losses than Karpov in his wins.
I should also clarify that I wouldn't use the forcing moves alone to find the style of a player, it will just be a part of the equation.

"The standard definition of a forcing move is a check, a capture or a threat."

How is this a good definition of a forcing move at all, let alone the "standard" one? Would anyone call it a forcing move if the opponent has the same set of good moves to choose from before and after? At best this is a proxy for forcing moves, and not even an especially good one.