This gamemode is so difficult that I usually make like 5 blunders a match. This one however had only 2 inaccuracies in 26 moves!
https://adjva4.dpdns.org/JbkwrzDB/white#51
Also if you have any high-quality Crazyhouse games feel free to post them :)
https://adjva4.dpdns.org/pBMvil1E
Acpl in crazyhouse in strange. You can have played such a beautiful mate and have very high acpl anyway. Since all the pieces value are different from standard chess, an opening that you'd use in standard would be a suicide in crazyhouse and a tactic that you'd use in crazy would be a suicide in standard.
I personally don't use acpl as indicator of how I played well a certain game in crazyhouse. Above all when you realize that some engine moves in the variant are such non human like.
In that game I also don't understand why it is 0-0-0 as there was a mate in 2 and I missed it, I promoted on c8 completely forgetting Bg4 was covering that. Nonetheless it's always pleasant to see that the engine thought you played a nearly perfect game.
Very nice game @Deadban!
At this point I just don't really understand what moves are good/bad anymore...
For example I feel like move 14 is only slightly an advantage for white while it's +14
@Deadban I think when you are up so much material the enigne doesn't consider slower mates even as inaccuracies.
My understanding is that the longer it takes you to convert an advantage, the lower the arithmetic mean of your centipawn loss will be. Suppose, then, that you have a game totally won (so it would be harder to find a bad move than a good one), but then take your time about finishing the job. You may well have 0/0/0 and a low acpl, although it could be argued that mating faster is more accurate and for sure you won't win many tournaments without winning won positions quickly and efficiently. Furthermore, if you play main lines (especially e4e5) then you will play a lot of seriously booked-up people. You may find that you have more of a chance of winning if you play the less-studied stuff (including some "inaccuracies" to keep the initiative). Take the wisdom of the engine with a pinch of salt.
Here is an example https://adjva4.dpdns.org/HRUXxX4g
@Marlonc I'm not sure about that. Take this game for instance:
https://adjva4.dpdns.org/zrj8HlAA
Same number of moves as yours, and despite both games with some ups and downs I think it can be argued that I got my biggest advantage sooner than you in your game. I even found the most accurate mate in the end (although that could be my mistake according to your theory 😅). I have no blunders and yet my acpl almost double than yours.
Is it because my opponent blundered more than yours? I have no idea.
acpl in ZH is a great mystery to me, but not one that I'm concerned definitely 😬
I agree with everything that has been said in the forum. acpl is just not a good indicator of how well you did.
@pepellou you agree with everything that has been said but you're not sure about what I said? Hmm. Let me rephrase and break down what I said:
1) Computer analysis only tells part of the story, and so-called inaccuracies are fine if they are part of your plan. This, I think, you agree with.
2) if your opponent is hopelessly lost, prolonging the game with more moves will reduce your acpl (because it is a mean value, and if the set is bigger and you are in no danger then the figure will be lower).
Turning to your game, it looks to me like what cost you in terms of acpl (which, I repeat, I do not regard as a reliable indicator of move quality) was 16.@f6. SF prefers 16.Bg5+, or 16.c4, or 16. @d6+, or 16.@g6 or even 16.@e4.
@Marlonc yes, I agree with both (1) and (2), but again I'm not sure and it's still a mystery to me 😅
For example in my game that move 16 ok, I understand I did a +9 move instead of a +29 one, and if I follow the best SF line for that +29 move, after 4 moves the engine sees a mate in 6. So wasn't what I did prolonging the game in a completely winning position?
Also, what qualifies as blunder? If doing a +9 instead of +29 is not a blunder, why would it be the cause for a higher acpl than in a similar game with 1 blunder?
Those things are really mysterious to me, but that doesn't mean that I disagree with any of what you said. I think that generally speaking that theory should be correct.
I suppose blunder is when the global assessment of the position changes (winning / equal-ish / losing).
That explains why in your game +0.9 instead of +3.6 qualifies as blunder while in mine +9 instead of +29 doesn't, while it probably creates a higher acpl.
But the "prolonging the game" thing is really tricky, as you may increase the number of moves but you can also increase your acpl by doing +10 moves when you have a +30. To me the most mysterious part is how checkmates are included in the calculation. It's my understanding that if you make a mate in 10 instead of a mate in 4 your acpl lowers, but I'm not sure.
I imagine checkmates to have the same value or very similar, so if you do +9 instead of +29 you lose a lot of cp, but if you do a mate in 10 instead of mate in 4 you keep the same cp, resulting in a lower acp.
So maybe your theory is more accurate for when it comes to checkmating, and then you should ... "toy around" with your opponent for a lower acpl 😅
@pepellou I think you would have to ask the developers of SF for how exactly it assigns the labels inaccuracy, mistake and blunder. My game differs from yours though in that from move ten onwards my plus is in double figures. I *have* played games where I have missed mate in two but still had mate in four, but I don't have time to dig them out right now and I thought a recent game would be the most relevant anyway.
SF is a chess engine that has been taught zh...perhaps a later engine will furnish us with even more theory and squeeze even more of the fun out of the game :-/
This topic has been archived and can no longer be replied to.