Accuracy isn't a fair metric for aggressive/tactical players?

Hey guys! I've recently been thoroughly enjoying playing some uber-agressive, sacrifice full chess. It's serving me pretty well, but I do find that the engine in post-game review tends to be largely unimpressed in terms of accuracy. While watching the PCL, I heard some of the commentators discussing this very issue, and some were calling for an AI analysis that instead judges who is winning on the DIFFICULTY of playing a position. That said, here are a few recent aggressive games I've played, and I'll list the accuracies and you'll see what I mean:

The engine gave this one only 75% accuracy!

This game, although I sacked 7 points of material, my opponent ended up having to accept the repetition or hang M2! Accuracy 82%, so much better.

I just don't think the accuracy represents the quality of the games, and I find this happens a lot in complex positions, or with aggressive games where the opponent's defensive resources do exist, but are extremely hard to find.

What do you think? Should we be developing an AI to judge positions on complexity, not accuracy?

Accuracy is all based on how good individual moves are. So hence the accuracy will get really high for positional players. Chess.com compromises this by making brilliant moves, which tactical/aggressive players will get very often

That's true I suppose, the little dopamine rush you get from seeing that double blue exclamation point is normally worth it... I got two of them in this OTB game (granted against a much lower rated opponent

yes, first analyse the game on your own (if you really want to, have an eval bar). Then check engine suggestion for 2nd analysis

AC---accuracy, position evaluation and quality of game are three different topics. Accuracy measures how your moves compare to the engine. In Game #1, 12/26 moves were "best, great or excellent"; Game #2, 19/27. Misses & mistakes were close (5 v 6). Not sure how it's calculated, but can see why the difference. At your level, Chess.com projects a range expectation of 80% to 90% accuracy. With this in mind, it is not a stretch to see accuracy have minimal impact on positional evaluation. The differences between good to excellent moves will be decimals, not enough to significantly move the evaluation bar.

I don't think AI is necessary for evaluating positions as you are trading one algorithm for another. An engine provides an assessment in numeric form as to which player has an advantage or an even position. What would AI do differently?

I agree that accuracy does not directly reflect the quality of a game. One would assume that higher accuracy yields quality games, but not always true. Defining "quality" is subjective and personal.

That said, solid players do sometimes just neutralise me after inaccuracies…

lol, one example = me (you don't even need to play innaccuracies sometimes. It just happens when you play me unless you are a hyper aggressive player or just generally stronger than me XD)

Really ringing me out to dry showing that game Adam

I highly disagree. You know, if you played a game really welll but lost, you should still be very happy with the game. It shows you perhaps have improved.

Extremely low accuracy wins aren’t all that satisfying, as it’s normally a question of who blundered last rather than anyone playing particularly well. That said, a few misses are probably okay, but that’s not reflected amazingly by game accuracy…

These accuracy ratings tell you how well your moves would have done against the best possible responses. Humans can't match engines for accurate responses so your aggressive play would not have defeated the machine, but neither would playing more conservative "accurate" moves; the engine will win in the end. The same holds true for games you might play vs someone 500 elo points higher-rated than you; your aggressive attacks will undoubtedly be refuted.

What really matters is whether your play succeeds against your human opponents--if you are having success and enjoying your games, that's all that matters.

Not sure why accuracy matters, in the sense in which the OP is discussing it. It seems like accuracy is important to the OP, however sac's and playing complex positions are going to result in lower accuracy scores. In this sense, chess resembles golf, playing difficult courses results in higher scores . Even at top levels, very complex positions and sacrifices are often not "accurate". This thread sounds like wanting to have one's cake, and eat it, too. Checkmate leaves no weakness in it's wake.

That's true u need something better

@OP: True. But at the end of the day no one really cares about accuracy. It's enough to win the game.

Thats a good pic,Jen.

Thanks everyone for your messages on this! I’m aware that accuracy doesn’t matter on a game by game basis, but knowing that attacking ideas were backed by engines helps improve my chess confidence (and therefore propensity to back aggressive sacrifices etc)