What accuracy % do you consider a “good” game?

@AjaxisFrosty

How do find your % accuracy?

best accuracy i like is 100%

85%

I think it probably holds up to the highest levels of human play. Here is a collection of KNN v KP endgames. One of the more impressive wins is this by GM D Gurevich (highest rating 2586).

After move 111 he has mate in 5 but instead embarks on a mate in 11, which is close to what you describe.

But it won't hold up to Stockfish for example. A mate in 4 or 5 will be just that.

That was obviously just an oversight on Gurevich's part. Here's another game snippet. Rob van de Plas v Philip Cody 1996 (both at least 2200+ at some point).

Van de Plas follows a plan of capturing the pawn then mating with a bishop and knight. Had Cody not resigned on move 87 he could have forced mate by move 96.

Both the plan and its execution are not totally accurate, but both are straightforward and lead safely to a win.

The variation shows an accurate continuation after move 58 from the Nalimov tablebase mating on move 80.

Most players would probably take a similar route to Van de Plas, because they would probably have practised the KBN v K endgame but not the KBN v KP endgame so there would be less chance of screwing up. 

I'm happy with anything above 90%. 80-90% is ok, and the rest are below my expectations.

Where are you all getting these % figures from?

CAPS.

not a good metric at all. Plenty of games with straightforward openings can have super low centi-pawn numbers simply because the lack of meaningful complications in the position. (e.g think of picteresque caro-kahn games where not much happens and fizzle into a draw)

Some of the best games have centipawn loss in the 40's where both sides did lots of (from a comp's perspective) small mistakes were done but the better player's superior understanding or grit wins out.

But I only get CAPS2 and it's not a %. Is CAPS a % and how do you display it?

Haven't worked out how to get an ACL either - where do those come from?

THERE IS NO WAY I JUST GOT A 97% GAME AS A 350 https://www.chess.com/analysis/game/live/66330185981 

edit: it wouldve been a 99.4 if i hadnt done that "inaccuracy" which was actually a trap to take a rook that 50% of the people at my elo fall for

@YidingL1 100% accuracy is not possible unless you cheat, but true, everyone wants 100% accuracy. However, if you played at 100% accuracy, you could literally draw Stockfish 1000 in a game.

Hey blueemu. Do you think a chess legend encountering a chess god is a good game?

yea, you're so right on this. I recently played a game in the Caro-kann where I got like 96 accuracy even though I didn't play particularly well, and then I had a very tactical and complicated game in the Benoni where I got like 60% even though I think I played better

100 accuracy is easy if you play the fritz or the fried liver.

For me, my average I have seen in my recent games has been around 70-80 accuracy when I play well. Yet drops here and there depending on if my chess is any good that day.

@ThorRagnaroc you only got 100% accuracy because you’re opponent blundered. It’s easy to get 100% accuracy if your opponent blunders every move, and you won in 6 moves. Getting 100% accuracy for 6 moves is relatively easy

When I said 100% accuracy is not possible unless you cheat, I was talking about an average game with around 40 moves.

When I did my best. That's a good game.

@powerpunch107 the only reason you got 97% accuracy is because your opponent played horribly. If your opponent plays horrible you can spot lots of opportunities to win, and you won’t make any mistakes.