ever consider that the evaluator is pure garbage? huh, ever consider this?
1% accuracy??
Sort:
Feb 9, 2021
0
#3
The programmer in me can't help trying to understand the algorithm.
Feb 16, 2021
0
#4
It may be that the moves have an additional score. So maybe the bad moves were exceptionally bad, and the excellent move was not judged as hard to find.
I had 30+ blunders in one game with 0.9 accuracy
Feb 16, 2021
0
#6
I like how you frame the situation. You state that, if you ignore the book moves, 25% of moves is excellent. You could also say, if you ignore book moves, your friend managed to lose the game in just 4 moves. Playing a blunder, an inaccuracy and mistake consecutively is pretty close to playing the worst possible game. Even if you find 1 excellent move, you are basically dead.
But I'm not challenging your point. There must be plenty of games available that deserve a 1% rating more than this game. I think it's fair to assume the accuracy rating is very unreliable when there aren't many moves played. Which makes sense, since statistics always become unreliable with a small sample size.
Maybe you could provide the game? I would like to see the 1% for kicks, really rare stuff...
Feb 16, 2021
0
#8
Foolsmateinfinity wrote:
Maybe you could provide the game? I would like to see the 1% for kicks, really rare stuff...
It isn't my goal to embarrass anyone. I'll just list the moves. White is a friend of mine, about 1450, and black is the opponent who got 1% - he is about 1350.
1.e4 e5 2.Nf3 Nc6 3.Bb5 Nf6 4.O-O Nxe4 5.d4 a6 6.Ba4 b5 7.Bb3 f5 8.Nxe5 d6 9.Nxc6 Bb7 10.Nxd8 Rxd8 11.Re1 resigns
White got 17.2 accuracy(!?), and black got 1.0 accuracy. I don't understand either of those numbers.
The "Game Report" said: "That was an intense game". Very strange. I don't think it was intense at all, just a standard opening followed by a couple of blunders on moves 8 and 9.
Feb 17, 2021
0
#9
Thanks for posting. It looks natural enough. He can't be the only one to have played this sequence.
On depth 30 accuracy rating are 51.7 vs 1.1%, btw. My numbers below are based on d-30, somehow the linked analysis is at a lower depth, so values are slightly off.
If you look at the best move diff, you see book moves are rated at +- 0. So move 7 is +3 diff, equivalent of losing a minor piece (which is what happend. The second non book move was +15 (equivalent of losing Q+R), third move played is +4. Had the opponent played better we're looking at losing 20 points in 3 moves, that's impressive. Even after excellent Rd8 on move 10, I doubt you can do much worse. It's actually not easy to find moves that are worse lol. For example sac Queen on move 8 (Qg5), that's a better move than d6. So I guess the reasoning is of all possible sequences of 4 moves, this one is as far from perfect play as you can get.
Thanks again for posting, I enjoyed replaying this game
A friend of mine just finished a game and the site evaluated his opponent's accuracy as "1.0%".
Sure, the opponent played pretty terrible and lost in 11 moves, but the rundown was:
1 excellent
6 book
1 inaccuracy
1 mistake
1 blunder
I'm curious how that becomes 1% - surely one could play much worse. Only 3 of his 10 moves were less than excellent. Even if you throw out the book moves, 25% of the moves were "excellent".