Can we have more bot levels? Stockfish level 3 bot with 90% accuracy and 0 mistakes and blunders.

Don't you think this is a little bit too much? Here is the game.

Notice how it perfectly cornered and finished me starting from around move 16.
Can you make more levels for this bot and show his skill based on performance of players against it?

@SlayerSv
Lichess also has other bots. See this page: lichess.org/player/bots
This page can be reached via the menu: "COMMUNITY" > "Players", then in the left-hand column "Online bots".

The accuracy is quite misleading here, the fact that you blundered a piece in move 5 affected the numbers. (Not that 90% accuracy would be a sign of an extraordinary performance in general.)

@mkubecek said in #3:
> The accuracy is quite misleading here, the fact that you blundered a piece in move 5 affected the numbers. (Not that 90% accuracy would be a sign of an extraordinary performance in general.)
Have you ever had 90% accuracy?
GM's have on average 90% accuracy.
You say I blundered a piece. Ofc I did, I'm expected do to that, more than once, my question is, why that thing did not blunder a single time and had a game close to perfect.

@SlayerSv said in #4:
> Have you ever had 90% accuracy?
Few times, I'm sure. My highest probably is lichess.org/WIgqOQrH/ with 99%. But, to be honest, it was quite a surprise when I saw those numbers because I didn't think I played particularly well in that game (and I still think I had games with much lower accuracy that I have more reasons to be proud of).

This example shows an example of a common case of high accuracy games: mostly uneventful, trading everything that moves (OK, I'm exaggerating a bit), soon it gets to an endgame which is either completely equal or completely winning for one side. (In this case, part equal, part clearly winning.) There were games where I had 89% and I believe I had to show much more than here.

> GM's have on average 90% accuracy.
...when they play against opponents of similar strength and the game is active with both sides fighting, with a lot of threats and opportunities to make mistakes. If they play a dull "grandmaster draw", they get much higher and likely close to 100%. But does it mean they play better chess? No way.

> why that thing did not blunder a single time
That's what I tried to tell you: it's related to how the classification works. The engine translates position evaluation to "winning chances" on the scale 0-1 (or 0-100%) and a "blunder" means a move that decreases "winning chances" by 0.3 (30) or more. Therefore if you are, say, a queen and rook up, you can blunder a rook and your winning chances are still so high that such move is not marked as a blunder (and probably not even a mistake). In a similar fashion, once you are losing so badly that your winning chances are below 30% (which is still far from hopeless among weaker players), you can play essentially anything (like overlooking a checkmate in one) and it still won't be marked as a "blunder" because your winning chances cannot drop enough.

TL;DR: the "accuracy" numbers are a hint at best, don't try to give them more weight than they deserve. And please don't believe that high accuracy is a sign of a "perfect game".