GAME REIVEW .VS. COMPUTER ANALYSIS!!!!!!!!!!!!!!!!!!!!!!

Ok

Chess.com Game Review accuracy is not a direct engine match algorithm.

https://support.chess.com/en/articles/8708970-how-is-accuracy-in-analysis-determined

Chess.com accuracy is in many cases overrated. It is not easy to get under 70 accuracy if you are not a beginner (if you play reasonably long game and that game isn't sharp - if it is, then both players will get lower accuracy probably).

I had a game where I played unrated game vs under 1 000 rated player who proceeded to blunder material 3 times in around 10 moves (after which he resigned), and he still got almost 70 accuracy which is realistically nonsense.

The accuracy has been changed in 2021. Before that, I had bad games with accuracy around 20 or less. When I view one such game after a few years, the accuracy was around 50-60.

Unless it is a sharp game, more often than not I get over 80 accuracy nowadays, which is in my view unrealistic.

If lichess shows even higher numbers (and I've seen many such examples where lichess shows higher numbers accuracy wise), I wouldn't take it seriously.

It is better to view individual moves and see where you went wrong and what you did right, than to judge your game based on some descriptive number which is pretty fishy.

Both measures are inaccurate.

Black already had a good game after your poor 4th move 4. Nxd5.

why is Nxd5 a poor move? it seems to me to be an ok move.

It doesn't lose the game or anything, but it is not good.

Why? Because you centralize his queen. The rule of the thimb is that centralized queen is not that great this early if there is Nc3 as an option for white. But as there is no knight to challenge it, it is not the best idea to initiate an exchange of knights here. Basically here he gets his queen from the staring square to the middle of the board for free.

There is a similar situation in Scotch opening, there it is not a great move either.

If you just view the position, after move 4. Nobody has a single minor piece out, there is only a black queen, even pawns are on the starting squares. As white plays first, it shouldn't happen that black has one piece out and white has none. It is equal because white is on the move, but black has a fine position.

One will not lose the game because of it of course, but it is nice to notice white the engine says that white doesn't have the opening advantage anymore.

oh yea , did not think of that. Good to know for future games.

Yep

Hi, it depends what you mean by accuracy. If you mean "playing good moves" (moves that do not lose) you probably had quite a high accuracy. But making moves that make the best of your position is another thing. On move two you played Nc3, which is rather weak, since the main idea is to play d4 and develop. You can play c4 after d4 if you like, hitting the knight. Then on move 4, you used your developed piece to take his piece, instead of developing and so his Q was developed to not a bad square for it, since you no longer have a knight to attack it with, so instead of white getting an advantage out of the opening, the game was completely level after four moves, because two of your moves were not accurate.

So how does this reflect on the analysis engines that gave you that percentage accuracy? It means that they are definitely not good teaching tools. It's all very well cheering up a 600 by being positive but you're stronger than that and obviosly you won't learn like that.

The only way to learn with the chess.com analysis tool is to look at the game in depth by seeing how you could have improved on your play and how your opponent could have improved on his. And although you shouldn't trust the programming of the analysis tool too blindly, Game Review seems much worse. These things are good for weaker players only. Like lets say 300 to 1400 ratings here.