Is it possible to loose a game without mistakes by checkmate ?

I think then the engine, and human annotators would call it an inaccuracy since it's much harder to see mate in 20.

If you play against Stockfish

then yes

#1
"Is it possible to loose a game without mistakes by checkmate ?" No.

"If you play a game without playing any mistakes, missed wins, or blunders, but you don't always play the BEST move, is it possible that you loose it ?" No as long as you make no mistakes, you cannot lose a game of chess. The notion of best move makes no sense. There are only good moves, mistakes (?) and blunders (??).

The answer is sort of. 

if you put a home computer using typical chess software against itself, it occasionally wins.  This of course means a mistake was made, but the mistake will trace back into the randomly selected opening -- some openings are simply weak, and the computers do not evaluate it just follows a list of moves.  Nothing will evaluate out as a 'mistake' when you do analysis in such a game -- the analysis will agree on every move that the best move the computer can find was played.  Still, I believe that a mistake is required to lose.   That gets into the definition of a mistake, though ... it can be very subtle, something no normal player and most software can't even point at and say "that was it". 

     This is really a question of semantics; what exactly is meant by "mistake". If you mean a poor move that clearly lets your opponent get a better game, then it is certainly possible to lose even if you never make such a poor move. If you consistently play very good moves it is still possible for your opponent to find a few even better moves that will eventually leave you in a position where you must lose in the long run no matter how well you play as long as the other players continues to play correctly.

definitive!

23.Qf7 is as mistake

Jonathan Schrantz has several videos on Stockfish where he beats it using an anti-computer strategy. In it, Schrantz plays several moves that the computer evaluates as blunders as he leads it into a trap. The computer, of course, thinks it plays all best moves and that Schrantz is blundering, but poor Stockfish ends up getting mated.

So yes, it happens.

again

only 3 types of positions

any move that goes from equal- lost is a mistake and to lose you need to be lost from equal so there must be a mistake!

Technically no, if you look at a game that you have lost and change a move so that it doesn't go into the losing sequence, the old move that loses the game is by definition considered a mistake/blunder. You will always be able to find a move to change so that it doesn't go into losing position, so you will always have mistakes in a game that you have lost.

If you can ever find a game that you lose no matter which move you change, you have essentially solved chess lol. 

Imagine, you keep playing a moves that loses 0.2 advantage. Theoretically speaking, you would lose 6 point advantage after 30 moves without having any mistakes(such a small lose of advantage doesn't count as mistakes rather as good move instead of best move)

#42
There is no such thing like 0.2 pawn. You lose a pawn, you lose the game. Those 0.2 are a computer approximation. A perfect computer would only have 3 evaluations: 0.00, +100 and -100. There would be no 0.2. In your 30 move sequence you can pinpoint 1 single move as the culprit that causes the loss: from 0.00 to -100.

exactly my point

I am guessing that the site's evaluations are being used.  Blunders, mistakes and inaccuracies have significant one-move evaluation changes in the opponent's favor.  Good moves (and excellent moves to a lesser extent) also have evaluation changes in the opponent's behavior.  I've seen moves flagged as "good" that I would consider "mistakes".

If one player is playing all "excellent" (not "best") moves and the other player is playing all "good" moves then the "excellent" player is likely to win big.

#45
The site evaluations "blunder", "mistake", "inaccuracy", "good move", "excellent move", "best move" objectively make no sense. Objectively there is only good, mistake (?), blunder (??).

Here is a famous example:
https://www.chessgames.com/perl/chessgame?gid=1072945

The losing mistake is 16...Be6?
Black had to overprotect central pawn e5 with 16...f6 and would be fine.

If you go watch the computer chess championships, there are plenty of games played where there are no mistakes/blunders in decisive games. If one computer plays best and the other computer plays a series of suboptimal moves that are still "good," one side WILL lose.

All best > All good, so there is just a slow but steady change in the eval with no single move being decisive.

Yes, interesting.

Thank you all for your answers ! I see that we don't all think the same, but now, I can have my own opinion about the subject !

Thanks

#47
No, also in computer chess championships: decisive game = mistake.
Show a decisive game and I will pinpoint the mistake.