Another perfect game of chess, but why only 2650 rating??

I played 14 moves of perfect, flawless, Super GM-level, Stockfish 16-approved chess (after which my opponent resigned), with 99.3% accuracy, including two single-exclam moves. So why was that worth only a 2650 rating? I have gotten 2700 many times and 2800 at least once. How are these numbers determined? It's literally impossible to play better than I did.

Probably not an opening/line a top engine would play im guessing

Sure they would, there's tons of engine games in the London System. 4. c4 is objectively the best move.

I ran the game in analysis and it said Knight to c3 was a mistake

Really, can you share a screenshot?

Please explain how? This game has no mistakes or inaccuracies. If you disagree, show them.

Fact. Although I do think it rates you higher if you're higher rated. But yea it's a gimmick

MikhailKasparov

How do you send screenshots

that is not screenshot tho, it's funny how half of you want to know how to send screenshot rather answering his concern lol

At that point, I think accuracy ratings lose a little bit of validity.

Had you any mistakes ?

im to lazy to check lol

100% accuracy = INFINITE ELO PERFORMANCE

Part of the "estimated Elo" calculation is based upon the Elo of the players involved (that's why it can't give an estimate if you upload a game without the Elo Tags for at least one player). Chess.com's estimation factors in your existing Elo when it gives the estimate for a single game.
If you want to check this out, open the PGN file, and just remove one of the Elo tags (in the header, where it will have something like [WhiteElo "1814"]
Just delete that line, then upload the game to Chess.com. Re-do the game review, and most likely your estimated Elo will go up (because the other player has the higher Elo). Remove the BlackElo line, leaving in the White's tag, will probably lower your Elo estimation.
Basically, the Elo calculation isn't only about the moves played in this game, but it takes into account the "history" of at least one player (both if the both Elo Tags are present) to some extent 
When they first introduced the Estimated Elo, they didn't do this, and I had games being rated well over 2000 (games I played with a friend over the board at the pub - so, guess the quality). Then it stopped giving estimates for our weekly games, so I got into some conversations about it, and someone from Chess.com explained how the new version of the estimation worked (due to problems like the above in that without those "anchors", it vastly over estimated Elo).

I raised this as a problem with chess.com when the feature was in Beta. The rating it gives you goes up or down not only with accuracy according to your current rating. Several different people ran the same Stockfish vs AlphaZero 2018 game and the ratings that it gave varied wildly. One lower rated user saw a rating around 1200 level even though they would have (if it were actually played by them) played above 3000 Elo level.

So don't treat that score too seriously because it was never intended to be.

For example, I imported a game and set both ratings to 1000. The review rating was 1600 for the winner and 1350 for the loser (accuracy was 90.2 and 81) . I re-set the ratings to 2000 for both players and the same game's ratings were 2450 and 2100 (accuracy stayed the same). So the in-game rating is based on the strength of the player's ratings.

Is 1500 rapid a good rating?

Yah, I find similar things. At one point I wondered if the difference in the estimated Elo might remain constant, so one could use it along the lines of "I outplayed/got outplayed" as if there were X differences in our Elo, but even that isn't the case as your example shows given the difference is either 250 or 350, and I've found the same when I tested it.

no