why is lichess vs chess.com accuracy in analysis different?

this game that i played:

98% accuracy on lichess, 74 on chess.com

why?

Because they define accuracy differently. See https://lichess.org/page/accuracy for the Lichess definition.

Say you're up two queens. Even if you blunder one queen, you'll still be completely winning - your expected win probability might come down from 99% to 98%, so as far as Lichess is concerned the move isn't even close to being an inaccuracy.

The other site's accuracy metric is more like a school grade. We don't know how it really works, but that's their description.

You were completely winning from move 7. You stayed completely winning for the rest of the game, eventually converting your advantage to a win. You missed 10...Nb4 which was even more completely winning (you might have won 20-30 moves sooner with it). You would have lost a piece with 18...Bxe4 (but still remained completely winning) but your opponent missed it.

@PurpleInferno said in #2:

Because they define accuracy differently. See lichess.org/page/accuracy for the Lichess definition.

Say you're up two queens. Even if you blunder one queen, you'll still be completely winning - your expected win probability might come down from 99% to 98%, so as far as Lichess is concerned the move isn't even close to being an inaccuracy.

The other site's accuracy metric is more like a school grade. We don't know how it really works, but that's their description.

You were completely winning from move 7. You stayed completely winning for the rest of the game, eventually converting your advantage to a win. You missed 10...Nb4 which was even more completely winning (you might have won 20-30 moves sooner with it). You would have lost a piece with 18...Bxe4 (but still remained completely winning) but your opponent missed it.

On chess.com, the best move is calculated based in how closely you played to the top engine move it says, so ok, (but which one is correct, did i play at 98 or 74?

Yeah, that's the principle with both sites, it's just the implementation on Lichess is based on estimated win probability using a public formula that's been calibrated with actual data from human games while the other site uses unknown proprietary mumbo-jumbo, then fudges the numbers.

Bottom line, you played about as perfectly as you needed to. But you just needed to play 7...Qa5+ and then not make any major mistakes.

If you hadn't been up a piece, 16...e6 and then not making any space for your light-squared bishop to retreat to would have been very bad. But you were up a piece, so you were briefly less winning until your opponent missed his opportunity.
If your dark-squared bishop and kingside knight had been missing on move 10, missing 10...Nb4! would have been disastrous, turning an excellent position into a losing one. But you were up a piece instead of down a piece, so you just missed a win-more move.

@Planet_CHESS465 said in #3:

but which one is correct, did i play at 98 or 74?
Both are meaningless numbers, so both are equally correct. Just like any other number would be.