5 book moves = 2% accuracy. 6 book moves = 99.7% accuracy.

Just played a game, my opponent resigned on move 6. So on the analysis I had done 6 book moves and was given a rating of 99.7% He(she) had done 5 book moves and was given a 2.0% accuracy. We had both only done book moves what justification for this difference?

1 e5 e5

2 Nf3 Nc6

3 Bc3 Nf6

4 Ng5 d5

5 exd5 Nxd5

6 Nf7

For reference, after 5 exd5, it's 100% vs. 2%. That's the earliest one can do analysis it seems. Anyway, this is interesting. Seems not all book moves are made equal. I am not a chess expert but the nominal continuation is a worse position for black, down on points, king's moved, worse position etc. Generally you want to avoid fried liver esque attacks in the first place. Link to the game: https://www.chess.com/analysis/game/live/18979003821

I've won games with very low accuracy and I've lost games with an accuracy in the 90's percentages. It's a flawed system in my opinion.

Same

Idk how it's 2% should ask some moderator.

Accuracy shows what the computer thinks is the best move at the set depth. That's all. Shouldn't be taken to mean that highest accuracy = winning, unless both players are playing the best moves, and that may end in a draw/narrow win anyway.

You should check the in-depth lines for each move and see which move is the one that decreases the accuracy. Then, by checking the lines, you can see if your opponent's worse moves lead to an attack or something like that.

do resigning cause it?

Maybe it has something to do with the fact that white played 6 book moves whilst black only played 5 book moves. 

The accuracy isn't based on the result of the game, it's based on the difference in the engine's best move and your move and book moves can be the best move or a blunder black 5th move was a mistake, so it is 2.

5...Nxd5? Is bad.  5...Na5! Gives Black full equality!

 Ok on that move but the rest were fine down to 2% because one in five moves is bad?

This is exactly why accuracy is flawed. Don’t pay attention to it.