CAPS is super confusing sometimes...

As they say, it is better to be lucky than good.

You could help us out by posting the game.

Many computer analyses disproportionately punish you when you have a very winning position but don't take the absolutely fastest way to win it.  Sometimes they ding you for opting for an easy mate in 5 rather than a complicated mate in 3.  Other times they punish you for simplifying into an easily winning endgame rather than pursuing a complicated but 'better' middlegame.

I like the idea of CAPS and other similar metrics, but they are clearly flawed as currently constructed, and I'm not sure if any are suitable (as in, even remotely helpful) to those under 1800 or so.  Also, the CAPS or whatever of a single game is nearly meaningless; more useful would be to see a rolling average of your CAPS over a 20 game window, say, because that might be informative.

I've found CAPS is pretty useless. For instance, I've had CAPS close to 100, when the game is short and I win easily after some forcing combination. Or when the game is quiet and very long, and I make a lot of natural moves without any obvious inaccuracy.

Average difference is also misleading, especially when you are absolutely winning but don't take the shortest path to victory, as @SmithyQ was saying.

Even CPU evaluation is misleading, as sometimes a position is objectively better for some player if he plays perfect defense, but practically much easier to be played by the other player. Also the CPU sometimes suggests very weird moves, which a human would never play based on positional criteria and pattern recognition. but the CPU plays because it is sable to calculate a couple of dozens moves ahead.

My suggestion is: analyse without a CPU, try to evaluate positions based on your own understanding, force yourself to explore variations without the engine's help, and only use it to check tactics. 

Daedalus9000 wrote:

I just found this particular computer analysis befuddling... selfishly perhaps because i felt pretty good about my game and CAPS seemed to imply I shouldn’t

It happens. Learn to don't give much credit to the computer. If you felt confident playing the game and managed to outplay your opponent (not if you won luckily because of some tactic) then you should be satisfied. Improvement takes years, and it doesn't make sense being worried all the time about the computer criticising our choices.

CAPS is a number without a definition, or even a specific objective. This is a shame, because I am sure its definition or its objective could be explained to make it more than a number that you need to guess how to interpret.

For example, it would be possible to define a statistic with the objective of ranking a player among all players on chess.com, on a scale from 0% to 100% (this is not what CAPS does). The rating from many games would be a better way of doing this (after comparing with the known distribution of ratings). A single game can be used either to give a crude one game rating (opponent's rating, +/- 400 depending on result) but it is certain that a better estimate of the rating of a player could be achieved by looking at the individual moves.

The reason I don't believe CAPS does this is it clearly would not do a good job of it. It is not a good inference from one good game that a player is very highly rated (still less from those games with very high CAPS for both players from long dull draws), nor the opposite for a game with a big blunder, so its numbers would not be what they are But what it is actually trying to do is a mystery.