Caps system

Some times on game analysis the caps score shows, but most often it does not, anyone knows the reason, is it only for premium members, 

The CAPs system is not good. Would not take it too seriously.

I agree. In my opinion, the CAPS is inconsistent and not very reliable at this point. I analyzed one of my games (4 times). Each time the CAPS numbers was different ranging from 60 to 80

I never saw it until I began a paid membership. Is there a limit, etc. on how many per day otherwise?

I would add that it's not anywhere as good or comprehensive as the freebie Lucas Chess game indexes using the same Stockfish 8 chess.com does:

Here's my chess.com indexes for my last game, which i won in 14 mostly good moves - my opponent made a lot of weak ones:

Since a CAPS of 96.2 supposedly represents an ELO 2400 OTB player, this looks fantastic.

But look at the Lucas Chess Indexes for the same game, where GM's unusually score in the upper 80%'s or low 90%'s.  Mine's ok, but certainly no 2400 ELO equivalent although the last 7 moves were like Tactics Trainer problems: I could smell blood each time and with 3 days/move could take my time looking for it, so that helped find good moves (I have a 2116 USCF Correspondence Rating from the pre-computer '70's and think I playing close to that now).  But clearly the Lucas Chess numbers tell the story a litle better than CAPS:

CAPS is no good. I have Elo 1360 OTB and my CAPS in last 4 games of SLow Chess League (90+30) are; 96,90 92,39 78,85 94,12 and my opponent have about the same CAPS- 

My understanding is you have to consider CAPS over a number of different games - white, black, open, closed, positional, tactical. If you're good at tactics and you have a very tactical game with a lot of obvious forcing moves then yes you're likely to get a very high CAPS score for that game, but you have to average it over a number of games.

I think CAPS only shows up on games where you choose Maximum analysis - which would explain why you have to be a member. I'm also unsure about how reliable it is; my CAPS ratings are much higher than my OTB.

The CAPS system is the best measure yet of the accuracy of a player's decisions during a game.  Unfortunately the Chess.com article that compared CAPS scores to ELO ratings is poorly done.  Yesterday I used CAPS to evaluate three OTB games I played in a USCF rated tournament.   The results were 67%, 97% and 81%.  I won the first two games vs. 1900 players (who had CAPS scores of 36% and 66% resepectively) and luckily drew the last game vs. an under-rated 1400 player whose CAPS evaluation for the game was 97%.  He made only one blunder by overlooking a difficult winning variation.  In that game he was clearly the better player despite a 600-point difference in our OTB ratings.  We all have good days and bad days, and good games and bad games.  In certain closed positions where the themes are well-known to me I can consistently find the best moves.  In risky and unfamiliar positions I find the best moves much less often, and have to work hard not to make large blunders.  

There is clearly a correlation between average (or median?) CAPS score over many games and ELO OTB rating.    However, the first article published by Chess.com offers a table that is seriously incorrect.  I hope that someone will undertake the serious statistical work required to find an conversion formula.