The CAPS system isn't very accurate. My last two games were 98.56 (about 2550 ELO) and 89.85 (about 2050 ELO). The better-algorithms in Lucas Chess listed them as 86.60% and 77.45%. I haven't seen a Lucas Analysis vs ELO chart, but the numbers represent something like 2400 and 1800. The first was a 14 move game where my opponent virtually invited the right moves. The second, 1800, is a little more accurate than CAPs.
10 CAPS points less even though winning
Sort:
Apr 2, 2017
0
#3
The main reason the CAPs is off is that you had an extended endgame and I've found if you play it safe when you have a winning endgame, Stockfish lists moves as blunders and inaccuracies if they don't lead to the fastest mate. For example, every endgame book would say to make the move I made, Kg7, below: keeping the opposition and guaranteeing my Pawn will Queen. But chess.com's Stockfish 8 analysis calls it an inaccuracy and so does Stockfish 8 set to 20-ply on Lucas Chess as shown below and Chess King's Houdini 2.
They all recommend f6, which would lead to Black's K being trapped stalemate like at f8 and you have to move another Passed Pawn - being careful not to block all his Pawns to win.
First of all, because this game had an end game, both CAPS scores are inflated. It's not hard to play Stockfish level chess when you only have one legal move, advancing a pawn forward.
Second off, you blundered significantly more times than him, which hurts your score a good deal.
Take it for what it's worth, which is very little. If you're crushing someone, and he has almost no pieces left, but you keep missing some Mate in 8 combo, your CAPS will keep going down, meanwhile his King Dance will keep him going up. Doesn't mean you're not crushing him.