I once played a game of which the missed combination has been graved in my memory forever as it is extremely weird. I need to correct myself as it was not a combination but rather a wrong continuation of my opponent which is countered by a very non-standard refutation. Which I didn't see and I have never detected in any book about combinations or any game. The "combination" or rather the refutation is so "unique" that I never ever seen it before: letting a piece to be captured with check and not take back but stop the check by putting another piece in between, as there exists a long-term threat which is stronger than the temporarily loss of the piece.
I didn't notice it in the game. The only game with this "line" in Chessbase is Piscopo (2364) - Zakharchenko (2197) played in 2012 - and so I am in good company: also the Italian international master Piscopo didn't find the move. Another move which can be categorized as invisible see part 1 and part 2. So unfortunately I played the automatic 10.bxc3 which let slip the white advantage away. Below you can replay the remainder of the game.
At the blog of Quality Chess there was recently a discussion about automatic moves. Often mistakes are made because players don't consider sufficiently alternatives. By spending more time you can find those moves was the logical remedy proposed by the author. However some readers didn't agree. Automatic moves leave extra time for other moments in the game when complex decisions need to be done. If you start to question each move so also the automatic ones then you risk time-trouble creating much bigger problems. It will be a disaster for the playing-strength to find that one unique move in one particular game in exchange for many blunders due to lack of time in dozens of other games. Still for my most beautiful move, I would've liked to make an exception.