I still remember 2 decades ago how people were making jokes about Kamsky having learned by heart the complete informator as preparation for his worldchampionship-finale against Karpov of 1996. Today such complete study of theory is absolutely impossible. It can be pretty astonishing what the theoretical baggage some players have but it remains fragmented and nobody can protect themselves from the countless novelties discovered each day by engines. In a videoconference after the game Kramnik - Anand London 2014, Anand jokingly said but at the same time containing some truth that computers have a lot of time.
So nobody has a waterproof shield in the opening. Everybody is vulnerable for surprises. We don't have to exaggerate this risk for most players as shown by my article password. On the other hand when you play against the 1% best players then ignoring the risk can quickly backfire which was demonstrated in my article harikiri. It is impossible to tell in advance when exactly you better deviate or can rely solely on your experience. Finding out information about your opponent before the game will surely improve your chances to make the right decisions but in the end it remains an estimated guess so partly instinctive.
Sometimes there are very clear signs which hint there is danger. An opponent won't deviate from his standard repertoire to help you gaining a head-start in the opening. Almost always this indicates the opponent is well prepared and has some surprise in store to gain an advantage. Such advantage doesn't need to be a refutation of your favorite opening as often it is sufficient to show a new equivalent or even an old forgotten idea to win a lot of time on the clock.
If you see these signals then it is often wise to deviate too from your repertoire surely if the position can quickly become tactical. There are limits to a specific game-preparation. I mean even spending hours of preparation will only permit to analyze x number of lines as mentioned earlier in my article the list of strength. Most openings consist of many more options than the x-variations which can be checked in a preparation.
A first example which shows this strategy, is from my own recent interclub-practice. I played in the second round of the past Belgium interclub against the ambitious leader of Opwijk, Arno Bomans surprising me by following my game against Glen De Schampheleire. However before he can show his improvement and can use his engine-analysis, I already deviate myself from my game to restore the chances.
The second example which I want to discuss, does not show this side-effect mainly due to the big ratinggap between both players. Maybe the Russian grandmaster Vyacheslav Ikonnikov wanted me to pay for the article an arranged result in open gent or not as from the opening it was immediately clear that nothing was left up to chance. I was well prepared as I checked his 400 black-games with 1.e4 but Vyacheslav chose to play something new to avoid any of my preparation. Nonetheless neither did I wait for his preparation and countered his surprise fittingly with my own.
Sometimes it can become funny when both players try to surprise each other in such way that they suddenly play an important theoretical variation which they aren't familiar with. This happened a few months ago in the 2nd game of the worldchampionship-finale Muzychuk and Pogonina with a novelty only at move 18 but on which 17 minutes was spent so surely no preparation.
Previous game was played between players of 2500 elo but this is also happening on + 2800 level. Let us have a look and enjoy a recent game between worldchampion Carlsen and the French supergrandmaster Vachier Lagrave both known not to be afraid of original creative chess.