It's surprisingly easy in practice. The reason I'm still on a 4CPU from 10 years ago is because, as it turns out, it's also surprisingly easy to predict moves that need high depth to be found.
Not only that, it's also surprisingly easy to predict people's opening moves. At one point I built a big database with all the corr chess games I could find (big mistake! it's full of garbage moves by people not using engines, but I didn't know better), then I had this opponent that played the most played move of the db three moves in a row. So I extrapolated it and managed to predict his next 20 moves, and managed to get the most advantageous position I could once out of book.
In the end I predicted all his moves and when I couldn't, they were worse moves than what I predicted. Unfortunately, I had played a drawish opening (at this point I was scared of higher rated people, and played more safely than now) so not even knowing what he was going to play was enough to win. The draw wideness of chess is indeed big.
But this is a common thing, either you predict their move, or they play something worse than what you predicted. Last week I was surprised by 2 moves, by someone that seemed to want a draw as white
Great spirits have always encountered violent opposition from mediocre minds.