Phil understands very little about the very games he plays, or game theory in general. This kind of strategy is straight from game theory. You want your opponent in a position where they have no correct choice. If you're either right or you're wrong 50% of the time, it literally doesn't matter what you do.
In non-game theory reasons strategy like this works, despite the meaninglessness of the decision you put your opponent to, your opponent is in the position of actually having to think about it anyway. When it turns out meaningless and there's no clearly good choice, it puts stress on them, which in turn increases the mistakes they make.
There are obviously more complicated situations that don't involve straight 50/50 odds, like in poker, where the game theoretical optimal choice is to bluff with a frequency that given the size of the bluff and the pot, the frequency correlates with the odds your opponent is calling that either option has equal EV. Again, you're giving them no "right" decision.
Obviously, if your opponent is less than skilled and has recognizable patterns (like Phil himself), you exploit those often by doing things that are "wrong" from an optimal strategy perspective (look up Nash equilibrium for what this means), but right from the perspective of beating that particular opponent.
However, if you don't know what the "right" move is, you don't know what the "wrong" move is or when the right time to make the wrong move is.
Phil lacks a basic comprehension of games. Obviously, you don't have to be a mathematical theorist specializing in games to have an intuitive grasp of this. A lot of 12 year olds who are good at games grasp this kind of stuff intuitively and have never even heard of game theory. But Phil lacks this on a fundamental level.
A bit :autism: I suppose, but he's just really bad at games. It's not that he's necessarily clumsy or stupid, but he doesn't grasp these ideas and has no interest in learning from his mistakes.