In general make the play with the highest reward to you, which puts you at the lowest risk. If you don't need to predict the switch in order to win the battle, then don't. In this case predicting the switch has not enough reward to be worth the risk of your opponent staying in. If however, you not predicting the switch will cost you the battle or make it far harder to win, you need to asses the risks of him staying against the risks those of him switching out, and then make the most beneficial play to you. Making stupid plays can confuse your opponent, but then they are just that, stupid plays. Just because your opponent won't see it coming doesn't mean they no longer have the advantage, if your opponent knows they have the advantage they are unlikely to make risky plays as they aren't necessary. In these situations predicting your opponents safest playing and reacting to it, is what you need to be doing. Your opponent will be working with the same logic as you, what is the most beneficial play to me, in order to win this game, you should generally work off your opponent's best play. Predicting predictions based off of other predictions is extremely high risk and almost never worth it. Remember you don't know what four out of the six they have brought or what moves they carry, until your opponent actually reveals what they have, all you can make is assumptions, and assume wrong and you can disadvantage yourself. Generally at the start of the battle making safe plays is the best thing to do as you can, scout out what the opponent has. As the game progresses you can predict more accurately, as you know what the opponent has and how they play.