A pure bluff of size B into pot P risks B to win P. It breaks even if the opponent folds often enough. Turn it around and you get the defender's rule: fold more than that and any two cards can bluff you profitably.

• Position: acting last is a structural edge, because you see what others do first. Play more hands on the button, fewer under the gun.
• Ranges, not hands: put the opponent on every hand they would play this way, weighted by likelihood. This is Bayes again: each action is evidence with a likelihood ratio.
• Balance: bet your strong hands and some bluffs with the same size, in a ratio that makes the opponent indifferent. A pot-size river bet wants about 1 bluff for every 2 value bets.
• Exploit: once you know a player folds too much, bluff them more. Game theory optimal (GTO) play is the unexploitable baseline; profit comes from deviating against mistakes.
• Variance and tilt: a winning player can lose for weeks. Track results in big blinds per 100 hands (bb/100) over tens of thousands of hands, and never chase losses.