Fixing an opponent's strategy at a specified decision node in a solver and re-solving for the best response. Used to convert a measured population leak (e.g. 68% fold to turn probes versus an equilibrium 55%) into a quantitative EV delta per occurrence. The lock must be a frequency at a node, not a vague read; the output is an exploitative strategy, not a new equilibrium. After locking, re-ask the six interrogation questions — range advantage, nut advantage, and mixing hands all shift.