Building evidence-based exploits
An exploit is a conditional strategy: if an opponent deviates in a repeatable way, then a counter-adjustment can gain EV. Node locking formalizes this by fixing an opponent strategy at a decision point and recomputing a response. It does not make a read true, and copying the response while the assumption is wrong can create a larger leak than the one you hoped to exploit.
The read-based notes attached to your reviewed hands follow this same discipline. They describe an adjustment you would make once a read is confirmed, and they never change the grade printed beside them: a postflop verdict still measures your action against a neutral response, even where the villain's estimated range was conditioned on their type. Treat a note as a hypothesis to test, in the confidence ladder's terms under Going deeper — not as a correction to the grade.
- State the observation.Use a specific line and sample, not a personality label.
- Translate it into a range error.Identify which value, bluff, call, or fold region is too wide or narrow.
- Choose the smallest response.Adjust the hands nearest the threshold before rebuilding everything.
- Set a falsifier.Decide what new evidence would make you stop or reverse the adjustment.
Worked example: from read to testable response
Across 18 river betting opportunities in the same bet–bet formation, a regular bets seven times. Six reached showdown: five strong value hands and one missed draw. Their eleven checks include every observed medium-strength showdown hand. Hypothesis: their betting range is polarized and may be under-bluffed. First response: fold only the weakest bluff-catchers that were close at baseline; do not fold your whole bluff-catching region. Falsifier: several shown thin-value bets or enough credible bluffs in the same formation. A solver node lock could measure the response to the assumed under-bluff, but its result remains conditional on the input and the modeled tree.