Plain-English scoring and workflow notes for coaches and members.
Range & React is designed for live-poker coaching. The point is not to memorize one perfect answer. The point is to build a repeatable decision process that respects villain type, scenario, action history, and range interaction.
Range Score measures the quality of the range left after a prune. Most of the score comes from keeping hands the model thinks are likely after villain's action. The rest comes from removing hands the model thinks are unlikely.
Action Score measures how close the member's bucket reaction reads were to the model's response probabilities. A selected response gets more credit when it is close to the bucket's most likely response.
Overall Score is the simple average of Range Score and Action Score. Keeping them separate is important because a player can range villains well but still miss how a specific villain type reacts.
Villain types are one of the main signals in the product. The same board and line can mean different things from a nit, calling station, loose reg, or maniac. Coaches should use villain-specific misses to assign better reps.
The Coach page rolls member work into pool-wide analytics, cohort comparison, assignments, flagged-hand review, and reporting. It is built to answer what a coach should do next.
Train mode is designed for desktop and laptop screens only. The range grids, table state, and postflop response matrix need enough horizontal space to be usable.
Coaches and admins can download one row per member with current Range Score, Action Score, Overall Score, reps done, and assignment counts. That makes external reporting easy for a coaching business.