Decision Threshold Profit Optimizer

Most machine-learning business decisions boil down to one number: where do you cut a continuous risk score into "act" and "don't act"? This simulator renders two overlapping 3D score-distribution histograms — a positive class (churners, fraud, defaulters) and a negative class — and a movable decision plane between them. Set the payoff for a caught positive against the costs of a false alarm and a missed case, and a live gold curve traces expected profit across every possible threshold, so you can see exactly why the profit-maximizing cutoff almost never sits at the textbook 0.5.