Every lesson so far has graded decisions — was this call profitable given the price and the ranges? That's the only thing poker lets you control, and it's worth being precise about why. A single hand has one outcome; a decision has a distribution of outcomes. Expected value is the average of that distribution, and the whole skill of the game is choosing the distribution — not engineering any particular result.

The trap has a name: result-oriented thinking — judging a choice by how it happened to turn out. It feels like learning but it's noise-fitting: the outcome of one hand is dominated by luck even when the decision was terrible (calling with 5% equity and winning) or excellent (getting it in with 85% and losing). Over thousands of hands the edge takes over and outcomes track decisions; over dozens, they barely correlate at all.

Read that table slowly. After twenty flips of a genuinely excellent bet, you're still behind about 5.8% of the time — one session in seventeen or so — purely from ordering, not from doing anything wrong. Real poker edges are far thinner than +0.5 per flip, so real downswings stretch correspondingly longer (the Variance & bankroll page builds on this arithmetic). The practical rules fall straight out: judge sessions by the quality of decisions made, review results only across large samples, and never "fix" a winning play because it lost once.

This app's grading follows the same philosophy by construction: verdicts score the action you chose against the chart or model baseline — never the river card that followed. A Blunder that got there is still a blunder; a Best call that bricked is still Best.