Your model likes the over, but who actually pitches the seventh? MLB bullpen workload answers that. Pitch counts, back-to-back days, high-leverage use and travel all decide which relievers a manager can trust, and a season ERA hides every bit of it. You will build a small availability board for four relievers, assign rough probabilities instead of yes-or-no flags, and split a run projection into a starter segment and a relief segment. The lesson also shows how first-five markets differ from full game, and why you should never subtract projections built on mismatched assumptions. Do this work before the lineup cards post, and a full-game total stops being a guess about the back end.
Learning objectives
Explain the mechanism, reproduce the worked examples, identify failure modes, and use the exercise to show comprehension.
