May 26 — 0-1, moneyline loss cost half unit

Yesterday

0-1 · -0.50u · ROI -100.0% · CLV +1.32%

Night 1 Recap: 0-1, -0.50u

The Royals thesis crumbled spectacularly in a 15-1 blowout loss to the Yankees. While we correctly identified KC's bullpen weakness (6/100 rating vs NYY's 92/100), the 86-point disadvantage proved even more devastating than projected. The Yankees' superior 109 wRC+ lineup overwhelmed Kansas City from the opening inning, turning what looked like market value into a rout. Despite the brutal result, we captured 1.32% closing line value, suggesting the handicap was sound even if the execution failed.

What missed

Worst result. KCR ML (NYY @ KCR) at +180, -0.50u — final 1-15, CLV +1.3%. Graded unlucky (positive CLV, lost).

The model catastrophically misjudged Kansas City's bullpen, predicting a 5.8 score versus New York's 92.3, yet KC's relief corps dominated after Bailey Falter's early exit. Falter's 7 ER in 2.33 IP aligned with his poor 4.34 xFIP prediction, but the model's 8.49 total runs estimate missed the 16-run explosion by nearly double. With 1.32% CLV, this was an unlucky loss on a fundamentally sound bet. Priority retrain: bullpen scoring models show systematic bias favoring highly-rated units.

Under the hood

Yesterday's update focused on refining the runline betting system. The development team removed a problematic feature that was generating home team runline recommendations and tightened the criteria that determine when runline bets qualify for publication. These changes aim to improve the accuracy of runline predictions by eliminating a source of inconsistent picks and raising the bar for which bets make it to users.