Ranked picks with source labels.

Model at 60.2% vs market 57.4% on NYY is modest comparison. BOS offense better (.736 OPS vs .692) but 25-36 record suggests context issues. Weather...

Model has ATL at 60.2%, market at 57.2% — mild comparison but not actionable. PIT's 6.9 RPG over 63 games is real offensive strength vs their .710...

Model likes WSN at 58.6% vs market's 45.2% implied—a 13.4pp gap triggering market_disagreement_large. ARI's .564 OPS is 62-game verified weakness;...

Model total of 8.97 vs market 11.0 is a 2-run gap—huge structural miss. Cabrera has no recorded IP/ERA/WHIP despite 10 starts (data issue). SFG's 0...

Model picks LAA at 53% but market has LAD at 64% implied (-190). Away pitcher is TBD/fallback with 1.2 IP sample — unknowable matchup quality. LAD'...

Model has CLE at 55.2%, market at 54.6% — convergence. Total at 9.1 vs market 7.5 is a 1.6-run gap; model likely overweighting TEX offense (4.8 RPG...

Model favors MIA 53.6% vs market 45.2%, an 8.4pp gap. TBR is 36-23 with superior .802 OPS; MIA is 29-34. Rasmussen's 1.80 ERA in 5.0 IP is tiny sam...

Model badly misreads Coors. Market total 12.0 vs model 9.03 — park factor 1.47 screams over. Sproat's 21.0 ERA in 3 IP is noise but MIL offense (.9...

Model has SEA 52.2%, market 52.8% — near consensus. Valdez 1.50 ERA in 60 IP is elite, Woo 3.00 with 13.5 K/9 also strong. Model's 9.03 total vs ma...

Christian Scott's pitcher data is completely missing (null ERA, IP, K/9, WHIP). Michael King shows 10 GS but only 5.0 IP total, which is impossible...