Ranked picks with source labels.

Model and market converge tightly on NYY (~60% implied). Weathers lacks basic stats (IP/ERA null despite 10 starts), creating structural uncertaint...

Model has WSN at 58% but market implies MIL 54%. MIL's .972 OPS and 6.0 RPG dwarf WSN's .82 OPS and 5.1 RPG. Missing pitcher ERA/WHIP data makes to...

Model likes LAA at 58% vs market's 45%, a 13-point gap flagged as large disagreement. Both teams 3.8 RPG, similar offensive profiles. Model total 8...

Model at 57.5% vs market 56.4% on SEA—barely any comparison. Both starters show 0.00 ERA in 6 IP which is obviously tiny samples, no persistent sig...

Cease's 1.69 ERA and 20.25 K/9 in 5.1 IP is microscopic sample, but market respects it at -121. Prielipp is pure fallback data—no ERA, IP, or K/9 t...

Model has CIN 56.1%, market implies PIT 54.4% — tight disagreement. Model's 9.82 total is 1.82 runs over market's 8.0, likely overweighting PNC's 1...

Griffin Jax has 1.0 IP this season (18.00 ERA); Roupp has null stats. Neither starter has meaningful data. Model predicts 9.23 runs vs market 7.5,...

Sasaki's stats are completely missing (null ERA/IP/K9/WHIP). McGreevy's 6.0 IP over 10 starts averages 0.6 IP per start—clearly incomplete data. Mo...

Market implies CHC 60.3%, model has 53.6% — market is more bullish on Cubs. Both starters have 7+ ERAs in tiny samples (5.0 IP, 4.2 IP), making pit...

Away pitcher Arrighetti on fallback data with 7.94 ERA and 1.94 WHIP creates uncertainty. Model picks HOU at 52.2% but market has BOS -124 (54.2% i...

DET pitcher Montero has fallback data with only 4.1 IP across 10 'starts' (0.4 IP/start suggests bulk reliever usage or data error). TEX pitcher Ro...

Market at 63.5% vs model 50.6% is a 13pp gap. King's 0.00 ERA in 5.0 IP is absurdly small sample but Burke's 6.75 ERA in 4.0 IP is equally noisy. P...