Team Side Won. Pitcher Props Missed Again.

The final recommendation was Garrett Crochet Under 18.5 pitcher outs (-152), and it lost by the smallest possible margin — Crochet recorded 19 outs (6.1 innings). The triage logic wasn’t wrong. Early-season workload management is a real signal, and the structural case for a shorter leash was sound. He just stayed out there one batter too long. Half an out. That’s the difference between a clean win and a loss that stings precisely because it felt well-reasoned.

The board split 6-6 on Sportsline and 2-1 on Rithmm, which is noise in both directions. Perplexity’s recommendation — Cleveland -115 — was the cleanest pick on the page and it won 2-1. The team-side framing held up again. The 3-loss experiment took Cleveland and cashed. The cleaner the logic, the simpler the bet, the more consistent the result.

The 6-loss experiment went with Cameron Schlittler Pitcher to Record a Win (+108), the Claude recommendation. The Yankees won 5-3 but Schlittler didn’t get the decision. This is the core problem with pitcher win props that keeps appearing in the data: the team can win and the prop still loses. Claude’s reasoning was compelling — elite ERA, weak opponent, strong offense behind him — but every word of that analysis described the team’s odds of winning, not the pitcher’s odds of getting the win. Those are different markets. The Yankees won. Schlittler did not.

The Sportsline board had moments. Ayo Dosunmu scored 24 against a 16.5 over. The Mets beat Arizona 4-3 in ten. Paul Skenes got the win in a 7-1 game. But Paul Skenes Under 6.5 strikeouts missed by exactly zero — he struck out six, one off the over threshold. That’s two near-misses on pitcher props in one day: Crochet at 19 (needed 18.5), Skenes at 6 (needed 6.5). Props at sharp lines don’t give you much room.

The LaMelo Ball under looked viable in triage — a broad combo number with multiple paths — but he finished with 47 PRA against 32.5. That’s not a near-miss, that’s a player who was locked in from the start. Star player combo unders keep showing up as triage candidates and keep getting cleared when the player is on. The pattern is consistent enough to start treating those as lower-confidence regardless of how clean the framing sounds.

The structural lesson from the day: team-side bets remain the most reliable format for the martingale experiments. Both times Perplexity has recommended a team side, it’s won. Both times Claude has recommended a pitcher win prop, it’s lost. The edge in reasoning may be real in both cases, but pitcher win props introduce a resolution path that has nothing to do with the quality of the pitcher — the team can win and the bet can still lose.