The Unders Got Lit Up — Except When They Didn't

Sixteen games on a Tuesday, and the run environment decided it was going to be complicated about it. We came in with four totals leans — all unders — and the slate split into two completely different baseball days depending on which game you happened to be watching. In one park, a pitcher carved up a lineup for 12 strikeouts. In another, a starter got touched for eight earned runs in four innings. The market had the same number on both. That's the job.

Final line: 0-0 on the moneyline (no sides cleared our bar), 1-3 on totals. Not a good night. Let's go through it honestly.

Four totals leans published. Here's how each one finished:

• Cleveland Guardians @ Cincinnati Reds — Under 9 — Final: CLE 6, CIN 5 — 11 total runs. LOSS. Two teams that combined for 11 runs in the first game of a doubleheader. The line was 9. The baseball was not interested in the line.
• Baltimore Orioles @ Detroit Tigers — Under 9.25 — Final: BAL 0, DET 14 — 14 total runs. LOSS. This one got out of hand in a hurry. We'll get to the pitching in a moment, but Baltimore's starter had a day that made the under feel optimistic in the wrong direction.
• CLE @ CIN (Game 2) — Under 9 — Final: CLE 0, CIN 2 — 2 total runs. WIN. There it is. The second game of the doubleheader was the one where the pitching actually showed up, the offenses went quiet, and an under looked like the obvious call in hindsight. It is rarely obvious beforehand. We'll take it.
• Seattle Mariners @ Los Angeles Dodgers — Under 9.5 — Final: SEA 7, LAD 6 — 13 total runs. LOSS. Dodger Stadium has historically played as a pitcher's park, and Tuesday it played like a bandbox. Thirteen runs, a one-run margin, and another under that didn't survive the evening.

One winner out of four. The under lean was right in concept on three of those games in the sense that offense shouldn't have been that easy — and yet. That's not an excuse, that's just the game telling you it doesn't owe you anything.

The betting side was rough. The DFS side was more of a mixed bag, which is a more forgiving way of saying some projections were dead-on and some were embarrassingly wrong. Let's do both.

Top Scorers on the Slate

• Eduardo Valencia, Detroit Tigers — 40.0 DK points. Went 4-for-5 with 2 home runs and 4 RBI. Our model had no projection on him, no salary on file. He's the guy who made that BAL @ DET final look the way it did, and if you had him, you had a big night. If you didn't, well, now you know the name.
• Gavin Williams, Cleveland Guardians — 33.4 DK points. Seven innings, 12 strikeouts, 2 earned runs at a $9,300 salary. Our projection: 17.53 points. He beat it by nearly double. That's a real hit — more on this below.
• Sal Stewart, Cincinnati Reds — 30.0 DK points. Went 2-for-3 with a home run, 3 RBI, and a stolen base at $4,800. Our projection: 9.4 points. He tripled it. Another clean projection beat at a price point that made it genuinely useful.
• Dominic Canzone, Seattle Mariners — 30.0 DK points. Went 2-for-4 with 2 home runs and 3 RBI at $3,600. Our projection: 7.81 points. Nearly four times the projection at the lowest salary of the group. That's the kind of line you almost don't believe when you see it in a box score.
• Logan Henderson, Milwaukee Brewers — 28.4 DK points. Went 5.2 innings, 7 strikeouts, 0 earned runs. Our projection: 16.19 points. He outperformed by 12 points. No salary on file, but the performance was legitimate.

Where We Hit

• Gavin Williams: Projected 17.53, scored 33.4. We liked him, the slate liked him more. Seven innings and 12 punchouts in a game we had leaned under — he was a big reason that second Reds-Guardians game finished 2-0. That's a projection doing its job and then some.
• Sal Stewart: Projected 9.4, scored 30.0. A $4,800 bat going for 30 points is what DFS value looks like on a good day. The projection didn't fully see it coming — a 9.4 is essentially a floor play — but flagging him at that salary was the right instinct.
• Dominic Canzone: Projected 7.81, scored 30.0. Same story as Stewart — low projection, low salary, massive outperformance. Two home runs from a $3,600 player is exactly why the bottom of the salary pool matters.

Where We Missed

• Slade Cecconi, Cleveland Guardians — Projected 15.1, scored -0.6. Three and two-thirds innings, 2 strikeouts, 4 earned runs. This one didn't work. A projection that size implies a reasonable length-of-outing assumption, and Cecconi didn't get there. Negative DK points. Owning it.
• Dean Kremer, Baltimore Orioles — Projected 14.82, scored -5.6. Four innings, 4 strikeouts, 8 earned runs. The guy who made that 14-0 final possible from Detroit's perspective. -5.6 DK points is about as bad as pitching outcomes get. We had him projected as a mid-range SP play. He was something else entirely.
• James Wood, Washington Nationals — Projected 13.88, scored 4.0. Went 0-for-4. The projection liked him; the at-bats didn't. Some days the best hitters on paper go hitless — that's baseball in a single game, and it's why roster construction across multiple spots matters.

The honest takeaway from Tuesday: under leans in doubleheader contexts are not a free pass. Game 2 of the CLE-CIN twin bill went exactly the way you'd hope — pitching dominated, two runs scored, done. Game 1 went sideways. The same logic, the same line, two very different games. That's not a process failure; that's variance. But it's a good reminder that doubleheader pricing has its own ecosystem and deserves sharper scrutiny on a game-by-game basis rather than a blanket approach.

The DFS lesson is more straightforward: the model found real equity in Gavin Williams at $9,300, and the salary-floor plays in Sal Stewart and Dominic Canzone turned into the kind of scores that win contests. The misses — Cecconi and Kremer — were both pitchers who got knocked around early. Starting pitcher volatility is the most consistent truth in DFS, and Tuesday was another data point in that file.

One more thing: Eduardo Valencia went 4-for-5 with two home runs in a game we had no projection on. He's a name to know going forward. The model will catch up. So will we.