Sixteen games on a Tuesday in late September, and the slate did what late-season slates sometimes do: it handed out a gem, a blowout, and a shutout in roughly equal measure, with very little in between to make the totals market feel predictable. The moneyline board finished 0-0-0 — no graded sides. On totals, we went 1-1. One over cashed comfortably; the other was buried under a two-run shutout. That kind of split is not a disaster, but it is a useful data point about how thin the margin was between a fine night and a quietly rough one.

The broader environment deserves a word. Three starters — Jared Jones, Drew Rasmussen, and Carlos Rodon — had the kind of outings that depress run totals by themselves. When multiple arms like that are working on the same night, even well-reasoned overs can run into trouble. The numbers bore that out.

No money was staked on the card Tuesday — card_graded is null, which means there were no official plays for this slate. The leans below are tracked edges: published positions we monitor for model feedback, not money in the ground.

The Leans

• STL @ PIT — Over 7.25 — LOSS. Final: St. Louis Cardinals 0, Pittsburgh Pirates 2. Total runs: 2. We needed 7.25 runs; the game produced 2. Jones was dominant (more on that below), and the Cardinals managed nothing. This was not a close miss.
• MIA @ CHC — Over 6.5 — WIN. Final: Miami Marlins 8, Chicago Cubs 2. Total runs: 10. Cleared the number by 3.5 runs, which is about as clean a result as an over gets.

The leans split 1-1. The STL/PIT result is the kind that stings because the thesis — a run environment producing more than 7 total — was not unreasonable entering the game. Jones simply had other ideas. We track these as leans for exactly this reason: the model's read on environment does not account for a starter throwing one of the best individual games on the slate.

446 players scored Tuesday. We matched 222 of them to projections. The gap between the top of the leaderboard and our pre-game estimates tells the real story of the night.

Top Scorers

• Jared Jones (Pittsburgh Pirates) — 37.15 DK points. Line: 7.0 IP, 9 K, 0 ER. Our projection: 14.77. That is a 22-point miss upward — the single largest gap on the board, and it came directly from the game that sank our STL/PIT over lean. Jones did not just pitch well; he had a historically clean outing by any reasonable standard.
• Randal Grichuk (Chicago White Sox) — 37.0 DK points. Line: 4-for-5. Our projection: 7.85. A nearly 30-point miss. Grichuk going 4-for-5 in a context where our model had him near replacement-level production is a significant outlier, the kind of game that is hard to project and harder to feel good about missing.
• Drew Rasmussen (Tampa Bay Rays) — 36.6 DK points. Line: 6.2 IP, 11 K, 1 ER. Our projection: 17.88. Salary: $9,800. A miss of roughly 19 points, but a directional hit — we had Rasmussen as one of our better pitching projections on the slate, and he was one of the best performers. The upside surprise was real; the magnitude of it is the note.
• Carlos Rodon (New York Yankees) — 36.2 DK points. Line: 8.0 IP, 8 K, 0 ER. Our projection: 15.68. Salary: $8,800. Another big upward miss on a starter. Rodon going eight shutout innings is an elite result at any price. At $8,800, anyone who rostered him had a strong floor to build around.
• Vinnie Pasquantino (Kansas City Royals) — 34.0 DK points. Line: 3-for-5. Our projection: 6.78. Salary: $4,100. The salary is the notable number. Pasquantino produced 34 DK points at a price point where most lineups could fit him easily. A significant outcome for anyone who identified it pregame.

Where We Hit

The best calls of the night were directionally correct on the pitching side, even where the magnitude surprised us.

• Jared Jones — Projected 14.77, scored 37.15. We had him as a legitimate DFS option; we did not have him as the top scorer on the entire slate. The direction was right. The ceiling was not.
• Randal Grichuk — Projected 7.85, scored 37.0. The honest read here is that identifying Grichuk as even a viable play was the call — at that projection, he was a low-floor, low-expectation name. A 4-for-5 night is not something a model projects with confidence for any individual player.
• Drew Rasmussen — Projected 17.88, scored 36.6. This is the best-calibrated call in the group. We had Rasmussen as a high-upside arm, and he delivered high-upside results. The gap is still 19 points, but the framework was sound.

Where We Missed

Owning the misses is part of the record.

• Yusei Kikuchi (Los Angeles Angels) — Projected 20.8, scored 7.2. Line: 5.1 IP, 2 K, 2 ER. We had Kikuchi as one of the stronger pitching projections on the slate. He gave up two runs in five-plus innings and struck out only two. That is a meaningful projection error — not a disaster in raw ERA terms, but the strikeout rate collapsed relative to expectations, which hit the DK scoring model hard.
• Logan Gilbert (Seattle Mariners) — Projected 18.7, scored -0.2. Line: 4.0 IP, 3 K, 4 ER. A negative DK score. Gilbert allowed four runs in four innings, and the model had him projected near 19 points. This is the kind of outcome that happens to starters — short, rough outings with earned runs accumulating faster than strikeouts — and it represents one of the larger single-player misses on the Tuesday slate.
• Michael King (San Diego Padres) — Projected 15.32, scored 3.6. Line: 4.0 IP, 5 K, 5 ER. Five earned runs in four innings. The five strikeouts cushioned the score slightly, but this was another case where a pitcher we liked significantly underdelivered. Three starters we projected well all had difficult nights. That is the kind of variance that makes a DFS pitching slate feel like whiplash.

The honest takeaway from Tuesday is about pitching variance, specifically how it moved in both directions at once. Jones, Rasmussen, and Rodon all cleared 36 DK points. Kikuchi, Gilbert, and King all collapsed. The model had none of the top three projected above 18 points, and none of the bottom three projected below 15. That is not a calibration failure in a narrow sense — pitching outcomes at the individual start level carry enormous variance, and no model eliminates that. But it is a useful reminder that pitcher projections set a mean, not a floor, and on a night when the top of the starting pitcher board goes nuclear, the lineup construction question is about identifying which arms have the upside ceiling, not just the best projected mean.

On the totals side, the STL/PIT loss reinforces something worth keeping in the model: a dominant starting pitching performance can compress a run environment faster than pre-game indicators suggest. Jones's outing was exceptional by any measure. We will check whether the pre-game indicators had any signal of that ceiling before deciding whether anything in the projection framework needs adjustment.

The MIA/CHC win, by contrast, was straightforward — a game that played to run-environment expectations and cleared the number cleanly. Two leans, two very different explanations. That is the nature of a late-September slate with elite arms still in rotation.