Pitching Ran the Show, Until Dodgers Blew It Open

Seventeen games on Wednesday, and the theme announced itself early: pitching. Runs were hard to find across the slate, which suited our totals card just fine — right up until a nine-run Phillies ballpark decided to remind everyone what can happen when two good offenses show up at the same time.

No moneyline plays graded — we published no ML picks Wednesday, so that ledger stays clean at 0-0-0. The totals card finished 2-1-2, which is a winning day the way a 12-inning tie is a winning day: you'll take it, but there's a little more nuance to unpack. The two pushes sat right on the number, which is its own kind of frustrating. The win that mattered most was a blowout in the right direction. The loss stung the way LAD-PHI losses tend to — loudly.

Five totals published Wednesday. Here's every one of them, straight.

• BAL @ BOS, Under 9 — Final: BAL 3, BOS 6. Nine runs total, right on the number. PUSH. The kind of game that makes you want to stare at the ceiling.
• CIN @ SEA, Under 8 — Final: CIN 5, SEA 3. Eight runs total, right on the number again. PUSH. Two games, same fate, different coasts. The baseball gods were not in a generous mood.
• STL @ LAA, Under 9 — Final: STL 1, LAA 0. One run. One. The Angels and Cardinals combined for a single run on what was projected to be a nine-run game, and we were on the right side of it by a comfortable margin. WIN. That's the one you frame.
• LAD @ PHI, Under 9.5 — Final: LAD 9, PHI 5. Fourteen runs. The model liked the pitching matchup, the ballpark said otherwise, and these two offenses had a conversation we weren't invited to. LOSS. Own it. That one hurt.
• BAL @ BOS (second entry), Under 9 — Final: BAL 5, BOS 1. Six runs, well under the number. WIN. A clean one to close the ledger.

Final totals line: 2-1-2. No ML plays published, no ML plays graded. The card said what it said.

Five hundred players scored Wednesday. The model had projections matched against 310 of them. Here's where it landed.

Top Scorers

• Joey Ortiz, Milwaukee Brewers — 31.0 DK points. Went 3-for-3 with 2 RBI and a stolen base. Our projection: 6.18. He was the best hitter on the slate by a mile and we undervalued him by about 25 points. That's a miss on the projection side, though anyone who had him in a lineup was very happy to be wrong alongside us.
• Cade Cavalli, Washington Nationals — 29.9 DK points. Six innings, eight strikeouts, zero earned runs. Our projection: 14.91. He nearly doubled our number and landed in the best-call column — one of the sharper SP calls of the day regardless of what the raw projection said.
• Gerrit Cole, New York Yankees — 29.4 DK points ($8,800 salary). Seven innings, eleven strikeouts, three earned runs. Our projection: 14.7. Cole did what Cole does at full health — he goes deep, he misses bats, and he costs you almost nine thousand dollars. He earned every penny of it Wednesday.
• Keider Montero, Detroit Tigers — 28.2 DK points. 5.1 innings, eight strikeouts, one earned run. Our projection: 13.25. He outperformed the number cleanly — a good call that got even better.
• James McCann, Arizona Diamondbacks — 28.0 DK points. Went 3-for-4 with a home run and 3 RBI. Our projection: 7.5. McCann is a 37-year-old backup catcher who decided Wednesday was the night. We projected 7.5. He gave you 28. Nobody had that coming, and it's okay to say so.

Where the Model Hit

• Ortiz — the model flagged him; the actual production was far beyond what we projected, but the direction was right. A best call by outcomes even when the number undersold the ceiling.
• Cavalli — 14.91 projected, 29.9 actual. Washed. But the call to play him was sound, and if you had him, you were well ahead for the night.
• Cole — 14.7 projected, 29.4 actual. Same story. The model liked the matchup; Cole went out and threw one of the better lines of the summer.

Where the Model Missed

• Sandy Alcantara, Miami Marlins — projected 14.73, finished with 5.8 DK points. Seven innings, three strikeouts, five earned runs. Alcantara has had a complicated year getting back to full form, and Wednesday reminded us that projecting him near 15 points assumed a version of him that didn't show up. That's on the model, and we'll own it.
• Shohei Ohtani, Los Angeles Dodgers — projected 13.7, finished with 2.0 DK points. Went 0-for-4. The best player in baseball goes hitless in a game his team scored nine runs. That's baseball. The model wasn't wrong to trust him; he just had a quiet night while his teammates did the work.
• James Wood, Washington Nationals — projected 13.67, finished with 2.0 DK points. Went 0-for-4 with an RBI. An RBI and 0-for-4 is a very particular kind of DFS heartbreak. The model saw the matchup and liked it. Wednesday had other ideas.

The STL-LAA game was a reminder of what a low-run-environment matchup can look like when everything breaks right — one run total is an outlier, but we were pointed at the right number and the right direction. The two pushes are a feature of betting totals, not a bug: when you're right about the run environment and the game lands on the number, that's the market doing its job. It stings a little, but it means the read was essentially correct.

The LAD-PHI loss is worth sitting with. Nine and a half was a number that acknowledged both offenses were capable of running it up, and they did exactly that. Fourteen runs in a game with two legitimate rotations suggests the ballpark context and the day's conditions mattered more than the model weighted them. That's an honest note for the process going forward.

On the DFS side, the pitching projections for Cole and Cavalli translated into real production at the top of the leaderboard. The Alcantara miss and the Ohtani zero are the kind of variance that comes with a full slate — you can be right on the process and still take the loss on the scoreboard. The Ortiz performance is the most interesting data point: when a player puts up 31 points against a projection of 6.18, that's not model error, that's a player having the game of his month. The model isn't built to predict that. No model is.

We go again Thursday. The process stays the same: call what the data actually supports, own what doesn't land, and never dress up a push as a win just because the math was close.