Fifteen games on Wednesday, and the slate had a familiar September feel: a handful of low-scoring, tidy finishes mixed with a few games that blew past their number in the later innings. Our model tracked three totals edges — all unders — and the results were uneven. One came in cleanly, two did not. No moneylines cleared our bar, so that column stays blank at 0-0-0.

The totals board finished 1-2, which is honest and not particularly fun to type. The wins and losses are in the leans section below. The official card had nothing staked, so there is no P&L to report today — just the tracking record and the DFS accounting.

No money was on the card Wednesday. Nothing on the moneyline side cleared our 5-percentage-point threshold, and the totals edges that qualified were tracked as leans only. That is the process working the way it is supposed to.

The Leans

Three totals edges made it onto the board. Here is how each settled:

• OAK @ TB — Lean: Under 7.75. Final: Oakland 3, Tampa Bay 4. Total runs: 7. WIN. A tight, low-event game that landed exactly where the model expected it to.
• BOS @ TEX — Lean: Under 8. Final: Boston 3, Texas 7. Total runs: 10. LOSS. The Rangers did most of the damage; ten runs against an under-8 line is a straightforward miss.
• SEA @ LAA — Lean: Under 8. Final: Seattle 7, Los Angeles Angels 2. Total runs: 9. LOSS. Nine runs clears the number by one. Frustrating margin, but the line is the line.

Leans: 1-2 on totals, 0-0 on moneylines. No official plays to grade.

The DFS slate had 451 players score on Wednesday, with 242 matched to our projections. The range between the best and worst outcomes for projected starters was wide — three pitchers, specifically, pulled the floor down hard.

Top Scorers

• Pete Crow-Armstrong (Chicago Cubs) — 42 DraftKings points, 3-for-4. Our projection: 10.06. He outperformed by a factor of four. That kind of day is why you roster Cubs hitters when the game environment sets up well.
• Alex Bregman (Chicago Cubs) — 33 DK points, 3-for-4. Our projection: 7.98. Another Cubs hitter delivering on a big day; the two of them together would have carried most lineups.
• Christian Encarnacion-Strand (Baltimore Orioles) — 30 DK points, 3-for-4. Our projection: 6.42, salary $4,300. At that price, 30 points is exceptional value — roughly 4.7× projected output against a mid-tier salary.
• Ethan Salas (San Diego Padres) — 30 DK points, 2-for-4. No projection on file for Salas, so we can clock the performance without claiming credit for it.
• Jakob Marsee (Miami Marlins) — 30 DK points, 2-for-3. Our projection: 7.06. Solid over-projection performance from a Marlins bat.

Best Calls

Crow-Armstrong (proj. 10.06, actual 42), Bregman (proj. 7.98, actual 33), and Encarnacion-Strand (proj. 6.42, actual 30) were the three clearest wins in the matched set. All three materially outperformed; none of it is retroactive credit — these projections were published before first pitch.

Worst Calls

The pitching misses were significant and worth naming plainly.

• Blake Snell (Los Angeles Dodgers) — Our projection: 19.66 DK points. Actual: 3.65. Line: 1.0 IP, 1 K, 0 ER. One inning. The projection assumed a full outing and got one frame. Whether that was injury, a planned hook, or something else, the outcome was a near-total projection failure.
• Xzavion Curry (New York Mets) — Our projection: 15.05 DK points. Actual: -3.45. Line: 3.0 IP, 3 K, 6 ER. Six earned runs in three innings produces a negative score. There is no framing that makes this look reasonable.
• Yusei Kikuchi (Los Angeles Angels) — Our projection: 14.56 DK points. Actual: 3.75. Line: 4.1 IP, 4 K, 4 ER. Four earned runs ate most of the strikeout value. This one also connects to the SEA @ LAA lean loss — Seattle's seven runs had to come from somewhere, and a chunk landed against Kikuchi.

Three pitching projections, three significant misses. The hitter calls were strong; the starting pitcher assumptions were not.

The under model went 1-2 Wednesday, and neither loss was close — ten runs against an under-8 in Texas, nine in Anaheim. Both games had pitching assumptions baked in that did not hold. That is a recurring theme from the DFS worst calls too: Snell, Curry, and Kikuchi all came in well short of projected innings and run prevention. When multiple starters underperform their expected depth in the same evening, under leans are more exposed than the pregame numbers suggest.

The hitter calls were genuinely good. Crow-Armstrong and Bregman putting up 42 and 33 DK points from a Cubs lineup is the kind of correlated output the model is designed to find. Encarnacion-Strand at $4,300 delivering 30 points is the value-tier result that makes salary construction worthwhile. That part of the process worked Wednesday.

The honest forward-looking note: pitcher projection reliability late in the season, when workload management is more aggressive, is worth watching. A starter projected for six innings in September may get four. The model knows their stuff; it does not always know their manager's intentions for that night. That gap showed up in the DFS results, and it may be inflating under-lean confidence in games where the bullpen is more likely to appear than the pre-series schedule implies.