Fifteen games on a Saturday in September, and the run-scoring environment cooperated with the model about as cleanly as it does. The totals went 2-0. No moneyline plays graded. That's the summary — the rest is details worth understanding.

The two tracked unders didn't squeak through. Toronto and Kansas City combined for 7 runs against a line of 9.5, and New York Yankees and San Diego combined for 6 against a line of 8. Neither game was particularly dramatic in terms of the total — both settled comfortably on the right side of the number. That matters because close calls over a season accumulate differently than comfortable ones.

No moneyline sides cleared our 5-point threshold on Saturday, so there's nothing to grade there. The totals were where the tracked edges lived, and both delivered.

No money was officially staked on Saturday — the_card was empty, so there's no official card record to settle. What we had were two tracked leans on the totals board, and both are worth grading as leans.

The Leans

• TOR @ KC — Lean: Under 9.5. Final: TOR 4, KC 3 (7 total runs). WIN. The game came in 2.5 runs under the number. Pitching and defense held up; neither offense found a second gear.
• NYY @ SD — Lean: Under 8. Final: NYY 5, SD 1 (6 total runs). WIN. A lopsided game, but lopsided in the right direction for the under — San Diego's offense didn't generate nearly enough to threaten the number.

Two-for-two on tracked totals leans is a good Saturday. The honest qualifier: these were leans, not official plays. The distinction is real — no units were allocated, which means the record reflects tracking, not staking. Both results support the model's read on those run environments, and we'll carry that signal forward.

The DFS slate scored 455 players, and we matched projections to 219 of them. Here's where things landed.

Top Scorers

• Riley Greene (Detroit Tigers) — 42 DraftKings points. Line: 4-for-5, 2 HR, 4 RBI. Our projection: 6.53 points. Greene was the day's outlier — that kind of performance puts him in a category where almost no projection system covers the distance. A 4-5 day with two home runs is a career-defining line for most players on a given afternoon.
• Zack Wheeler (Philadelphia Phillies) — 30.5 DK points. Line: 6.0 IP, 9 K, 1 ER. Our projection: 18.14 points. Wheeler outperformed by a significant margin, though the structure of the performance — high strikeout volume, limited damage — is consistent with what he does when his stuff is sharp.
• Sonny Gray (Boston Red Sox) — 30.1 DK points. Line: 7.0 IP, 7 K, 0 ER. Our projection: 19.59 points. Salary: $9,100. A clean shutout line over 7 innings. Gray outpaced his projection by roughly 10.5 points, which is meaningful at that salary level.
• Rafael Devers (San Francisco Giants) — 30 DK points. Line: 2-for-4, 2 HR, 3 RBI. Our projection: 7.73 points.
• Christian Koss (San Francisco Giants) — 30 DK points. Line: 2-for-4, 2 HR, 3 RBI. Our projection: 6.14 points.

Where the Projections Hit

The three clearest directional wins were Greene, Wheeler, and Gray — all three landed well above their projections, so the model had them pointed the right way even if it couldn't see the ceiling.

• Riley Greene: projected 6.53, scored 42. The gap is enormous, but the direction was right.
• Zack Wheeler: projected 18.14, scored 30.5. A cleaner beat — within a reasonable range of a good Wheeler start.
• Sonny Gray: projected 19.59, scored 30.1. The gap here is the most encouraging of the three, because $9,100 is a real salary commitment and Gray rewarded it.

Where the Projections Missed

Three starters fell well short of what the model expected, and all three deserve a plain look.

• Matthew Liberatore (St. Louis Cardinals) — projected 13.68, scored 2.2. Line: 4.1 IP, 3 K, 5 ER. Five earned runs in just over four innings erases most of a pitcher's DFS value before the lineup card is even dry.
• Anthony Kay (Chicago White Sox) — projected 12.84, scored 1.0. Line: 4.1 IP, 3 K, 5 ER. Nearly identical rough outing to Liberatore's. Two starters with the same rough shape on the same day is noise worth noting but not over-reading.
• Martin Perez (Atlanta Braves) — projected 11.66, scored -0.4. Line: 3.2 IP, 3 K, 4 ER. A negative DK score means he didn't survive long enough or cleanly enough to offset the damage. That's a brutal outcome for any roster that rostered him.

All three missed projections came from starting pitchers getting hit hard before recording an out-of-the-fifth. That's the occupational hazard of pitching in DFS — a bad inning or two doesn't just cost points, it removes innings that would have generated them.

The model's read on run environments held up cleanly on Saturday. Both under leans settled comfortably — not by a whisker — which suggests the edge in those games was real, not lucky. That's the kind of result that reinforces process rather than just the outcome.

On the DFS side, the spread between the model's top hits and worst misses came almost entirely from starting pitchers. Gray and Wheeler significantly outpaced projections; Liberatore, Kay, and Perez each imploded before the fifth inning. The hitter projections for Greene, Devers, and Koss were directionally reasonable — all three had positive projection bases — but none of the models anticipated the two-homer days. That's not a failure of the model so much as a reminder that individual game outcomes for hitters carry wide variance, and projecting 6 or 7 DK points on a player who scores 30 or 42 is a miss in magnitude, not direction.

The thing worth carrying forward: when starting pitchers crater early, they don't just disappoint — they actively hurt the roster. The upside case for a struggling pitcher at projection time was always dependent on innings pitched, and innings pitched requires surviving contact. We'll keep that risk in mind when building pitcher projections for deeper roster decisions.