Zandalasini Went Off and the Model Saw It Coming

One game on the Sunday slate, Golden State hosting Los Angeles, and if you think a single-game board is a quiet night, tell that to the injury report that reshuffled both rotations before tip-off. Nearly the entire top of the scoring ledger — six of the six highest DK scorers, both best calls, all three worst calls — carried a minutes-bumped flag. Somebody got hurt, somebody filled in, and the game became a stress test of how well the model could read which role players would absorb the load and actually deliver.

The short answer: pretty well on the LA side, less so on the GS side. Cecilia Zandalasini lit the box score up for 15 points and 7 rebounds in 23 minutes for the least expensive path to 31.3 DK points on the board. Erica Wheeler quietly went 15-4-1 in 27 minutes. Cameron Brink gave you the activity line — 7 points, 7 rebounds, 2 steals, a block — in 22 minutes. That's the game in three names. The rest of the story involves a couple of projections that didn't land, and we'll get to those honestly.

The one-number receipt for the night: a mean absolute error of 6.1 DK points across 19 graded players. On a slate this small, with this many injury-driven minute bumps creating uncertainty across both rosters, that's a workable number — not a celebration, not a crisis. Here's what the box scores actually said.

Top Scorers — Projection vs. Reality

• Cecilia Zandalasini (GS, $6,300) — Projected 17.8, actual 31.3. Line: 23 min, 15 pts, 7 reb, 2 ast, 1 stl. An injury bumped her minutes, and she made every one count. We had the direction right; the magnitude was the surprise. More on her in a moment.
• Nneka Ogwumike (LA, $10,900) — Projected 37.1, actual 27.3. Line: 32 min, 16 pts, 7 reb, 3 ast. Her minutes were bumped too, and she was on the floor — the efficiency just didn't match the projection. A 9.8-point miss at the top salary on the board stings a little.
• Erica Wheeler (LA, $7,600) — Projected 25.7, actual 26.5. Line: 27 min, 15 pts, 2 reb, 4 ast, 1 stl. Dead center. We'll take it.
• Veronica Burton (GS, $9,600) — Projected 31.1, actual 23.0. Line: 29 min, 7 pts, 0 reb, 7 ast, 3 stl. She played, she was active, and the assists and steals were real — but the scoring wasn't there. An 8.1-point miss on a player at that salary tier matters for lineup construction.
• Ariel Atkins (LA, $7,200) — Projected 24.7, actual 23.0. Line: 30 min, 9 pts, 4 reb, 4 ast, 1 stl. Solid all-around line, projection was close. A 1.7-point variance is about as clean as it gets.
• Kayla Thornton (GS, $8,000) — Projected 22.8, actual 22.3. Line: 28 min, 11 pts, 7 reb, 1 ast, 1 stl. Half a point off. That's not a projection, that's a transcript.

Best Calls

• Erica Wheeler (LA, $7,600) — 25.7 projected, 26.5 actual. The model pegged an injury-bumped guard in a role expansion and got the output right to within a single DK point. At $7,600, that's 3.5 points per $1K. Clean value, clean call.
• Cameron Brink (LA, $6,900) — 21.0 projected, 21.3 actual. Line: 22 min, 7 pts, 7 reb, 2 stl, 1 blk. The model projected a floor-heavy, activity-based line for a player whose minutes were bumped, and that is exactly what came out of the box score. 3.1 points per $1K at a salary that fit comfortably in most builds.

Worst Calls — Owned Plainly

• Gabby Williams (GS, $8,400) — Projected 30.9, actual 15.3. Line: 19 min, 11 pts, 1 reb, 1 ast, 1 stl. A 15.6-point miss. Her minutes were bumped going in, and the model read more usage than she got. The minutes cap hurt her; the production per minute was fine, but 19 minutes is a hard ceiling when you're projecting 31 points.
• Dearica Hamby (LA, $9,900) — Projected 34.0, actual 18.8. Line: 26 min, 12 pts, 3 reb, 1 ast, 1 stl. A 15.2-point miss on a near-$10K salary. She logged 26 minutes — the playing time was there. The rebounding and usage volume we projected simply didn't materialize. That's the model overweighting the opportunity; the role was real, the stat-line reward wasn't.
• Janelle Salaun (GS, $7,400) — Projected 26.8, actual 13.8. Line: 21 min, 8 pts, 3 reb, 1 stl. A 13.0-point miss. Fewer minutes than projected and modest efficiency within them. Two GS players in the worst-calls column tells you something about how the Golden State rotation actually settled once the game was live.

The honest takeaway from a one-game, injury-reshuffled slate is this: the model's live-news edge — redistributing minutes when an injury hits a roster — is real and walk-forward validated, but it is not uniform across both teams. Tonight, the LA side of the ledger was reasonably clean. Wheeler and Brink were the two best calls on the board. The GS side was messier. Williams, Salaun, and to some extent Burton all underperformed projections that were already accounting for expanded roles. The rotation didn't settle the way the pregame minute estimates suggested it would.

What that means going forward: when an injury reshapes a full rotation and you're working a single-game slate, the model's output on the team with more projection uncertainty deserves a wider mental range — especially for GPP builds where ceiling matters. Wheeler at 26.5 actual DK points on a 25.7 projection is the kind of number you can build around with confidence. Williams at 15.3 on a 30.9 projection is a reminder that minute bumps create opportunity, not outcomes. There's a difference, and a 6.1 MAE on a volatile night is the model being honest about exactly that.

One last thing: Zandalasini going for 31.3 at $6,300 — nearly double her projection — is the kind of outcome you circle. The model had the minutes-bumped flag. It had the direction. It missed the magnitude by 13.5 DK points, and that's a real miss. But the signal was there. That's the difference between a cold miss and a known unknown, and it's worth remembering when you're building next time a low-salary player lands on the minutes-bumped list.