Burton Went Off and the Model Saw It Coming

One game on the board Friday — Golden State hosting Dallas — and it turned into exactly the kind of night that reminds you how much injury news reshapes a WNBA slate before tip-off. Both teams were shorthanded. Both rosters were redistributing minutes. The box scores that came out the other end looked nothing like a normal Friday in August.

On the Golden State side, the beneficiaries ran with it. Veronica Burton was everywhere — 22 points, 7 assists, 4 rebounds in 30 minutes, the kind of line that shows up when a playmaker gets extra runway and uses every inch of it. Kayla Thornton turned 27 minutes into a quiet double-double: 11 points, 10 rebounds, looked like she'd been starting all season. Kiah Stokes did the dirty work — 7 points, 7 rebounds, 3 assists, 2 blocks in 26 minutes, the sort of stat-stuffing performance that prints DraftKings points without a box-score headline.

Dallas was a different story. Paige Bueckers logged 29 minutes and 17 points but never found a rhythm beyond that. Arike Ogunbowale had one of those 5-point, 27-minute nights you don't want to look at twice. And Jessica Shepard, who drew the slate's biggest salary and the model's biggest projection, grabbed 11 rebounds and 6 points — the kind of line that keeps a team in a game and absolutely destroys a DFS lineup.

Golden State's injury-driven minute bumps produced. Dallas's, broadly, did not. That asymmetry is the story of August 7th.

Twenty players graded. Mean absolute error: 6.6 DraftKings points. That's the honest one-number receipt for the night — not a disaster, not a triumph, a reflection of what happens when you get one side of an injury-shuffled slate right and the other side badly wrong.

Where We Hit

Veronica Burton ($9,500) is the call of the night, and it isn't close. We projected 30.2 DK points. She posted 41 — a stat line of 22 pts, 7 ast, 4 reb, 1 stl in 30 minutes. An injury had bumped her minutes coming in, and the model flagged it. She rewarded every lineup that listened. That's a 10.8-point beat on a near-$10K salary, which translates to real value at 4.3 DK points per $1,000.

Kayla Thornton ($8,100) projected at 21.7 and delivered 32 — 11 pts, 10 reb, 3 ast, 1 stl in 27 minutes. Another injury-bumped minute projection that cashed. The double-double floor made her a safe tournament piece; the upside made her a winner. At 3.9 DK points per $1K against a projection that was already optimistic, this one exceeded even the model's confidence.

Kiah Stokes ($6,400) projected at 20.1 and finished at 24.8 — 7 pts, 7 reb, 3 ast, 2 blk. The blocks made the line. At $6,400 she was the slate's clearest value play, and at 3.9 DK points per $1K on actual production she justified every dollar. Min-bumped, caught it, said so.

Gabby Williams ($9,000) was projected at 30.3 and came in at 29.3 — 19 pts, 4 ast, 1 reb, 1 stl in 22 minutes. Near-perfect projection. An injury had bumped her minutes too; the model's read on her efficiency held up almost to the decimal.

Where We Missed

Dallas was rough, and we'll say so plainly.

Jessica Shepard ($11,100) was our biggest swing and our biggest miss. Projected at 42 DK points — the slate's highest projection. She played 26 minutes, grabbed 11 rebounds, scored 6 points, and finished at 20.8. A 21.2-point miss. An injury had bumped her minutes, and the model loaded up on her usage. She saw the floor time. The production just wasn't there. A double-digit rebounding performance that doesn't come with points or assists is worth about half what you hoped for in DFS scoring, and that math caught us badly.

Arike Ogunbowale ($9,400) projected at 31.3. She finished at 14.5 — 5 pts, 4 reb, 1 ast, 2 stl in 27 minutes. A 16.8-point miss. She played the minutes. She just didn't make them count, and on a night where Golden State's defense held, even injury-driven usage projections on the Dallas backcourt ran into a wall. At $9,400, that's a salary you don't recover from in a single-game slate.

Paige Bueckers ($10,500) projected at 38.1 and delivered 21.8 — 17 pts, 1 reb, 1 ast, 1 stl in 29 minutes. A 16.3-point miss on the slate's second-highest salary. She was min-bumped entering the night, so the minutes were there. The efficiency wasn't, and the peripheral stat contribution was nearly invisible. When Bueckers doesn't facilitate and the line is basically points-only, the DraftKings total runs lean.

The Dallas trio combined for a 54.3-point miss against projection. That's the night in one sentence.

The honest takeaway from August 7th is that injury-driven minute bumps are the model's real edge — and tonight proved both why that edge exists and why it isn't enough on its own.

On the Golden State side, the minute bumps were real, the usage followed, and the production matched. Burton, Thornton, Stokes, and Williams all saw elevated floors because of injury-related redistribution, and all four delivered at or above projection. That's the signal working the way it's supposed to.

On the Dallas side, the minute bumps were also real. The minutes also came. But production in a single-game slate is subject to game script, defensive matchup, and just the randomness of basketball in ways that a projection model — any projection model — can't fully account for. Shepard's 11-rebound, 6-point line is a night that happens to centers sometimes. Ogunbowale's 5-point game is a night that happens to scorers. Neither outcome was predictable from the inputs.

What this tells us going forward: when the model flags min-bumped players on both sides of a single-game slate, the right construction approach is to diversify the Dallas exposure rather than concentrate at the top. Shepard and Bueckers together at a combined $21,600 in salary left almost no margin for the kind of cold offensive night Dallas had. Spreading that salary down the Dallas depth chart — or leaning harder into the Golden State injury-bumped tier — would have been the better build in hindsight.

The model is walk-forward validated against naive baselines, and a 6.6 MAE on a slate this chaotic is a number we can work with. But the miss tonight was real, it was concentrated, and it came from the same source as the hit: trusting injured-roster minute projections. The difference was execution on the court, not the model's read of the pregame inputs. That's a distinction worth keeping in mind the next time the injury wire reshapes a slate before tip.