Injuries Reshuffled the Deck and Nneka Dealt the Winning Hand

Two games, 39 graded players, and a slate that announced itself early with injury news that quietly reshuffled who mattered. Four of the six top scorers on the night were playing in expanded roles because of teammate absences — which meant the model's live-news edge was going to decide whether this was a good night or a forgettable one. Mostly, it was a good night.

A'ja Wilson was, again, A'ja Wilson — 27 points, 8 rebounds, 2 blocks in 30 minutes for the Las Vegas Aces, finishing at 45.0 DK points. Nneka Ogwumike went off for the Los Angeles Sparks: 24 points, 5 assists, 4 steals, 44.5 DK points. Jackie Young quietly put up 23 and 8 assists for 44.0. Over in Washington, Sonia Citron gave the Mystics 22 points and 3 steals, and Shakira Austin posted a tidy double-double. Up in Toronto, Laura Juskaite was the low-key story of the slate — 8 boards, 2 steals, a block, and 32.5 DK points out of a $7,300 salary. That one mattered.

The one-number receipt first: a mean absolute error of 4.9 across 39 players. That's the honest ledger. Some nights that number flatters you, some nights it stings. Tonight it reflects a model that caught the big directional moves but left some points on the table in both directions.

Top Scorers — Projected vs. Actual

• A'ja Wilson (LV, $12,600) — Proj: 51.2 · Actual: 45.0 · Line: 27 pts, 8 reb, 2 ast, 1 stl, 2 blk in 30 min. An injury had bumped her minutes projection upward, and she delivered a strong game. The model ran a little hot on her — a 6.2-point miss — but you'll take that direction every time. You projected a monster; you got a very good game.
• Nneka Ogwumike (LA, $10,000) — Proj: 35.3 · Actual: 44.5 · Line: 24 pts, 4 reb, 5 ast, 4 stl in 30 min. An injury had bumped her minutes, and she more than justified every one of them. A 9.2-point beat to the upside. This was a best call — more on that below.
• Jackie Young (LV, $10,200) — Proj: 38.4 · Actual: 44.0 · Line: 23 pts, 4 reb, 8 ast, 1 stl in 26 min. An injury had bumped her minutes as well. She ran hot on assists — 8 dimes is the kind of line that looks easy in hindsight and is genuinely hard to project. A 5.6-point miss, upward.
• Sonia Citron (WSH, $9,900) — Proj: 30.6 · Actual: 38.8 · Line: 22 pts, 3 reb, 4 ast, 3 stl in 32 min. No minutes bump flagged here — Citron just played well. An 8.2-point beat. Best call territory, and one the model earned the old-fashioned way: saw the role, respected the salary, and let the player do the rest.
• Shakira Austin (WSH, $11,000) — Proj: 36.9 · Actual: 37.5 · Line: 19 pts, 10 reb, 3 ast, 1 blk in 29 min. A dead-on 0.6-point miss. The model called 36.9; she delivered 37.5. That's the kind of projection that makes the MAE look good. No minutes bump — just a clean read on a reliable player at a fair salary.
• Laura Juskaite (TOR, $7,300) — Proj: 22.6 · Actual: 32.5 · Line: 14 pts, 8 reb, 2 ast, 2 stl, 1 blk in 28 min. An injury had bumped her minutes, and she did more with them than the raw projection captured. A 9.9-point beat on a player at $7,300. That is the value-play signature of this model — catch the minutes bump, get the player in lineups, let the stat sheet do the rest.

Best Calls

The three names that outperformed projection by the widest and most meaningful margins were Juskaite (+9.9), Ogwumike (+9.2), and Citron (+8.2). What connects Juskaite and Ogwumike is straightforward: an injury had bumped both players' minutes, and the model identified that before lineups locked. Citron is the quieter win — no injury context, just a clean projection on a player who delivered.

Juskaite at $7,300 is the one that matters most for lineup construction purposes. At 4.45 DK points per $1,000 of salary against a projection of roughly 3.1 per $1K, she was the kind of salary-efficient performance that wins GPP contests outright. The model put her in that conversation. She answered.

Worst Call

Own it plainly: Dearica Hamby (LA, $8,600) was projected at 30.1 and finished at 15.5. An injury had bumped her minutes going in — the model saw the opportunity and moved her up accordingly. She did not deliver. 9 points, 4 rebounds, 1 assist in 24 minutes. A 14.6-point miss to the downside, and the worst call on the slate by a significant margin.

This is the honest part of walk-forward validation: the minutes-bump signal creates real upside. It also creates real variance. When the model moves a player up based on injury news and that player goes cold, you eat the miss. Hamby went cold. We ate it. The MAE of 4.9 includes that number — that's what makes it a real figure rather than a comfortable one.

The minutes-bump signal is still the most actionable edge this model produces, and tonight was a clean illustration of why — and a fair reminder of its limits. Ogwumike and Juskaite both got the bump, both rewarded lineups that trusted it. Hamby got the bump and didn't convert. Two out of three is a good rate. It's also not a perfect rate, and it never will be.

What the 4.9 MAE tells you is that across 39 player projections, the model's average error was under 5 DK points. The walk-forward validation methodology means that number is tested out of sample — it beat the naive baselines, not a backfitted curve. Tonight's number is consistent with that track record. Some misses above, some below, one bad one with Hamby, and a handful of genuine value hits that justified the process.

The forward-looking takeaway is this: when the model flags an injury-driven minutes bump on a player at the lower end of the salary range — a Juskaite type, a player who wasn't priced for expanded usage — that is where the projection edge is sharpest. The salary hasn't caught up to the opportunity. Tonight, that gap was worth nearly 10 DK points. That's the number to remember when you're building lineups the next time the injury wire lights up before lock.