WNBA Live Model · Performance Report · Every Pick Graded
WNBA MODEL LEDGER
This is how the live model is actually performing, not how we wish it was. It reads each game's scoring pace in real time and locks one OVER/UNDER call before halftime, then grades it against the final score. Win rate, ROI, net units and how close each projection landed all sit out in the open. Tap any game to see the full read. No cherry-picked winners, no quiet edits, no spin.
Loading the ledger
Fetching the latest graded results
Unable to load the ledger
We're having trouble loading the ledger right now. Please refresh the page to try again.
EQUITY CURVE
Every graded pick stacked in order, oldest to newest, at one flat unit per pick (−110). Green steps are wins, red are losses. The dashed line marks the peak, and the bracket marks the deepest drawdown along the way. Tap any point for the pick behind it.
Result tape
Equity
Win
Loss
Push
Peak
Break-even
FULL PICK LOG
Every graded pick, newest first. Tap a row to open the full read: the model's projected total, the final, and how far off it landed. Filter by result if you want; the summary numbers up top never move.
| Matchup | Pick | Proj | Final | Result | Net |
|---|
HOW THIS IS GRADED
The rules behind every number on this page, so you can judge the record on its own terms.
The model
A live, in-game engine. It tracks first-half scoring pace and projects the full-game total, then takes the side with edge and locks the call before halftime. One pick per game, no second guesses after the fact.
What counts
A pick is logged once the game goes final and the total is settled against the line. Wins, losses and pushes all count. A push returns the stake, so it counts as a decision but not a win or a loss.
Net units
Every pick is one flat unit priced at −110. A win returns +0.91, a loss is −1.00, a push is 0.00. ROI is net units divided by total units risked.
Model accuracy
Each pick stores the projected total at the moment it locked. We show that against the actual final and the average miss in points. The signed bias is shown too, so you can see if the model leans high or low. Nothing is hidden.