I learned the hard way that a day with nothing to bet is still a good day. Old donk-me would have picked the biggest favorite and called it a value play. New me runs the model first. Today the model says sit down, so sit down I will.
WNBA picks today: the honest answer is no play
The honest WNBA picks today answer from our model is: no play. We ran our number against the market on every WNBA game in the slate and not one cleared the trigger. When our number and the market agree, the gap is noise, not edge. The disciplined play is to pass, and a pass is the pick.
That is not a cop-out. It is the whole job. The closing line value guide explains why an edge only shows up when our number and the market actually disagree. Today the model has no game where they disagree enough to beat the vig.
What a one-book price really is
Here is the part of today's feed worth your attention: every line we track sits at exactly one book with zero dispersion. Dispersion means how much books disagree on the same price. Zero books, one price, is not yet a consensus. It is a suggestion from one shop.
Take the MLB board. The Rays at the Tigers shows a home spread of 1.5 across one book line-e193503339c577591a9f30185315b5a9. The Red Sox at the Marlins shows a home spread of 1.5 across one book too line-5cbe6ea03f242e45c63a38557db087e7. One shop, zero dispersion, on both. The market has not settled these games yet.
We apply the same test to WNBA. A single-book price is a price without a market behind it. We need two or more sharp books to agree before the number is real. Read how we source the number in the WNBA desk to see the full method.
The college openers are telling the same story
College football openers make the point even louder. San José State at USC posts a home spread of minus 38 across one book line-e654a030b2b34bed018f5a896b76cc95. Bethune-Cookman at UCF posts a home spread of minus 42.5 across one book line-1502f9bc407d446e172acae5ddcadc29. A favorite that big is a trap for anyone who reads a pinned number as a pick. We do not.
Even the tighter openers stay at a single shop. NC State at Virginia shows a home spread of minus 5.5 across one book line-0a83af774f4a64cb3d0b295a9a7ff488, and Jacksonville State at North Dakota State shows minus 7 across one book line-b82f3ee7025cbccd287428be9483096d. Same story. One shop, zero dispersion, still cooking.
Want to understand how a settled consensus takes shape before that happens? The sharp versus public explainer walks through how real money settles a line. Also worth a read: the WNBA prop edge cornerstone and the pace-adjusted totals model for how our WNBA numbers are built.
The Donk Check
Here is the rerun of my own worst habit so you do not have to buy it. The day a favorite opened at a big number, I bet it on vibes, called it analysis, and watched the market run past me. The number was one book early in the week. It was never a pick. It was a headline I wanted to believe.
So the check I run on every would-be play: is our number the market, or is our number the market plus a real gap? If it is the market plus noise, I pass. That is the discipline the track your bets guide teaches, and I trust it more than I trust old me.
What would change my mind
One concrete move flips this take: if any WNBA game moves 1.5 points or more toward our number at a second sharp book, that game becomes a bet. We would publish it with our number, the market number, the gap, and the trigger price at that exact moment, and it would be graded later on the track record page. Until a line agrees at two shops or moves that far, the honest answer for today is no play.
Source table rows cited above: game_odds:line-e193503339c577591a9f30185315b5a9, game_odds:line-5cbe6ea03f242e45c63a38557db087e7, game_odds:line-e654a030b2b34bed018f5a896b76cc95, game_odds:line-1502f9bc407d446e172acae5ddcadc29, game_odds:line-0a83af774f4a64cb3d0b295a9a7ff488, game_odds:line-b82f3ee7025cbccd287428be9483096d.
Bet responsibly — set limits, never chase losses.
Model calibration: predicted vs observed
Predicted win probability bucket vs the empirical win rate inside that bucket on the test set. Points on the y=x reference line are perfectly calibrated; points below mean the model is overconfident in that bucket.
EV per $100 across win rate × odds grid
Expected value of a $100 stake at each combination of true win rate and market odds. Anywhere the cell is positive you have a long-run profitable bet; the magnitude shows how aggressive Kelly will size it.
