Our WNBA model published game picks all season, from the May openers through the playoffs: moneylines, spreads and totals. All three finished in profit. This post goes through each one, with the record as published and the line movement that came after.
Every number here comes from the public record on the WNBA backtest page, read on November 10. This post covers the game markets only. Player props have their own section of the record on the same page, and we'll write about them separately.
How the record is booked
Each pick is booked to win one unit at the vig-free price of the side taken. A favourite risks more than a unit to win one, an underdog risks less, and there is no sportsbook margin in any figure below. That also means breakeven is the price itself. A 47% hit rate on underdogs can make money, and a 55% hit rate on heavy favourites can lose it.
The game record
| Market | Settled | Record | Units | Risked | Return |
|---|---|---|---|---|---|
| Total | 138 | 77-61 | +16.40 | 137.82 | +11.9% |
| Spread | 132 | 70-62 | +8.35 | 131.68 | +6.3% |
| Moneyline | 111 | 53-58 | +2.35 | 148.57 | +1.6% |
Totals did most of the work. A 77-61 record on prices near even money is a 55.8% hit rate, and it returned 11.9% on what was risked. Spreads ran at 53.0% for a 6.3% return. Moneylines went 53-58 and still finished slightly ahead, because on a moneyline the units depend on the prices of the sides that won, not on the count of wins.
Did the line move our way?
The measure we care about most is early line value: how far the market moved toward our side between when we published a pick and tip-off. It says whether we were early on something the market later agreed with, which a win-loss record over a hundred-odd games can't tell you on its own.
| Market | Average early line value |
|---|---|
| Total | +0.36 points |
| Spread | +0.09 points |
| Moneyline | +0.63 probability points |
All three are positive. The totals number is the one that stands out. On average the total moved more than a third of a point toward our side after we posted, which lines up with totals being the strongest market in the results. Spreads moved our way too, though by less than a tenth of a point, so the spread record leans more on results than on the market coming around to us.
What we take from it
The samples are modest. A hundred-odd picks per market is enough to see a direction and not enough to call any of these returns settled. A spread record of 70-62 could plausibly have landed a few games either side of where it did. We'd rather say that plainly than lean on the return figure.
The two things pointing the same way are worth more than either alone. Totals made the most money and also drew the most line movement toward us. That is the pattern we'd want to see from a model that is publishing before the market settles rather than getting lucky on the result. It is also the market we'll watch most closely next season.
The full game record, market by market, is on the backtest page beside everything else the model published, losses included.
Nothing here is betting advice. If you bet, do it where it is legal, at 21 or older, and with money you can afford to lose.