Why the WNBA, and why now
Here's a gap hiding in plain sight: the WNBA is the fastest-growing league in American sports, and there is almost no rigorous, public, week-to-week probability on it. Plenty of takes. Very few numbers. That's exactly the kind of empty seat the Ninjas were built to fill — post an honest forecast, put it on the record, and let you argue with it.
And the timing is free. It's mid-July: football's still weeks away, baseball's grinding through the dog days — and the WNBA is right in the thick of a title race. So we pointed the same machine that runs our NBA board at it.
The call
Minnesota is the team to beat. The Lynx (18-6) come out as a clear 39% favorite to win the championship and a 57% pick to reach the Finals — the only side the model is truly confident in. That's earned: they own the league's best record and the best point differential, and the model rewards teams that win and win comfortably.
Behind them, it's a scrum. Golden State's expansion Valkyries have been the story of the season and land second at 17%; the Aces sit at 11%, Dallas at 10%, Indiana at 9%, Atlanta at 6%. Nobody in that group is out of it — but nobody's close to Minnesota either. Eight teams make the playoffs, and the current top eight are mostly settled; the live drama is Washington clinging to the last seat at 75% while the pack chases.
How the number is built
Same engine as our NBA model. Each team's strength is its Pythagorean net rating — how many points it scores and allows, which predicts the future better than raw win-loss — regressed toward the league average so a hot month doesn't get taken at face value. From there we play out every team's remaining games 10,000 times, seed the league one through eight, and run a best-of-five bracket to a champion. It's deterministic, so when a number moves on our Ledger, it moved because a game happened — never because the dice rolled differently.
It's a v1 — and it's on the record
Two honest caveats. This is a first-cut model: it reads team scoring, not individual availability, so a star sitting out doesn't move it until the results do, and it runs one generic series length rather than the real playoff format exactly. We've said as much on the methodology page. Second — and this is the whole point — every one of these numbers is now timestamped to our Ledger and will be graded against what actually happens. No backdating, no quiet edits. If we've got Minnesota too high, the season will say so, in public.
Think the Valkyries are for real, or that someone's sleeping on the Aces? Good. Go look at the board and tell us where we're wrong.