Methodology · WNBA
How the WNBA model works
Championship and playoff odds for every team, rest of season.
- Version
- v1 (Pythagorean net rating + rest-of-season Monte Carlo)
- Updated
- Weekly in season
- Data source
- ESPN (free) — records + points for/against
- Calibration
- New — grades on the Ledger through the season
The question
Each team's odds to make the eight-team playoff, reach the Finals, and win the championship — plus a projected final record.
The data
ESPN's free feed: current records and points for/against for every team, refreshed weekly.
How strength is measured
Talent comes from the Pythagorean expectation on points scored and allowed (the same net-rating idea as our NBA model), regressed toward the league average so a small sample can't run away.
Recency & regression
The rating uses season-to-date scoring and updates as games are played; the pull toward average keeps a hot or cold stretch from being taken at face value.
Injuries & roster
Not modeled at the player level in v1 — strength is inferred from team scoring, which already reflects who's on the floor. A known simplification.
Home advantage
A home-court edge is applied in the playoff series (with a 2-2-1 pattern for the higher seed).
The simulation
A 10,000-run Monte Carlo plays out each team's remaining games, seeds the league 1–8, and runs a best-of-five bracket (quarters → semis → Finals) to a champion. Deterministic seeding means ledger movement is real, not simulation noise.
Postseason format
Eight teams, seeded league-wide by record, in a single-elimination-by-series bracket. (The real first round is best-of-three; v1 uses best-of-five throughout — a small simplification.)
Backtesting
Brand-new for this season, so there's no track record yet — which is exactly why it goes on the Ledger now and gets graded as the games land.
Known weaknesses
No player-level or availability modeling, a single generic series length, and net rating alone can miss a team that's been unusually lucky or unlucky. A labeled first cut.
Every published number is timestamped to the Ledger the moment it's made and graded against results — nothing here is backdated. Percentages are rounded for readability.