2026 NBA Playoffs · Conference Semifinals
Playoff Predictions
Series win probabilities, projected length, and championship odds from a Monte Carlo simulation built on Predictium's B13 model. Game 1 spreads come from the same feature stack that drives the regular-season picks.
10,000 simulations · net_rating_fallback model · as of May 6, 2026
Championship & Advancement Odds
Championship & Round Odds
| Team | Round 2 | Conf. Finals | Finals | Title |
|---|---|---|---|---|
| Oklahoma City Thunder | 100.0% | 81.3% | 47.8% | 27.0% |
| New York Knicks | 100.0% | 78.5% | 43.4% | 22.3% |
| San Antonio Spurs | 100.0% | 70.1% | 36.0% | 20.3% |
| Detroit Pistons | 100.0% | 74.8% | 41.3% | 20.0% |
| Minnesota Timberwolves | 100.0% | 29.9% | 9.5% | 3.3% |
| Cleveland Cavaliers | 100.0% | 25.2% | 9.6% | 3.2% |
| Los Angeles Lakers | 100.0% | 18.7% | 6.7% | 2.4% |
| Philadelphia 76ers | 100.0% | 21.4% | 5.7% | 1.4% |
Top 12 teams by title probability. Values update each time the simulation is re-run.
Conference Semifinals Predictions
Higher-seed series win probability, projected series length, and Game 1 spread/total from the B13 oracle. Injury notes reflect availability overrides baked into the simulation.
Eastern Conference
Proj. Length
5.7 G
G2 Spread
—
G2 Total
—
Series Length Distribution
Proj. Length
5.6 G
G2 Spread
—
G2 Total
—
Series Length Distribution
Western Conference
Proj. Length
5.6 G
G2 Spread
—
G2 Total
—
Series Length Distribution
Availability Notes
Luka Dončić (LAL) — Expected back Game 3 · Grade 2 left hamstring strain
Jalen Williams (OKC) — Expected back Game 2 · Grade 1 left hamstring strain
Proj. Length
5.7 G
G2 Spread
—
G2 Total
—
Series Length Distribution
Availability Overrides Applied
Manual overrides injected into the simulation to reflect confirmed playoff absences. The model adjusts per-game spreads within the specified series rather than altering the season-long priors.
| Player | Team | Series | Return | Impact |
|---|---|---|---|---|
Luka Dončić Grade 2 left hamstring strain | LAL | OKC-LAL | Game 3 | −0.0 pts |
Jalen Williams Grade 1 left hamstring strain | OKC | OKC-LAL | Game 2 | −0.0 pts |
How We Model Playoff Games
Regular season and playoff games use the same B13 model core — 204 features, RAPM player impact, split spread/total heads. The playoff simulator runs a Monte Carlo over the full bracket, drawing per-team strength perturbations each iteration to propagate true-strength uncertainty. Each simulated game is an oracle lookup into pre-computed B13 predictions plus a travel-fatigue term for lower-seed road games.
Series win probabilities come from the share of simulations in which each team wins their best-of-seven. Projected length averages the actual number of games played across all sims. Availability overrides shift a team's per-game offset within a specific series for the games the player is projected to miss — rather than haircutting the season-long priors.
Backtest: 55.3% ATS overall across 5 regular seasons, 59.1% at ≥3pt model edge. Playoff-specific backtest will publish alongside the first full round of results.