by Charlie of Mr. Danoff's Teaching Laboratory
Last Updated on June 8th, 2026 at 6:30 PM Central Time
Looking only the first two games of the NBA Finals, this tracker assesses the San Antonio Spurs' chances of overcoming their deficit. A 13.5% chance means that, historically, teams in similar situations have won approximately 13.5 times out of every 100 attempts. While challenging, comebacks from this position have occurred before in NBA history.
The Spurs win when they move the ball (~27 assists), rebound (~50 boards), and shoot efficiently (~48% FG) — against the Knicks they are averaging 19 assists and 42% shooting, both well below their winning profile.
Based on those Finals performances, the model puts their comeback probability near zero. As you can see in the table if they get back to shooting at 45% or higher, the model suggests they have a realistic chance of coming all the way back.
Remaining schedule: G3 Away · G4 Away · G5 Home · G6 Away · G7 Home
| Scenario | FG% | P(win home) | P(win away) | P(comeback) | 90% CI |
|---|---|---|---|---|---|
| Finals avg current | 41% | 3.0% | 10.8% | ~0% | [0%, 13%] |
| Midpoint improvement scenario | 45% | 27.4% | 60.5% | 12.8% | [0%, 81%] |
| Other playoff win avg baseline | 48% | 82.2% | 95.0% | 90.8% | [10%, 100%] |
Observed averages from the SportRadar box scores — descriptive statistics, not model coefficients. The logistic regression assigns highest posterior confidence to assists and rebounds as predictors of winning.
Assists
27
playoff win avg
19 vs NYK · ↓ 30%
FG%
48%
playoff win avg
42% vs NYK · ↓ 6 pts
Rebounds
50
playoff win avg
48 vs NYK · roughly on par
Turnovers
12
playoff win avg
15 vs NYK · ↑ 3 per game
Simulated from 5,000 browser-side draws using beta-distributed posteriors derived from the Python MCMC (10,000 post-warmup draws). The distribution is right-skewed, not bell-shaped — most probability mass sits near zero under current Finals performance. The midpoint scenario spreads widely because only two Finals games have been played.
Bayesian hierarchical logistic regression with two partial-pooling groups: the NYK Finals (n=2) and all other 2026 playoff games (n=18).
Fit via adaptive Metropolis-Hastings, 10,000 post-warmup draws. Predictors: FG%, rebounds, assists, steals, blocks, turnovers, home/away indicator — all standardised.
The wide credible intervals are not a modeling weakness. They honestly reflect irreducible uncertainty from n=2 in the focal group. A well-calibrated Bayesian model should produce wide intervals when data is sparse.
Every team to erase a 0-2 Finals deficit — 5 comebacks in NBA history. historical base rate ≈ 13%
1969 · Boston Celtics
Bill Russell's last ride: down 0-2 to the Lakers, won it in Game 7.
1977 · Portland Trail Blazers
Bill Walton's Blazers reeled off four straight after dropping the opener pair.
2006 · Miami Heat
Dwyane Wade caught fire; Miami answered with four in a row.
2016 · Cleveland Cavaliers
LeBron's Cavs — the only club to escape both a 0-2 and a 3-1 Finals hole.
2021 · Milwaukee Bucks
Giannis' 50-point close-out capped the comeback over the Suns.