NBA Win Probability Explained
How we predict basketball games (and why the NBA is uniquely predictable)
Why Basketball is Different
The NBA is the most predictable major sport. More possessions per game means less randomness. A bad team can steal an NFL game with one lucky play. In basketball, you need to outplay someone for 48 minutes across 100+ possessions. Fluke wins are rarer.
That's why our NBA predictions tend to be more reliable than other sports. When we say the Celtics have a 72% chance, we're working with a lot more signal than noise.
The Factors That Actually Matter
Rest and Scheduling
The NBA schedule is brutal—82 games, constant travel, back-to-backs sprinkled throughout. Teams on the second night of a back-to-back win about 4% less often. Teams with 2+ days rest facing a tired opponent? That's a real edge.
We track rest days for both teams in every game. A fully rested team at home against opponents finishing a road trip? That matters more than you might think.
Home Court Advantage
NBA home teams win around 55-57% of games historically. It's real, but it's smaller than other sports. Some buildings matter more—Utah's altitude, Miami's late-night tipoffs for East Coast teams. We account for venue-specific effects.
Efficiency Ratings
Points per possession is the currency of NBA analytics. Offensive rating, defensive rating, net rating—these tell you how good a team actually is, stripped of pace. A team that scores 120 in a high-pace game isn't necessarily better than one scoring 105 in a grind-it-out affair.
We look at recent efficiency numbers, adjusting for opponent strength. A team that just held the #1 offense to 100 points is playing better defense than their season average suggests.
Injuries
Stars matter more in basketball than any other team sport. LeBron, Giannis, Jokic— these players swing games by 8-10 points just by being on the court. Missing your star shifts win probability dramatically.
We also track role player availability. Losing your starting center against Embiid is different from missing him against a small-ball team.
What We're Watching For
Pace mismatches tell a story. When a run-and-gun team faces a half-court grinder, who controls tempo usually wins.
Three-point varianceis basketball's biggest wildcard. Teams can go 8-for-35 one night and 16-for-35 the next with the same looks. We can't predict hot shooting, but we know which teams rely on it.
Motivation differencesmatter late in the season. A playoff team resting starters against a team fighting for the play-in? The standings don't capture that.
Reading Our Breakdowns
Every game page shows you which factors favor which team. When you see "REST: HIGH impact favoring BOS"—that's telling you the Celtics have a significant schedule advantage.
Stack multiple factors in one direction, and you get lopsided probabilities. Home court + rest advantage + injury to the opponent's star = a team we're pretty confident about.
Related Topics
Frequently Asked Questions
Basketball is more predictable than football—more games, more possessions, smaller variance. Our model hits around 65% on games we call 55/45 or better. When we show a team at 70%+, they win about 73% of those games.
An 82-game season with constant travel takes a toll. Teams on the second night of a back-to-back win about 4% less often than rested teams. We track rest advantages for every game.
We can't predict when a star will sit out for rest. Once it's announced, we update. But if Kawhi is listed as playing on a back-to-back, we factor in that he might not play his usual minutes.
Playoff basketball is different—tighter rotations, more preparation, home court matters more. We adjust our model for postseason games. Regular season numbers don't translate directly.
Sources
- Basketball Reference - NBA team statistics, efficiency ratings, and historical data
- NBA.com Stats - Official league statistics and advanced metrics
- ESPN - Research on back-to-back game impact and rest advantages