NBA Playoff Odds API — Monte Carlo Simulator & Value Bets

Replays every remaining NBA game thousands of times and returns, for all 30 teams, the probability of making the playoffs, of falling into the play-in tournament, of winning a division, a conference and the title — then prices each of those against live Kalshi contracts and reports the edge, the expected value net of fees, and a suggested position size.

Sports data, simulation, statistics, prediction markets, betting odds, basketball.

---

The one thing to get right about the NBA

In baseball, football and hockey, "make the playoffs" is a single threshold and Kalshi lists a single market. Basketball is different, and it is the most common way to get this wrong:

Finish in your conference	What it means	Kalshi market
1st – 6th	In the playoffs, no questions asked	KXNBAPLAYOFF-27
7th – 10th	Into the play-in tournament, where only two of the four survive	KXNBAPLAYIN-27EAST / KXNBAPLAYIN-27WEST
11th – 15th	Season over	—

Kalshi says it in the contract rules: "Qualifying for the play-in tournament doesn't constitute playoff qualification."

So the play-in market is not a weaker version of the playoff market — it is a band, and the two are almost disjoint. The best team in a conference is a near lock for the playoffs and a near-zero for the play-in. A 45-win team is the reverse. That is why this Actor simulates the play-in games themselves (7 v 8 for the seventh seed; the loser then hosts the winner of 9 v 10 for the eighth) instead of guessing, and why it reports fairPlayoffProbability and fairPlayInProbability as two separate columns that are compared against two separate markets.

What it does

1. Pulls current standings from ESPN, including points scored and allowed.
2. Rates every team from its point differential — Pythagorean expectation with the basketball exponent of 13.91 — blended with its actual record and regressed toward a prior built from last season.
3. Simulates each remaining game from the ESPN schedule with home court applied in log-odds, redrawing every team's true strength once per simulated season so the output is a distribution rather than a single confident guess.
4. Seeds each conference, runs the play-in, then runs the full four-round bracket with best-of-seven series and the 2-2-1-1-1 home court pattern.
5. Fetches live Kalshi prices, matches every team to its contract, and computes edge, expected value net of fees, and a quarter-Kelly stake capped per position and in total.

Output

One row per team. Highlights:

Field	Meaning
projectedWins	Mean win total across all simulations
averageSeed	Mean finishing seed within the conference
fairPlayoffProbability	Reaches the playoffs — top six, or survives the play-in
fairTopSixProbability	Avoids the play-in entirely
fairPlayInProbability	Finishes 7th to 10th
fairDivisionProbability	Wins its division
fairConferenceProbability	Reaches the Finals
fairChampionshipProbability	Wins the title
marketPrice, edge, expectedValuePerContract	The contract and what it is worth
call	VALUE, PASS or WATCH
suggestedContracts, suggestedStake	Quarter-Kelly sizing, if a bankroll is set

Four dataset views are provided: Playoff overview, Market edge, Play-in race and Title odds.

Preseason honesty

Before opening night this model knows exactly one thing: how last season ended. It has not seen free agency, the draft, a trade, or an injury. Its largest disagreements with the market are therefore not edges — they are the summer.

Two safeguards make that explicit rather than leaving you to discover it:

• minGamesPlayedForValue (default 10). Until every team has played that many games, no row may call itself VALUE. The edge is still reported in full; every row is simply labelled WATCH.
• strengthUncertainty (default 0.22). Each simulated season redraws every team's rating. Set it to zero and the Actor will happily tell you a team makes the playoffs 100% of the time, which is never true in September.

Run against real Kalshi prices in September 2026 the model's rank correlation with the market was about 0.80, with a mean absolute difference of 13 points. The largest gaps were teams whose entire case rests on the offseason — exactly what the gate is there for.

Notes on the data

• ESPN names a season by the year it ends: 2026-27 is season=2027.
• Before opening night ESPN publishes the new division tree with no teams in it. The Actor falls back to last season for the club list and the conference map, with records zeroed.
• ESPN lists 80 of the 82 games until the NBA Cup is decided. The missing games are simulated against an average opponent so win totals stay on an 82-game scale rather than quietly projecting an 80-game season.
• Kalshi's public read API needs no key. Prices come from yes_bid_dollars / yes_ask_dollars; both sides of every contract are priced and the better one is used, so an overpriced favourite shows up as a chance to sell rather than as no signal at all.

Related Actors

Same engine, other leagues: MLB Playoff Odds, NFL Playoff Odds, NHL Playoff Odds, Football (Soccer) Monte Carlo Predictor, and the Sports Probabilities MCP Server that exposes all of them to AI agents.

Disclaimer

Statistical simulation for research and analysis. Not betting advice, and not affiliated with the NBA, ESPN or Kalshi.