Trader's View | Think all basketball is the same? Think again.

Did you know, Genius Sports runs individual trading models for each of the NCAA, FIBA and NBA formats? In this edition of Trader’s View, our VP – Product, Thomas Holland, explains exactly why – including the powerful impact on basketball pricing accuracy and delivers an additional 12% in-play uptime

If you’re trading basketball, you already know the NBA isn’t the same as NCAA or FIBA. Different tempos, foul dynamics and scoring rhythms. But what matters more than knowing those differences is how your models respond to them.

At Genius Sports, we run individual trading models for each format, NBA, NCAA and FIBA. Not for the sake of detail, but because it makes a real difference to pricing accuracy and in-play performance.

Take the NBA Finals decider. Oakland City Thunder edged it 4-3, and like most tight matchups, the momentum swung wildly across the series. That’s normal in pro basketball. But when you zoom out across the full calendar, including NCAA and FIBA, we’re talking 45,000 fixtures a year, each with its own trading challenges. That’s why we don’t believe in generic solutions.

We’ve built bespoke models for each competition, because you can’t trade FIBA like it’s the NBA. And the results speak for themselves: since deploying our Monte Carlo-simulated NBA model earlier this season, we’ve achieved over 12% additional uptime, while the model has seen a 23% uplift in pricing accuracy across core markets.

Now, we’ve taken that a step further by integrating new logic to support razor-sharp Same Game Parlay (SGPs) pricing across all three formats. SGPs and betbuilder are only growing in significance, with DraftKings reporting SGP handle growth of 40% year-on-year.

Whether it’s totals, spreads, or the growing complexity of SGPs, our models bring confidence to your trading desk, especially when the game goes off-script.

College basketball vs NBA vs FIBA

The NCAA’s two 20-minute halves, the NBA’s four 12-minute quarters, and FIBA’s 10-minute quarters only scratch the surface of what sets these competitions apart.

Each league’s unique pace, scoring patterns, and late-game tactics create a “multiplier effect” on pricing every event. That’s why league-specific logic isn’t just a nice-to-have – it’s essential for accuracy at scale.

We all know NCAA, FIBA, and NBA games differ in scoring around 150, under 200, and 220+ points respectively, which is why mixing those into one model risks mispricing “Total Points” markets, especially late in close games.

But the differences go deeper still: NCAA teams shoot fewer threes, FIBA offences play more tactically with fewer possessions, and NBA games are faster and more fluid.

Our models are designed to account for even the subtle differences.

One example specific to our NCAA model: the “One-and-One” bonus rule that’s entirely unique. No other professional basketball league uses it. In a tight March Madness game with 90 seconds left – say a lower seed clinging to a lead against a favourite – it has a massive impact on pricing free-throw scenarios. Our model doesn’t simply calculate the probability of making the first free throw. It’s built to factor in the chance of a second attempt occurring.

With our new agreement to supply official NCAA data to licensed sportsbooks for March Madness and the full postseason, the granularity of our data collection alongside our NCAA model becomes even more valuable.

The “Blowout Factor”

Let’s talk about a common trading headache in basketball: blowouts, where one team wins by a large margin, and the outcome is never in doubt.

When Jokic gets pulled with eight minutes left in the fourth, because Denver is up by 20 points, our NBA model is trained to automatically adjust probabilities across key market-types. Player points and assists markets are adjusted, while team totals shift as the pace of game slows, and handicap markets change as rotation players get extended minutes.

This isn’t just about clever substitution logic. The “blowout effect” impacts pricing for all markets, in some capacity. Handicap markets must account for the changes to spread markets. In addition, team total markets need recalculating because points expectancies change, as the game pace shifts. Lines for player props are highly-affected when starting players see reduced minutes, and bench players suddenly play a key role.

We’ve designed our models to account for blowout thresholds and substitution patterns, to keep pricing accurate when games become one-sided. This is crucial in reducing the chance your sportsbook is left exposed to sharp money by keeping your pricing ahead of the game.

The intentional foul game: where our models get smart

It’s no secret that the live match state is key to in-play trading. In basketball, the final moments of the game are especially vital. If a team is down eight, with two minutes remaining, they might stretch the game with timeouts and fouls, turning a seemingly finished contest into up to 10 extra possessions.

Our models are built to anticipate these situations. They recognise fouling patterns by the losing team when they need to stop the clock. Our logic also understands the “up-three” fouling strategy. This is when a team leading by three points in the final seconds might intentionally foul to prevent a game-tying three-point attempt.

Our models analyse factors such as the frequency of fouls, timing relative to the game clock, and score differential to distinguish deliberate fouling sequences from regular gameplay. This behaviour is embedded within our league-specific logic, because intentional fouling plays out differently across NBA, NCAA, and FIBA contexts.

Monte Carlo = 12%+ more basketball uptime

So how do we make all this league-specific intelligence work at scale? That’s where our Monte Carlo simulation and Edge pricing come into play.

In the previous edition of Trader’s View, I wrote about the rollout of Monte Carlo-simulated trading models.

Since deploying Monte Carlo-simulated trading models to the NBA earlier in the season, the gains have been clear: over 12% additional uptime during the playoffs alone. All our basketball models now simulates thousands of potential outcomes per second, reacting instantly to game state, score, possession data, and more.

Previously, volatile moments like 4th Quarter Totals or Race to 20 needed manual intervention. Now, our model handles those swings automatically across all three formats, which is backed up by the data shown by a 23% uplift in pricing accuracy across core market, letting traders focus on reading game flow, spotting fatigue, and tracking momentum shifts.

And when we combine these format-specific models with Edge, our automated liability pricing tool, the dividends multiply. Edge automatically factors liabilities across correlated bet types and makes incremental adjustments to offset sportsbook customers’ liabilities.

Take a FIBA Champions League example: if a SGP backs Unicaja’s star player to score over 17.5 points and the team to win by 10+, our model continuously simulates the likelihood of these outcomes and how they impact each other. Meanwhile, Edge dynamically optimises pricing by balancing these dependencies against the customer’s unique liabilities and how they correlate.

In short, our FIBA model reads the game, while Edge automatically manages liability across both legs.

Closing thoughts

Here’s what gets me excited about where we’re heading: we’re not just building better basketball models – we’re fundamentally changing how the game gets traded.

The Monte Carlo simulation gives us unprecedented real-time accuracy. Edge pricing solves the SGP correlation puzzle, and our league-specific logic ensures we’re not just throwing generic solutions at nuanced problems.

But the real breakthrough is how these innovations work together. When you combine simulation-driven predictions with automated liability management across format-specific models, you get something that didn’t exist before: true basketball trading intelligence that scales.

Not all basketball is the same. And now, neither are the models that trade it.