In Boston, where anything short of a championship is a failure, the future of sports prediction isn’t coming from instinct — but from algorithms.

For Dr. Robert Kissell, a leader in quantitative modeling and author of “Optimal Sports Math, Statistics, and Fantasy,” a guide to using advanced math to predict sports outcomes, his path to sports analytics didn’t begin in an arena. It began on Wall Street.

“I did a PhD in economics, and I was on Wall Street where I headed algorithmic trading,” Kissell said. “We were developing … statistical models. I became interested in the portability of these models. Can these same models be used to predict sports outcomes?”

That question led to the creation of AlgoSports23, an online platform that applies artificial intelligence to professional and collegiate basketball and football game predictions. The system is built on more than 250 predictive models, ranging from regression analysis to neural networks. Its goal is simple — assign probabilities to wins, losses and margins of victory.

“Our models determine the probability that a team will win [and] the expected score that they’re going to win by,” Kissell said.

Kissell said his predictive system is superior to prior ventures in analytical sports modeling because it emphasizes visibility and accountability.

“Never trust the ‘quant’ when they say theirs is the best,” he said. “You want that transparency.”

The platform allows users to track both past predictions and the odds of future games, comparing results against real outcomes and Las Vegas betting lines.

So far, the numbers are striking. His models have predicted up to 77% of men’s college basketball and 80% of women’s college basketball games correctly, he said. Those numbers outperform Vegas’s predictions.

“We’ve been beating Vegas for the three basketball leagues,” he said, referring to the NBA and men’s and women’s college basketball.

That success becomes especially relevant during March Madness, when people across the country make brackets that have millions of possibilities. While casual fans rely on gut feeling or school loyalty, Kissell’s models search for how a team performs relative to expectations, not just whether they win or lose.

“If you beat a team by more than the expected margin, you move up,” he said. “You could also lose to a team by less than the expected margin. So even though you lost, you did better, and you could move up.”

But Max White, a professor at Boston University who teaches a writing course on American sports and society, said there are inherent shortfalls in any predictive models.

“I trust [data models], but I just think any model is always going to be limited by the unpredictability of sport,” White said.

Reflecting on historic upsets like No. 16 University of Maryland, Baltimore County’s win over No. 1 University of Virginia in 2018’s March Madness tournament, Kissell noted how his model can hint at surprises before they happen.

“That was the first year a 16 [seed team] had a realistic chance of beating a one,” he said. His model predicted Virginia “had a 95% chance of beating UMBC … that’s giving UMBC a one in 20 shot of winning,” — his model’s highest upset odds ever for a 16-seeded team.

For many basketball fans in Boston, the Boston Celtics represent both championship expectations and analytical intrigue. According to Kissell’s current rankings, the Celtics sit among the league’s elite.

For a recent matchup on March 20, 2026 against the Memphis Grizzlies, his model gave Boston a 74% chance of winning, along with a projected margin. The Celtics won by five points, consistent with the model.

But for some fans, like BU sophomore and diehard Boston Sports fan Ava Dunphy, city loyalties outweigh statistics when predicting the outcomes of games.

“If you’re from Boston, you’re mostly a die-hard Boston sports fan,” Dunphy said. “You root for your teams no matter how bad they’re doing.”

Still, Kissell is quick to point out the limits of analytics. Injuries, for example, remain difficult to quantify.

“There’s no easy way to do that,” he said. “The data is not there.”

In high-stakes moments like the later rounds of March Madness, outcomes can begin to resemble coin flips. Even Kissell himself doesn’t claim to outperform his own system.

“I know I can’t beat the algo,” he said. “So I go with the algo.”

In Boston, where sports fandom blurs the line between passion and analysis, that mindset presents a challenge: Do we trust the numbers or trust our gut? Or, as his platform asks users each week, are you smarter than the algorithm?

Your donation will support the student journalists of Boston University. Your contribution will allow us to purchase equipment and cover our annual website hosting costs.