And this week, the quants at Los Angeles-based firm Analytic Investors picked which team was the best bet to win the game the same way they would typically pick stocks. Essentially, they calculated each team’s “alpha,” or its investment return compared to the wagering odds. In other words, the lower the odds a team had of winning (the underdogs), the more alpha they got when they beat expectations with an upset victory. (With stocks, alpha is the company’s outperformance relative to, say, the S&P 500.)

But predicting the Super Bowl winner isn’t as simple as picking the team with the most alpha—just like investors who bet on Netflix (NFLX) stock this year after it returned a market-besting 134% last year have been disappointed. The Carolina Panthers had the highest alpha in the NFL during the 2015-16 football season, with 61.5%, according to Analytic Investors’ report—an unusually high figure due to the team’s underdog status throughout most of its winning season. The Denver Broncos, meanwhile, had the third highest alpha, with 31.9%.

Here’s where the analysis gets interesting: Analytic Investors, which has used its method to correctly predict the Super Bowl results nine out of the last 12 years, always picks the contender with the lower alpha. Why? It anticipates that high-performing football teams will behave like stocks do and revert to the mean (average performance): “We expect the higher alpha teams in the regular season to be over-valued in comparison to the lower alpha teams in each matchup,” the investors wrote in their report.

In other words, the Carolina Panthers’ high alpha is a sign they’ve been overhyped, “likely leading bettors to overestimate their probability of victory,” according to Analytic. It’s the same reasoning by which investors have recently deemed Netflix stock to be overvalued after its high-flying streak last year.

But even then, Analytic Investors isn’t simply picking Denver because it’s the underdog for Sunday’s game—they aren’t even betting that Denver will definitely win the Super Bowl. They’re simply saying that the Panthers, the favorite to win by six points based on bookmakers’ odds, aren’t as sure a bet as people think—and that Denver has a good chance of closing the gap. “To put it differently, we think the underdog Broncos will lose by less than two field goals, if not win the game outright,” Analytic Investors wrote in its report. “This selection is not based on sentimentality for Peyton Manning, but on the fact that Denver is undervalued from our model’s perspective.”

One caveat about Analytic Investors’ own odds of being right: Their prediction has been wrong the last two years in a row.