Simulating the NFL Playoffs: Super Bowl Futures

Simulating the NFL Playoffs: Super Bowl Futures
Dan Rubin
January 8, 2021
Hand holding Super Bowl Trophy 2021
Photo By Patrick Smith / Getty Images

After a full 256 game NFL season, the playoffs have finally arrived. Unique from other major sports, the NFL playoffs do not feature a traditional bracket. Instead, the NFL is structured so that the highest seed in each conference plays the lowest remaining seed in the same conference. For example – as the #1 seed in the NFC, the Green Bay Packers will face the lowest remaining seed after this weekend. Depending how the games shake out, this could be the Los Angeles Rams, Washington or Tampa Bay. The other two remaining teams in the NFC will play each other.

To accommodate for this added complexity and assess the likelihood of each team winning the Super Bowl, we turn to the Monte Carlo simulation. Monte Carlo simulations are an excellent tool to assess the likelihood of events happening when you are faced with a series of complex interactions or path dependent scenarios.

To estimate the probability of each team winning the Super Bowl, we simulated the playoffs 10,000 times using our Power Rankings. (Note: in addition to our Power Rankings, we rely on empirical data to convert the Power Rankings into a win probability. This is a topic for another day, but we would be remiss if we did not mention this.)

So – who do we like to win the Super Bowl? Although we have the Chiefs as one of the favorites, we are not nearly as high on them as the betting markets. We give them a 20% chance (+397) of winning while many sportsbooks have them around +200 (which requires a 33.3% win probability to break even).

So who do we like? Well – generally we see value on the Saints (widely available at +800) and Ravens (widely available at +1000) who (unsurprisingly) rank #1 and #2 in our Power Rankings. If you’re in the mood for a long shot, you could do worse than the Colts (available at +5000). We’ve converted our estimated win probabilities into our fair odds below:

How strong is this model? It’s important to note that the above percentages represent our raw predictions and don’t include any market data. While our EPA model has been strong thus far, no model is perfect, and every model can benefit from weighing the market odds. The team at Data Golf does a fantastic job of explaining why it’s critical to weight the market odds.

How should we bet these? SHOP. SHOP. SHOP. In futures markets, it is extremely important to get the best available line. THIS CANNOT BE OVERSTATED. Futures markets have notoriously high vig – so we must chip away at this vig by getting the best odds available. Case in point: across nine offshore sportsbooks, the average hold for Super Bowl futures is 17.4%. The synthetic hold for the best line at these nine sportsbooks: 6.9%. Add in the best available lines at the legal books in Colorado and the synthetic hold drops to 0.7%!

How much should we bet? Although we’re confident in our numbers, longshots of this nature only comprise a small amount of our portfolio. As a result, it’s unlikely we’ll even approach 0.5% of our bankroll on any bet.

Other Related Blog Posts

Receive our research, bets, and analysis