Ken Pomeroy, the godfather of college basketball analytics, taught the sports analytics community about the predictive power of the preseason AP poll. Let’s explore how these ideas apply to the NFL playoffs.
From 1985 through 2022, Ken noted that 36 teams had entered the NCAA tournament as a 1 or 2 seed that weren’t ranked in the preseason AP poll. These teams had zero Final Four appearances. Zero.
This result for the preseason AP poll relies on the wisdom of crowds. No one sports writer submits a perfect ranking before the start of the college basketball season. However, the combination of many sports writers leads to a powerful predictor of team strength.
It seems crazy that a preseason poll with zero access to data from the regular season can have anything to say about post season performance. It seems even crazier that the ideas apply to the NFL. Let me suggest otherwise.
Each season, I put together an NFL preseason AP poll by gathering the subjective power rankings of sports writers. I use my team based analytics to turn these subjective rankings into a rating for each team.
For example, my preseason wisdom of crowds model had Buffalo 3.1 points better than NFL average. The Bills were the 9th best team before the season.
This stupid model is a remarkable predictor against the NFL market. For the playoffs in the 2014 through 2023 seasons, this preseason wisdom of crowds model is 52.9% against the spread (46-41 with 2 pushes). I find this result shocking, even if it’s a small sample size.
This post season result is consistent with what Ken found in college basketball. I haven’t done the same study on the top 2 NFL seeds in each conference that Ken did for the NCAA tournament. But for fun, let’s assume the same ideas apply.
Minnesota had the same 14-3 record as Philadelphia, the 2 seed in the NFC. Head coach Kevin O’Connell has done a remarkable job with QB Sam Darnold, and the Minnesota defense under Brian Flores has also been great.
However, Minnesota started the season 23rd in my wisdom of crowds preseason model, more than 3 points worse than NFL average. My equivalent of the preseason AP poll had this team bottom 10 in the NFL.
The NFL market strongly disagrees with this preseason assessment. My market model takes closing points spreads and adjust for opposition. This powerful predictor has Minnesota 4 points better than NFL average.
This market model considers starting QB, so it won’t include last week’s game with Jimmy Garoppolo for the Rams. With Matt Stafford, the market model has the Rams 1.8 points better than average.
Due to the fires in Los Angeles, the Rams will host the Vikings in Arizona. The market model favors Minnesota by 2.2 points at a neutral site, close to the market that favors Minnesota by 2.5 points.
All my other metrics are more bullish on the Los Angeles Rams. In particular, the preseason wisdom of crowds model has the Rams by 4.3 points. My member prediction has the Rams by 1.1 points, and I’d lean Rams +2.5.
Even if Minnesota moves past the Los Angeles Rams, the work of Ken Pomeroy suggests that Vikings will not make the Super Bowl.
Football betting with a PhD edge
This article was sent to The Power Rank’s email newsletter on Sunday, January 12, 2025 during Wild Card Playoff weekend of the 2024 season. This newsletter is a free service that strives to be:
- Valuable
- Concise
- Entertaining
The predictions are based on my PhD research from Stanford, and the content covers college football and the NFL.
To get this free service delivered straight into your inbox, enter your best email address and click on “Sign up now!”
