With eight seconds left on the play clock, Harrison Butker of the Kansas City Chiefs kicked a chip shot field goal that put the nail in the coffin for the Philadelphia Eagles. By most quantitative measures, the Eagles were flying high all afternoon, but conceding a costly scoop-and-score defensive touchdown in the first half kept the Chiefs within striking distance. Ultimately, though, it was a qualitative measure that ruined the night for the Eagles—don’t give Patrick Mahomes the football with a one-score lead late in the fourth quarter of an important game. He will punish you, and he may even bundle your insurance in the process. That Mahomes drive to set up the Butker field goal marked a second consecutive model defeat for Allspring’s Super Bowl prediction. The silver lining? We’ve never missed three in a row.
For the fourth time in the past five years, the Kansas City Chiefs are Super Bowl bound, and it will be a rematch of Super Bowl LIV against the San Francisco 49ers in Sin City this coming weekend. When these two teams met four years ago, the Chiefs scored 21 unanswered points in the fourth quarter to overcome a 10-point 49er lead—and simultaneously put another mark in the win column for the quants at Allspring. Will our model let it ride this year with another Chiefs prediction, or will the pendulum shift to the other side with a bet on the 49ers?
Every February, we do our best to pretend we know about more than just traditional investments. We do so by quantifying the performance of all 32 NFL teams during the regular season, and we call this metric the “NFL Alpha.” These NFL Alphas (Table 1) are structurally similar to an alpha one might compute for a security in that they represent an asset’s (a team’s) performance relative to market (wagering) expectations.
To help illustrate this concept and add insult to Achilles injury, let’s take a look at the 2023 season for the New York Jets. We’ll assume that for each of the Jets’ 17 games in 2023, a hypothetical bettor placed a $100 wager on the “money line” for the Jets to win. For every game the Jets lost, the bettor would lose $100 and maybe some dignity. If they won, the bettor would collect back the $100 for that game, plus an additional sum based on the team’s win probability per market expectations.
After the bettor’s results for all 17 games were calculated, we added up all of the bettor’s winnings and compared that dollar amount with the $1,700 total hypothetically wagered throughout the season. Anything less than $1,700 implies a negative Alpha, and anything greater is a positive Alpha. For 2023, applying this process to the Jets’ results during the season would have delivered $1,613 to the bettor—a loss of $87, which computes as a -5.1% NFL Alpha (as shown in Table 1 for the Jets’ 2023 Alpha). Compared with an approximate $400,000/minute that the Jets paid for the on-field services of Aaron Rodgers in the 2023 campaign, an $87 loss starts to look like a steal.
There were plenty of questions swirling around the Baltimore Ravens this year. Would Lamar Jackson live up to his contract? Does he actually plan to throw the football to his new receivers? Will the Ravens be able to survive their yearly swath of unwatchable, season-ending leg injuries to their running backs? The Ravens appeared to be up to the challenge. Despite playing in a tough AFC North, they finished the season with an NFL-leading 13-4 record that earned them the #1 seed in the AFC. This is all nice, but analysts will say the only number that really matters is their league-leading 37.1% NFL Alpha (Chart 1).
Farther south, we see an example of a team that didn’t come up with too many answers this year. The Carolina Panthers sold the farm in order to draft 2023’s shiniest rookie quarterback: Bryce Young. A league-worst 2-15 record and -70.7% NFL Alpha may be an indication that they should’ve kept a few more animals around. We look forward to feedback from our colleagues in Charlotte on how “next year will be different,” but what we do know is that Allspring scored the same amount of points (zero) as the Panthers did in their final two games of the season.
The San Francisco 49ers were victims of their own success from an NFL Alpha perspective. After a very strong 2022 showing, the market expected them to pick up right where they left off, as they were favored in all 17 of their games. Turns out it’s actually pretty hard to go undefeated in the NFL. Thus, a 12-5 record was received as a bit of a disappointment, leading to a -2.8% NFL Alpha (Chart 2).
Elsewhere in the NFC West, the Arizona Cardinals were out there upsetting the applecart. Two of the three biggest upsets of the year came at the hands of the Cards. As underdogs in all 17 of their games and bearers of the NFL’s most brutal schedule, they had plenty of opportunity to play the spoiler. Four total wins proved plenty enough to make them a profitable investment for those daring enough to dabble (13.3% NFL Alpha). The question now is whether or not their $230 million dabble into QB Kyler Murray will turn a profit.
No risk it, no biscuit
Sometimes “number go up.” The 2023 NFL season had a slew of upsets that not only disrupted “Survivor Pools” across the nation but turned the nerd-famous “Favorite-Longshot Bias” expectation upside down. To elaborate a bit, research has shown that bets on the Carolina Panthers of the world (got ‘em) tend to underperform alternative bets on actually good football teams (heavy favorites). Ample corollaries of this heuristic exist in markets around the world, but our Systematic Edge Equity team has focused for over 20 years on its application to equities in particular. High-risk (high-beta) equities tend to underperform low-risk (low-beta) equities over the long run. In short, 2023 wasn’t a long-run kind of year. High-risk equities outperformed low-risk equities just as high-risk NFL wagers (22.04%) outperformed their low-risk counterparts (-5.75%)—unless that high-risk team was the Carolina Panthers. Definitely don’t bet on the Carolina Panthers.
Ultimately, the crux of this piece circles back to our now-dusty research showing that NFL teams tend to revert back to their average longer-term success ratios. In particular, teams that outperform expectations in one season tend to underperform expectations in the next (and vice versa). To take this a step further, we’ve found this mean-reversion tendency to be apparent as soon as a season’s playoffs. In other words, teams with higher NFL Alphas often underperform expectations in the postseason, thus making the teams with the lower Alphas relatively undervalued and the better selection to cover the point spread. While this strategy has proven fruitful in the past, the 2023 season has been humbling at best (Table 2). In the 20+ years that we’ve been hoping people will read this piece, our 4-8 record is as bad as it’s ever been—and dragged our overall record down to 126 right-86 wrong (59% correct). Perhaps some mean reversion is on deck for this Super Bowl matchup?
With NFL Alphas of -0.3% and -2.8%, respectively, the Chiefs and the 49ers were the epitome of efficient wagering markets. We’ve never had two Super Bowl contestants so close to a zero Alpha, and we’ve actually never had two teams with negative Alphas, either. It’s certainly counterintuitive to think that both of these teams made it to the Super Bowl yet somehow underperformed market expectations throughout the season.
As much as we’d love to flash some real chutzpah in our Super Bowl prediction, we’ve once again found ourselves with a matchup of thin quantitative margin. Both of these teams largely “were who we thought they were.” At the time of this writing, wagering markets seem to agree, with the 49ers a 2-point favorite in Las Vegas. As the lower-Alpha team (barely) in this bout, we think the 49ers have the edge here as the undervalued team. Put differently, we think the smart money is on Jake Moody (not Butker) of the 49ers to be kicking the game-winning field goal as time expires this coming Sunday. At which point, we’ll gather our belongings, stumble with our betting slip to the casino cashier, and redeem it for a brand-new winning streak.