Super Bowl teams show analytics gaining ground in NFL

While still nascent compared with Major League Baseball, the use of analytics to inform decisions is gaining momentum in the NFL.

In baseball, fans such as author Bill James and those who started the Society for American Baseball Research in the 1970s were the first to look beyond basic statistics such as home runs and batting averages to try to better understand the true value of players. Pro teams finally began to follow suit in the late 1990s and early 2000s, and eventually analytics use became widespread in baseball.

Similarly, fan groups such as Football Outsiders in the early 2000s were the first to look beyond basic football statistics before NFL teams began to catch on in the late 2010s. Now, while not all NFL teams have bought in completely, all have analytics departments, and the use of analytics to inform strategic decisions is growing.

In fact, it was on full display during the AFC Championship Game between the Kansas City Chiefs and Buffalo Bills, when the Chiefs repeatedly stopped the Bills on fourth-down attempts having studied the Bills' tendencies on such plays.

Similar situations are likely to occur this coming Sunday, when the Chiefs play the Philadelphia Eagles in Super Bowl LIX.

Not surprisingly, both the Eagles and Chiefs rely heavily on analytics, according to Peter Zaimes, a lecturer at the University of New Hampshire (UNH). He's also founder of the university's Sports Analytics Lab, where students work with real-world sports clients to learn how analytics are applied in sports.

Among UNH's clients are baseball's Baltimore Orioles, baseball player development program 603evo and UNH's own athletics teams.

As an expert in sports analytics, Zaines has studied the growing use of advanced statistical analysis in pro football. In a recent interview, he discussed the evolution of analytics in the NFL, situations where statistical analysis most frequently comes into play and how analytics is most likely to be used in the Super Bowl, on the field by the Eagles and Chiefs, as well as fans wanting to place a wager or two.

In addition, Zaines spoke about the advantage analytics can give a team over another that doesn't use analytics, and how NFL football could look once all teams buy in, especially with the deployment of AI and machine learning tools.

Peter Zaimes

We began to hear about analytics in baseball in the late 1990s and early 2000s; when did analytics start to creep into the NFL?

Peter Zaimes: It's not as well-known, but it started in the early 2000s in the NFL as well, but on a much smaller scale.

There's a group called Football Outsiders that was coming up with some proprietary measures such as defense-adjusted value over average, which examines the down and distance of each play and calculates how much more or less successful each team is compared to the league average. There was also defense-adjusted yards above replacement, which measures players compared to the league average. But no teams were using them. The analytics geekery, if you will, was into it, but it didn't get picked up by the mainstream until more recently.

If there were groups like Football Outsiders made up of non-NFL personnel using advanced statistics to analyze football 20-25 years ago, when did NFL teams themselves start using analytics?

Zaimes: They got into it in the late 2010s.

If you recall the Super Bowl between the Patriots and Eagles in 2018, the Eagles ran what was called the Philly Special, where quarterback Nick Foles caught the ball for a touchdown on a trick play. That was, according to experts at MIT's Sloan Sports Analytics Conference, the first analytics-driven Super Bowl. The play was called down to the sideline by coaches from above. They said the play was going to work based on the Patriots' defensive formation. The data told the Eagles that the defensive formation was susceptible to a play to the offense's right side, and that play would work.

After that, analytics use started going crazy, just like in every other sport.

In baseball, analytics quickly became associated with certain teams, most prominently the Oakland A's. In football, were there teams that were early adopters in the same way?

Now, there's not a team that doesn't have an analytics department, but teams still invest in it much differently.

Zaimes: There were some teams that bought in early. The Ravens are one, the Bills are another. The Browns, of all teams, and the Eagles are another. Those are the teams that have the most robust staffs.

Football is a sport where a lot of the teams are led by ex-players. They are bulldogs and want to be gut-feel people who know the game. As a result, there was a lot of resistance. In other cases, some teams were just late to the game and didn't want to deal with analytics. Now, there's not a team that doesn't have an analytics department, but teams still invest in it much differently.

In baseball, it's pretty universal, although teams do different things with analytics.

You mentioned that one of this year's Super Bowl teams, the Eagles, has been at the forefront of analytics use among NFL teams. What about the Chiefs?

Zaimes: Just looking at Chiefs coach Andy Reid and what he does, he would in theory classify as a gut-feel guy. But actually, his analytics department outsmarted the 49ers in last year's Super Bowl when they knew that overtime rules in the playoffs are different from the regular season. The 49ers made the wrong decision by taking the ball instead of deferring.

The Chiefs definitely have a robust analytics department, and Reid has come around on analytics. He's still, I think, a gut-feel guy, but the Chiefs are definitely heading in the direction [of using analytics to determine strategy], and Reid has warmed to it big time.

In MLB, analytics has led to fundamental changes, such as the reliance on power over speed and reduced emphasis on starting pitching. Are there fundamental on-field changes happening in the NFL that are analytics-driven?

Zaimes: The biggest thing we all have seen in the last few years is teams going for it more on fourth down instead of punting. Every down and distance, factoring in where a team is on the field, is studied and there's enough data now to analyze whether it's advisable to go for it on fourth-and-1 or fourth-and-5 from a particular spot on the field. That's the most noticeable example of analytics taking over. The other thing that comes in is which plays are most likely to be successful based on the down-and-distance scenario. A lot of that comes from analyzing the other team.

You also see it with 2-point conversions, when to do it and when not to do it. It used to just be based on how many points a team is up or down. Now, it's more like a chess game where teams are looking two or three possessions ahead to decide whether to go for two. What plays to run also comes into it. It's easy to say, 'Let's go for it,' but analytics are really helping with what play should be run given the defensive tendencies.

In terms of what we'll be watching on Sunday in the Super Bowl, what is a key situation that could arise when analytics will potentially play a significant role in a decision?

Zaimes: Clock management in the last two minutes of halves, especially if the game is close late in the fourth quarter -- what plays to run based on the situation, when to get into a hurry-up offense, what plays will work against the other team's base defense. The Eagles and Chiefs are both up-tempo teams. When you think of Chiefs quarterback Patrick Mahomes, he can move the ball up the field in 15 seconds to get the ball in field-goal range, so think of him having to do that at the end of the half or the end of the game. Analytics will be huge in that facet.

The game could be a blowout and clock management never becomes a thing, but it will if it's a close game.

Gambling is a huge part of fans' enjoyment of the Super Bowl, and people are able to bet on everything from the outcome of the game itself to how long it will take to sing the national anthem. How can fans use analytics to inform the bets they make?

Zaimes: What's available are some of the public data. Fans can grab things like the defense-adjusted yards above replacement, calculations on some of the individual players. People can look at that and get a better understanding of prop bets such as who is most likely to score a couple of touchdowns or run for over a certain number of yards. Those numbers are out there. It doesn't mean something will happen, but those numbers are out there that can help make predictions.

People go on gambling sites thinking that Eagles running back Saquon Barkley is great and is going to score however many touchdowns. Maybe he will. But people should also look at some of the metrics that are out there and adjust their decision-making based on publicly available data about that player against the defense he'll be playing. Inform yourself. It doesn't mean it will come true, but at least it's another level of information.

What is a recent example of a game that was affected by analytics, when one team made a key decision or gained an advantage because of its embrace of advanced analysis?

Zaimes: Let's go back to the AFC Championship when Buffalo lost that game, going for it several times on fourth down and not converting. The Bills were doing their version of the "tush push" [when the quarterback keeps the ball and teammates try to push him forward for a first down]. But the defense knew that when Buffalo does its version of the "tush push" quarterback, Josh Allen likes to go a certain direction to get that yard or half a yard. The Chiefs stacked the line where Josh Allen likes to go and stopped him each time.

The Bills used analytics to call the "tush push" in that short-yardage situation, and the Chiefs countered with analytics that revealed the Bills' tendencies. They didn't just stack the line. They knew Josh Allen goes left, so they put more people to the left and Josh Allen didn't adjust. It was very interesting.

Is there a way to quantify the advantage analytics gives one team over another that doesn't rely on analytics to drive decisions -- could it equate to a field goal difference, a touchdown, anything like that?

Zaimes: I don't know if I've ever thought about it that way.

Looking at who invests most in analytics, it's a mixed bag in terms of winning percentage. The Ravens and the Bills are teams with winning cultures, but then there's the Browns, who don't have a winning culture but haven't been terrible the last few years. There is a statistically significant relationship in investing in analytics and how you do, but there's also some luck in terms of who you have at quarterback or injuries.

It could equate to a field-goal difference. I definitely wouldn't say a touchdown. But in the NFL, if analytics gives a team a field-goal advantage each game, that could be worth five wins. Look at all the one-score margins in the NFL and the differences between playoff teams and non-playoff teams separated by one win.

What about off the field? Are there ways analytics is changing how NFL teams assemble talent?

Zaimes: It's definitely informing roster construction. But where football teams are really investing is in wearable technology and what goes into workload management so teams can try to avoid injuries by shutting players down. They can see that players are overexerting during practice and a soft-tissue injury is more likely than it otherwise would have been. That's where football is really headed. We're doing work with the UNH football team on exactly that. We're able to dashboard those types of metrics.

Both teams are relatively healthy going into the Super Bowl -- no one is fully healthy at this point in the season -- but no stars are going to be missing. That is a huge thing for the NFL. Wearables, a lot of which is AI-driven, is really the off-the-field focus.

Going forward, how do you see analytics continuing to change football -- will there be greater adoption, and what might be some of the results in terms of what we see on the field?

Zaimes: Let's go to the extreme and introduce AI to football. With the processing power and machine learning and predictability that comes with it, a team can get a fully integrated, hyper-personalized strategy tailored to exploit weaknesses. There are all kinds of coaching advantages that can result. But if everyone does it, the strategy will become standardized. If all the data is there for everyone to use and every analytics department is the same, there will be less diverse play styles with everyone doing the same thing, which could take some of the fun out of the game. That would take a major cultural shift, but why wouldn't it eventually go there if we're developing all these technologies?

We're sort of heading there, but I don't know if it will ever fully get there.

Is that where baseball has gotten, with all teams emphasizing the same things?

Zaimes: I think baseball is there. I'm a baseball-first fan, coached my kids, and I don't like the game as much. I still like it and watch it, but I don't like where it has gone. And it's to the detriment of players. The spin rate on pitches is so emphasized, they're spinning the ball so much it has led to more elbow injuries. There are no more Vince Colemans [who don't hit for power but steal a lot of bases]. It's a homogeneous game now.

Baseball is trying to get in front of that by introducing the pitch clock, eliminating shifts and other rules. The end result in football is that all the offenses and defenses will be the same. That's the extreme. But the difference in football is that there are way more facets in football than baseball where it's throw ball, hit ball, catch ball. In football, it's hand-to-hand across the field.

Editor's note: This Q&A has been edited for clarity and conciseness.

Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than 25 years of experience. He covers analytics and data management.