What It's Like To Be... with Dan Heath
Curious what it would be like to walk in someone else’s (work) shoes? Join New York Times bestselling author Dan Heath as he explores the world of work, one profession at a time, and interviews people who love what they do.
What It's Like To Be... with Dan Heath
An NFL Data Analyst
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Forecasting how much scoring would rise under the NFL's new kickoff rule, sifting through game footage to judge the Philadelphia Eagles' "tush push," and presenting to the Competition Committee with Mike Lopez, senior director of data and analytics at the NFL. When should you not follow the data's advice? And what is "EPA"?
GOT A COMMENT OR SUGGESTION? Email us at jobs@whatitslike.com
FOR SPONSORSHIP OPPORTUNITIES: Email us at partnerships@whatitslike.com
WANT TO BE ON THE SHOW? Leave us a voicemail at (919) 213-0456. We’ll ask you to answer two questions:
1. What’s a word or phrase that only someone from your profession would be likely to know and what does it mean?
2. What’s a specific story you tell your friends that happened on the job? It could be funny, sad, anxiety-making, pride-inducing or otherwise.
We can’t respond to every message, but we do listen to all of them! We’ll follow up if it's a good fit.
So I thought maybe we would start with the tush push.
NFL Announcers:Oh, here it comes, Jim, your favorite, the tushy pushy. Meanwhile, the tush push, McKee. Here goes tush push, and they get it. It's not the Jalen Hurts push; it's the tush. No matter who's tush, they're pushing. It should work.
Dan:So I want to talk about your role in kind of looking more carefully at the tush push. But let's just start with the basics. What is the tush push?
Mike:I don't know if I've ever actually used that term.
Dan:Mike Lopez is the senior director of data and analytics at the NFL.
Mike:So, you know, our job at the league office is to analyze the competitiveness of the game and understand the impact of how teams are playing, how coaches are coaching, and things like that.
Dan:The tush push hit his radar a few years back. It was popularized by the Philadelphia Eagles. It starts with a quarterback sneak, and for non-football fans, that's where a quarterback keeps the ball and runs forward to get some short yardage that's really crucial, like maybe you're right at the goal line, or you just need one yard to get a first down. But what's different about the tush push is that one or two of the quarterback's teammates are behind him, shoving him forward.
Mike:And it happened a couple of times, and we noticed, and we started saying, "Well, hey, listen, this is new. There are some folks in the media and out in football who might not like the look of this for sure. Why don't we just track it?"
Dan:Let me just pause you there for a sec. The reason you're getting interested in that is you're thinking, might this be something that we someday would render unfair? Like it's against the rules to do this because if perhaps it gives the offense an unfair advantage? Is that? Sort of the lens?
Mike:Yeah, exactly. If there's a play or a strategy that we deem, you know, sort of inequitable from the sake of maybe it's too hard to stop, maybe it's health and safety risk, maybe it's too hard to officiate, things along those lines. In that case, you know, the question was, was there an unfair advantage to performing that play that would make it too hard for the defense to stop?
Dan:So to answer that question, he turned to the data. What did he find? Stay tuned. I'm Dan Heath, and this is What It's Like To Be. In every episode, we walk in the shoes of someone from a different profession... a tennis coach, a baker, a toy distributor. We want to know what they do all day at work. Today we'll ask Mike Lopez what it's like to be an NFL data analyst. We'll talk about who decides whether a certain rule should change, why data can often be deceptive, and why the NFL is now collecting skeleton-level data on players during games. Stay with us.
It Turns Out Promo:On the It Turns Out podcast, we talk about research that will change how you see the world. An increasing share of the men who are not in the labor force are not unemployed; they're non-employed. In other words, they're not seeking work, and that's a really important difference because it's not just that these are men who like really want a job or looking for a job, but the jobs aren't there. There's actually men who have actually disengaged from the labor market. Great research on money, marriage, longevity, and more. Join us on It Turns Out.
Dan:So Mike and his colleagues are trying to figure out is the tush push an unfair play? The first thing they have to do is define it. What counts as a tush push?
Mike:And incidentally, it's not the easiest definition in the world. What happens if you go to push somebody and you miss them? What happens if the quarterback fumbles a snap?
Dan:And then he has to figure out... did the play succeed?
Mike:For the most part, the success of a quarterback sneak is determined by if you pick up the first down or you score a touchdown. Sometimes teams were doing it out of their own end zone and sort of you know 99 yards from a touchdown, but they wanted to get a couple yards. So you maybe have to change your definition a little bit there.
Dan:But to really understand if the play is unfair, you have to compare it to what the teams would have done otherwise.
Mike:You know, it's not just the the success rate on the play itself is... Well, if you take that away, what are they going to do? What are the alternatives? So you could run a traditional quarterback sneak where nobody's pushing the quarterback. You could get rid of quarterback sneaks entirely and say, what happens if they just run a regular rushing or passing play? So you sort of have some comparison groups, so to speak, and you want to compare the success rate of of that specific play or that specific act to what happens on on those other plays.
Dan:This is such just a good general point about analytics. Your point about base rates, which is let's say the tush push succeeds ninety percent of the time, and you think, oh well, that's unfair. But if a quarterback sneak succeeds ninety one percent of the time, that throws it in a very different light.
Mike:Yeah, that's exactly it. In this specific case, you also had to deal with the reality that the Philadelphia quarterback can squat over five hundred pounds, is an extremely strong player and gifted at this play.
Dan:Right. Oh, that's a great point too. That it's not the play per se; it might be the players.
Mike:Yep. And their their centers over the years have done a real good job of of gaining leverage on the defenders, and so. A lot went into that discussion, trying to to think, you know, from a big picture perspective for the league is is this good for the game? And that's kind of where we came in.
Dan:And and what was the gist of the evidence that you ultimately came back with? Like, did you come out in favor or against, or how did it net out?
Mike:Yeah, we we would never really be in favor or against something per se, but I think our job would be to sort of say, here are the competitive implications of it.
Dan:Yeah.
Mike:And and all three years that we've looked at this, the teams that ran a push sneak had lower success rate than the teams that did not run a push sneak.
Dan:Oh, really?
Mike:And and that included Philadelphia in the year that I want to believe they went something like fifteen for sixteen at it. And so, you know, like you said, base rates of success on this play are typically very high. Teams are running this on a third down or fourth and down, and typically only needing one yard to go. And a lot of times, when you put several players behind the quarterback to push, you really can't run any other play. And so, it kind of keys the defense into exactly what you're doing. And and so, yeah, typically when we've looked at it, is is not shown to be a competitive edge. And I do think that played a little bit of a role in sort of how the league's clubs netted out. In that, you know, they ultimately they did not vote to ban that type of play.
Dan:Gosh, that is such an interesting case study. It helps me to think through like how you approach that and where you got the evidence. And if I can just circle back, one thing I'm curious about is, so you know, beyond the ones that people were complaining about that the Eagles did, where do you go looking for the others? I mean, surely you don't have to watch every moment of every game. Like, how do you know where to even look for the other instances?
Mike:Yeah, the state of football data is in such a good spot, and I would compare it to maybe when I came into the field a decade ago. At this point, there are both internal NFL and then also public, sort of ones that folks, analysts out there publicly can access that have all of our play-by-play data. And within the play description of each play, it identifies who the player was, the type of play, and then you can sort of use the characteristics of the play. The NFL has sort of identifiers that happen on each play, and you can match that to the film. So for us, this would be something fairly straightforward where we would find the plays you know in which a quarterback you know ran the ball directly under center and ran up the middle, and and almost all those are sneaks. And that's not entirely true, but you can filter out the ones that are not by watching film.
Dan:Oh, so you can just go through some kind of master database and like filter? Okay, quarterbacks running the ball up the center, probably some other filter, and then it it gives you what 138 hits or something?
Mike:Yeah. Well, in this case, I think it would give us you know somewhere around 300, which would be the total number of sneaks.
Dan:Okay.
Mike:And then so you're watching 300 plays, and you know adding your labels to that.
Dan:And is that your next move in that situation? Is to actually go and look at the tape for the 300 plays?
Mike:You got it. Yeah.
Dan:Wow. And just to get a feel for your day, like okay, so you run the search. There's 300 total sneaks. Okay, now you've got to go through the list. Is this a kind of one-click watching thing where you can just go click, click, click, and the video auto loads at the right spot, or is it more mechanical than that?
Mike:Yeah, and this would be like one of the examples of something we do each week. So each week we have a variety of things at the league office we're tracking. Push sneak is the one we've talked about, but every time a quarterback slides, we're tracking how far before the defender was arriving did the quarterback slide? Did they slide headfirst? Did they slide feet first? And there's sort of a half dozen examples of things like that, and so after the game, you know, we'll have a set of playlists that come in, and then we'll just go, you know, one by one and watch them. We have, and at this point, most NFL teams have some type of film tool that exists where you can do the filtering and say, say you can filter by, you know, the quarter, the the whether players on offense or defense, you can filter by the down and distance, how far the offense is from the goal line. And so you can get sort of specific cuts that might look at at specific examples in the game that you're looking for.
Dan:Where do these questions originate that you're investigating? Do teams bring them to you, or do you self generate, or how does it work?
Mike:It's a good mix. You know, I think largely our job is to anticipate the questions before they are asked. And so in this example, like. We could have sat around to the end of the year in 2023 and heard discussion about Philadelphia running the push sneak, and then we would have had to review a bunch of film and we had to come up a bunch of answers in short notice. But we're better off sort of anticipating and saying, "Hey, listen, I saw this article, I saw this trend, I saw this discussion. Why don't we track it and have that research ready?" At the same point, you know, we do work for the league office and specifically for football operations and the group that is in charge of the rules. And there's a good number of things that we do because they ask us to. Right? They say, "Hey, can you look into that? Hey, can you look into that?"
Dan:And the things they're looking into are potential changes that would upend rules that have stood for decades. One of the most contentious recent changes concerned the way that players line up on a kickoff.
Mike:Historically, they had lined up at the thirty or thirty-five yard line, and they're kicking off to the receiving team who's lined up at the other end. On the receiving team, blockers all line up, and you sort of have collisions between the return team and the kickoff team. And often, those collisions happened at high speed, and so players are moving fast. They might not see who they're blocking or who's hitting them, and it added up to a play that disproportionately had high injuries. From a league perspective, you know we we care about the competitiveness of the game. We care about the officiating of the game, and, obviously, we care about the health and safety of our players too, and so for several seasons we had tried to make the kickoff play itself safer. Each year, our group was was playing a role in that, identifying the the sort of competitive impact of of different proposals. And beginning two years ago, in the the twenty twenty four season was the first one we had the what's called the dynamic kickoff. And instead of having the players line up thirty, forty, fifty yards apart, now they you watch a game they're they're lining up between five and ten yards apart, and the collisions between players that are five and ten yards apart are certainly less high speed.
Dan:The point being, you don't want a giant human being picking up like thirty or forty yards worth of speed before they hit someone. That was the basic gist of this.
Mike:Yeah, yeah. It's and and those players are tough to block, right? They're physically imposing. So, from our perspective, I think that was probably the. I don't know if it was contentious, but it was one that generated a lot of discussion because you know you're taking a play that had been part of the game for you know over a century, and you're you're changing the fundamental look of the play.
Dan:That's so interesting. And is that still the rule currently? What you changed in 2024?
Mike:Yeah, we changed in 2024. The results from a competitive and officiating and health and safety perspective. We're good enough with our sort of trial year in 2024 that we implemented it for good in 2025.
Dan:And how did you establish what would be success and what would be failure? Like that first year, what were you looking for?
Mike:So when we're looking at a play, we're we're thinking about the implications of both that play itself and then also what it does to the rest of the game. So for example, on the play itself, it's you know, are we okay with the look of it? Right? Are we okay with the look of? Players lining up and waiting for the ball to be caught, and then the you know sort of the collisions happening in the the returner returning the ball.
Dan:So that's almost like an aesthetic question, right?
Mike:It is, yeah, for sure. And that was one of the reasons that people were uncomfortable with it was because of that.
Dan:Huh.
Mike:And then obviously on the play, we are interested in the competitive outcomes of it. You know, from an offensive or return team standpoint, you know, do we get exciting plays? Right? If every single play ends up at the same yard line, then that's not very exciting.
Dan:Right.
Mike:And then you know, from health and safety standpoint, we have you know, what is the rate of concussions? What is the rate of of other injuries, soft tissue injuries, you know, contact injuries, things like that. And then there's the officiating of the play, right? If we have a play that is new for players and coaches, does it lend itself to to many penalty flags? You know, and those are the the play itself. But then there's also the big picture thing is you're changing all sorts of aspects of the play. And specifically on this play, we were changing where the offense got the ball to start the drive. And so typically, you know, in the previous kickoffs, the the return team ended up somewhere around the twenty or twenty-five yard line to start their drive. With this new kickoff offenses were ending up closer to the thirty-yard line, so five yards closer to where they have to score a touchdown. So it's going to lend itself probably to more points, potentially more big play returns, which end up in points. And even you know, this is a very niche thing, but clubs have been very worried about kickers getting injured by tackling. If you add a bunch of kickoff returns and put the kicker in a different position and sort of open up the field a little bit, you might get more kickers trying to tackle the returning player, and maybe that's an injury risk too. So, a lot of different components went into it, and that was a big part of our our role at the league.
Dan:Gosh, that is such a good description, and why it seems like data analytics would be so fiendishly hard is like there are certain things you can measure, like the rate of injuries, and you know which yard line the offense is on when it starts, and then some things you can't, which is just. When you're a football fan and you watch this, does it pass the smell test? You know, and then these kind of unexpected ripple effects. Like, I never would have thought about the downstream effects on the kicker having to tackle the person on the breakaway. So interesting.
Mike:And it's you get from a club perspective why they care because they don't have a backup kicker dressed for games. So if the kicker gets hurt, suddenly there's nobody else to put out there.
Dan:And so, what was the evidence from that first year that convinced you collectively? Hey, this is now a permanent thing.
Mike:Yeah, I think from a competitive and officiating standpoint, the rules and the execution of the play were in line with what we expected. You know, in fact, we predicted to the the league executives like if we do this, we're expecting points to go up by about two points per game. And you know, that's what netted out. I think we went up one point nine points per game.
Dan:That's incredible. I mean, that's kind of blow people's minds that you were able to predict points were going to go up by two points. Like, how did how did you make that prediction?
Mike:I mean, mostly luck.
Dan:Come on.
Mike:Well...
Dan:I don't believe that.
Mike:Yeah, I mean, more more technically, you know, we're data nerds. So, you know, one of our our analysts, Ali, she built a basically a game simulation model that says, Hey, listen, what if we took all the kickoffs that instead of them ending at the twenty to twenty five yard line, what if we ended them at the thirty?
Dan:Okay.
Mike:And then you know, sort of resimulated the rest of the game, and and suddenly you know the points add up. But it's not going to change every game, but it's going to change a couple, and it's going to go up. And so. It was actually a much more complex calculation, but it ended up being somewhat close to the truth.
Dan:Coming up, what kind of data does Mike have access to? You might be surprised. Stay with us.
Sidedoor Promo:Have you ever had the urge to sneak behind the cordoned off areas of a museum, or roam the halls after closing time? The Smithsonian's flagship podcast, Sidedoor, will sneak you behind the scenes of the world's largest museum and research complex. Come learn about the ghosts that supposedly walk the museum halls after dark, how a train robbery gave rise to criminal forensics, why leeches are actually the coolest thing ever, and how to get away with murder in the Arctic. Maybe. You'll discover stories of history, science, art, and culture you won't find in a display case. You can listen to Sidedoor wherever you get your podcasts, or find us online at si.edu/sidedoor.
Dan:What kind of data would people be surprised that you have access to do you think?
Mike:So in 2015, the league began a partnership with Zebra and started what's called our Next Gen Stats Group. And so, beginning in 2016 and in 2017, we began tracking all the players on the field. And what I mean by that is that each player, when they get dressed for the game, their shoulder pads have two RFID chips.
Dan:Oh, really?
Mike:One goes in the right shoulder pad, and the other one goes in the left shoulder pad. And Zebra collects at each stadium. You sort of uses the the frequency and the signal of those chips to identify the player's location on the field. At a frame of ten observations per second, and if you know the two dimensional location of a player, you can get their speed and their acceleration. And so, having access to you know at this point, you know a decades worth of player tracking data, you know as a sort of football analytics junkie is pretty cool.
Dan:And if you think that's wild, check out what's coming next.
Mike:We've been working with a third party Hawkeye, which is sort of owned by Sony. They've installed cameras at at all of our stadiums. And the future of football isn't two-dimensional data; it's three-dimensional data. And so, by three-dimensional, I mean we have the players' x and y, which would be their two-dimensional coordinate. But now we also have their height. And specifically, we have twenty-nine points of a player's pose. And so, you get their wrist, you get their elbow, you get their shoulder.
Dan:And these are like trackers that they're wearing, or how do you get this information?
Mike:Yeah, good question. So it's actually unlike the next-gen stats player tracking, which they are wearing sensors. This is camera-based. And so, Hawkeye has installed cameras at all NFL stadium plus our international games. And those stadiums will take the cameras and generate a best estimate of where each of those players' skeletons are. And this has been done, and it's the same company that does the the ball in or out in tennis. And if you watch baseball games, they'll have you can see the same skeletons moving around a baseball field that now we get on a football field. So,
Dan:And and so once you have you know millions of micro skeleton movements by players, like what's the dream of what you could do with that?
Mike:I think because it's so new, we don't even know how to dream.
Dan:Yeah.
Mike:You know, a real easy example would be if I wanted to look at a defensive end and and what their pass rush move is. Like, take Myles Garrett, right? Like, he is an incredible athlete. He comes off the edge and can kind of do whatever he wants with the offensive tackle and get to the quarterback. But he has three or four common pass rush moves, and with this data, you can now measure the effectiveness of each move. Right? So you could be able to measure and say, Hey, listen, when he does a spin move, how quick is his spin move? When he does a bull rush, right? How strong, how powerful is that? So, if you have good sort of three dimensional pose data, you now can sort of start to answer the questions, like real micro questions about individual skill sets. And we've only had the data for a little bit. We're just sort of starting to see the potential with that. And to be honest, we don't. We haven't even shared it yet with our NFL clubs. They don't have access to it yet. We still at the league office need to think about, you know, what are all the implications of having this data and opening that up. But that's kind of one of the exciting things about about our field.
Dan:What's the biggest risk in interpreting data?
Mike:So, the sport of football has a long history of being somewhat derided in the sports analytics community for not being a good sport for analytics. In other words, in baseball, you have box score data that is real clean and easy for analysts to look at and say, "Hey, listen, home run, good.""Strikeout bad", right, for the hitter. In football, our issue is that there is so much context on a play, and often that context means that when you're looking at it and making the suggestion of a finding, whether it's you know this player is undervalued or this play is a competitive advantage or this trend is good or this idea is bad. A lot of the time, there's actually a very valid football explanation for that. And in other words, if you're presenting to coaches or scouts and you stand up there and you sort of get your chance and you say, "Hey, listen, we should be doing this," the first thing they might say is, "Hey, listen, we'll have you thought about that." And if you hadn't thought about that and you can't account for that in your analysis, then it's not that you're laughed out of the room, but it sort of has much less of an impact than if you can say, "Hey, listen, yeah, actually, we we did that filtering and the finding is still the same."
Dan:So you have to know the game as well as you know data it sounds like. Because you're constantly going to be tested on. You know the the contextual factors that might not be obvious from the numbers.
Mike:Every time somebody asks how to get a job in football analytics, I say make a scatter plot and your x axis is how much do you know football, and your y axis is how much do you know analytics. And teams are after people in the top right. Right? You're after people that really know the game and also really know data.
Dan:I asked Mike about an example where the data was pointing one way, but if you knew football, you would know the data was oversimplified. And he talked about a situation where some smart data scientist had built algorithms that predicted how many yards a punt returner would get after they received the ball.
Mike:And so you get really advanced data science algorithms that say, "Hey, listen, when Dan catches the ball, right? We expect him to get five yards. And if he gets ten, then that's a good thing, right? If he only gets two, that's a bad thing." And so we had one of the examples that they showed was a player, and I won't give the more specifics because because it's a it's a common thing, and this isn't a callout to one the specific group. But you know they had a punt returner back there, and they they said this is the worst decision an NFL punt returner has made this year. And that punt returner caught the ball with like let's call it maybe fifteen expected return yards. And when you looked at the play, it was like yeah, a hundred percent, this player should have returned the ball. But they were showing the broadcast feed, and the team that that called for a fair catch and did not return the ball was up three touchdowns, and there was a minute left in the game. And so it was a good example of you know you you can say sure, my algorithm says, hey, listen, you really need to return the ball because there's a lot of space in front of you. But of course, if you're a club, well, that returner is not returning the ball because you don't want that returner getting hurt. You don't need anything unnecessary in that game. You don't want to turn over. The best thing he could have done was fair catch the ball and not return it.
Dan:That's a great example. Yeah.
Mike:So. That's kind of a constant struggle that we sort of go with in this field. Is you don't want to be up there saying, "Hey, listen, you should be doing this," when there's a context behind why the player chose that.
Dan:I mean, it must be fun as a football fan to be in this role where you can kind of toggle back and forth between your understanding of the game and your love of the game, but also like the hard numbers and what they're telling you.
Mike:Yeah, it's fun. I mean, I grew up. My dad was a high school football coach for forty one years, and you know, I was you know six years old going to his games and and being the water boy, right? So like, I've sort of grown up around this stuff. I would watch film, you know, probably more than was was sort of normal for a high school player, a college player, and always really been into sort of the niche game strategy. And so this is stuff that you know, I hate to say it, but would kind of do for fun.
Dan:Let's talk a little bit about that moment when you're presenting your findings to the league officials. Like, who's in the room for that?
Mike:So we report to a variety of committees. The one that we spend the most time and probably the most important time is is what's called the competition committee. They are a group of of roughly eight to ten combination of coaches, GMs, executives. They are tasked with sort of carrying the shield of the league and sort of best representing the game of football from a league perspective.
Dan:Are they volunteers? Are they drafted into duty? Or...
Mike:yeah, you're sort of if you've reached sort of a level of of success and and also the example that they like to give is you sort of take your club hat off and you you put the league hat on. And so, you know, our job is at the end of the year we we present. Hey, listen, this is the the state of the game, right? This is what's good. This is what's bad. These are trends that we're following. And you know, each year we're tasked with sort of identifying a couple of things that say, "Hey, listen, you know, this is one area that maybe we want to think about getting better at." Right? You know, sometimes we've we've talked about overtime. Sometimes we've talked about our replay system and our timing and how to make that as efficient as possible. So that's kind of the the thing for us. It's sort of our Super Bowl, right?
Dan:And is that stressful for you to have to go in front of these football luminaries and and do your thing?
Mike:I I don't know if stressful is the right word, but it is. You know, it's no different than you know preparing for a big game, right? You have to make sure you've put in the work. You have to make sure that your your stuff is as buttoned up as possible. And you know, I mentioned the hardest part of our job is context. We need to make sure we have the context for everything we're saying and the worst thing that we can do is present something and then have them. Have you thought about this? Have you thought about that? And if you don't have that answer, right, it kind of slows down all the momentum of your work. So, our job is to to sort of have answers to a lot of their questions before they ask them. And if we've done that, then we've we've done our job.
Dan:Mike and I were talking about his proudest moments in the work, and he brought up something that happened back in January 2023. So during a Buffalo Bills and Cincinnati Bengals game, the Bills safety Damar Hamlin made a tackle, stood up, and then collapsed. He went into cardiac arrest and was actually resuscitated on the field. He survived, thankfully, but the game was canceled because the players were traumatized from watching one of their peers almost die on the field. This cancellation was very unusual for the NFL. A game hadn't been canceled since 1987. That was a player strike, and the time before that, 1935. So Mike's team had to figure out what do we do about this canceled game.
Mike:And it was a Monday night, and so you're moving on to the, I think it was at that point, week seventeen. You're moving on to the next week schedule, and you're going to end the season with an unplayed game. And so immediately, the question is, what does that mean? Right? Like you now, you went from having two hundred seventy two games to having two hundred seventy one. You went from a playoff structure that all was based on each team having the same number of games to suddenly the Bills and the Bengals are going to play one less. So then the question would be like, well, how do we make them replay the game? Right? Can we go to the week after the regular season? What if we went there? What if we went and had them play on the Tuesday night before the playoffs? What if we did this? What if we did that? You know, we had some analysts that were, you know, we were all working late into the night, and I remember, you know, sharing some of the implications. And it wasn't just going to impact the Bills and the Bengals; it was going to impact everybody at the league. It was going to impact all thirty-two teams because whatever we decided, it was going to give one team a small edge that they wouldn't have had otherwise. Or maybe the opposite of an edge, right? It was going to take away either the chance of making the playoffs or their chance of a home game.
Dan:And so, what eventually happened in that situation?
Mike:So, we didn't replay the game. That was the biggest thing. Is that our group did a lot of work to say that if we replay the game and we replay it after the regular season is ended, we're going to be replaying a game that is not fair to one of the teams. And what was going to happen is that there was an overwhelming likelihood that if we replayed the game. One of the teams was not going to have anything to play for, and so we would have been bringing back two teams for what would have otherwise been a really competitive game. Like if we replayed this game, we are going to have one game where we're going to be getting a backup quarterback, and we're going to be getting starters that are going to be sitting on the bench, and we're going to be putting it in primetime television, and it's going to be kind of a bad look. And so the the alternative was all right. Now we don't replay the game. What does that mean for our postseason? So then there were, you know, once you sort of got around that, then there were smaller implications in the postseason that we had to work out too.
Dan:So, Mike, we always end our episodes with a quick lightning round of questions. Here we go. What is a word or phrase that only someone from your profession would be likely to know, and what does it mean?
Mike:One of the the modern football data analytics metrics is called EPA, which stands for Expected Points Added. And at this point, you you can't really get into the field of football analytics without knowing what EPA is. And a big reason that EPA works is that not every yardage gained in football is the same. One yard on a third and one is a really good yard. One yard on a fourth and two is not a good yard, right? And so EPA compares the expected number of points scored based on the location of the team and the down and the distance of the play. And by being able to use a sort of a metric that accounts for that context where you are on the field, the down and the distance, and sort of looking at the the impact on overall net points scored, gives you a much better sense of what works for teams and what doesn't. And you know, largely speaking, it's been the main driver if you watch an NFL game. Every team goes for it on fourth and one.
Dan:Yeah, and and that's a real victory for analytics because it makes it a more exciting game. You know, so it's almost like the numbers gave people an excuse to take the game in a positive direction.
Mike:And you know, it's funny. The alternative to going for it is often punting, and a punt play has the most penalties and one of the highest injury rates of any NFL play type. And it pushes the team, you know, fifty yards further back than they would be otherwise. And the alternative is a fourth and one in midfield, where you keep, you know, a quarterback like Patrick Mahomes or Drake May on the field. Like, yeah, I'll sign up for that. So, yeah, fourth downs I think are pretty exciting for the game, and we're certainly happy with that that strategy.
Dan:What's the most insulting thing you could say about NFL data analyst's work?
Mike:I it absolutely drives me bonkers when people say that analytics doesn't have a role in football. And it it's it's been around for decades.
Dan:And you still hear that, like you know, post Moneyball.
Mike:Oh yeah, and in fact, a lot of people say you can't do Moneyball for football. And so, listen, it is tough. It is really tough to do football analytics well. And I mentioned context earlier. It's tough to do football analytics well because it's tough to account for context. But the big difference in football is if you were in baseball, you know, the difference between sort of making the playoffs and and sort of the edge you would need to get, you got to win five, ten more games, maybe. In football, if you can just win one extra game, right? It's it's been worth it. And so, you know, I think the the sort of edge here you don't need to go from a a four win team to a twelve win team to prove your value as a football data analyst, right? You need to get one win every couple of years. Listen, and and you'd be well worth it. So, you know, the edges may be tougher to find, but you know, once you do have them, your value is easy to identify.
Dan:What phrase or sentence strikes fear in the heart of an NFL data analyst?
Mike:I, I'm not totally sure if the league is okay with me saying this, but I've gotten a couple calls from the commissioner, and when that happens, it is from an unlisted phone number. And so, the couple times in my life where my phone rings and it's an unlisted phone number, I'm like, Oh boy, here we go.
Dan:Your heart starts beating a little faster.
Mike:Yeah, and and he has been a great advocate for our group and all the projects and the the work we've done with him. He has pushed us in the exact direction we needed to be pushed. But that doesn't mean that you don't get nervous when it happens. So that's the one that that stands out. The phone rings, and and listen, this is like once a year. I'm not on his list of favorite dials, right? But when there's an important data decision that that we're either involved in or or part of, you know, I think he does trust our our sort of objective viewpoint. So you know, occasionally that happens.
Dan:Can you still watch football as a fan, or or does your brain just get clouded with analytics?
Mike:Yeah, I'm not much fun to watch a game with.
Dan:But he does watch games at home, and he can see how his work has influenced the game, like a metric that he and his team created.
Mike:And I'll just point to my daughters or whoever's watching the game, like, hey, listen, I helped make that metric, right? And and so that's always cool, right? Is like you're you're able to to see the impact that that you're you know helping in a real small way have in the game.
Dan:Would you want your kids to follow in your footsteps if if they were interested in it?
Mike:Oh, I think being able to understand and interpret data is an incredibly important skill to have. You know, being able to sort of read through either misinformation or data analysis that's been done in a subjective fashion, right? That can be challenging to do. And being able to take an objective eye towards data and research is a skill that will help you anywhere. And so, yeah, I think any of these fields, I think, would be great for kids to get into.
Dan:Mike Lopez is the senior director of data and analytics at the NFL. I want to go back to that punt return the data scientists flagged. Their algorithm said the returner could have gained maybe ten or fifteen yards. Instead, he called a fair catch and got zero. It was the worst decision a punt returner made all year. But then you look at the game footage and you see no. His team was up three touchdowns with a minute left. There was no upside to a return, only the risk of a fumble or an injury. The naive data says this was a mistake, but in context, it was a wise choice. In my book Reset, I tell a story about the San Francisco 49ers. The team had rigged up a way to get instant feedback from fans at the stadium. Are the bathrooms clean? Are the concession lines moving, thousands of data points are pouring in, and someone has to make sense of them. In one game against the Cowboys, there were tons of complaints. In another game against the Jaguars, there were hardly any. So, what had they done? Were they cleaning the bathrooms differently or stocking the concessions better? Nope. The difference was the Forty Niners were beating the Jaguars and losing badly to the Cowboys. And losing made the beer seem a little warmer and the restrooms a little dirtier. Data seems to speak to us in black and white. It's instantly credible, but it's also surprisingly brittle. Shift the surrounding frame, and our interpretations can shatter. That's why Mike's ability to speak both languages, football and data, matters so much. Because the data only tells the truth when you understand the world it came from. Turning game film into a research dataset, anticipating questions before anyone asks them, modeling what else will change if you change the rule, and helping shape the game without stepping onto the field. Folks, that's what it's like to be an NFL data analyst. A shout out to recent Apple Podcast reviewers Artichoke3.14, Strandinavian, and Lisa the Librarian. This episode was produced by Matt Purdy. I'm Dan Heath. See you next time.
People on this episode
Podcasts we love
Check out these other fine podcasts recommended by us, not an algorithm.
The Economics of Everyday Things
Freakonomics Network & Zachary Crockett
Ologies with Alie Ward
Alie Ward
Hidden Brain
Hidden Brain, Shankar Vedantam
The Moth
The Moth