The countdown to the Super Bowl is on! Let’s take a look at the role of analytics.

Ashley Kilgore:

A quick note to our listeners that this episode was recorded on January 26th and posted the following week.

At the time of this recording, we are well into this year’s football playoffs, with the Super Bowl, now just a couple of weeks away, there are four teams still in the playoffs, the Bengals, Chiefs, Eagles, and 49ers. With the next round coming this weekend. I invited my previous guest, Walt DeGrange, director of analytics capabilities at CANA, to take a look at the continually evolving role of analytics in football as we continue our countdown to the Super Bowl. We’ll also take a few minutes to discuss Walt’s new book, Field Guide to Compelling Analytics.

Walt, it’s great to have you back on Resoundingly Human and I thank you in advance for your patience walking this non-sport centric host through the application and impact of analytics and football. So you last joined me in 2019 for what turned into one of our most popular podcast episodes of all time, which was also focused on sports analytics. So Walt, no pressure, but I have high hopes for this episode.

Walt DeGrange:

Well Ashley, it’s a pleasure to be back on the podcast and looking forward to our conversation.

Ashley Kilgore:

So Walt, I’d love to start with some more general questions about analytics and football and then I actually enlisted my husband and his friends for some more specific or targeted questions. And then finally, we’ll take a look at the upcoming Super Bowl.

Walt DeGrange:

Sounds great.

Ashley Kilgore:

So what are some of the earliest applications of analytics in football and how have they developed over the years?

Walt DeGrange:

It’s pretty interesting to look back over time and as with many applications of analytics, it has really been dependent upon the data that was available at that era. So in the beginning there were things such as play tendencies and also players statistics like the time in the 40 yard dash, the amount of weight that they can lift. There’s a combine every year for draft picks. So anything that was available as one might imagine like 50 or 60 years ago, over time that has increased with the ability to get data from digital devices and GPS devices and accelerometers. And so yeah, but really the beginning was back in the day just through observation when people had a clipboard or a stopwatch and they were collecting data that way.

Ashley Kilgore:

So can you also share what are some of the behind the scenes impacts of analytics in football from coaching to training to maybe even sports medicine and player conditioning, maybe something the average fan doesn’t know about?

Walt DeGrange:

Yeah, that’s a great question. Football in general is a very physical sport and it didn’t help that the NFL has recently added one regular season game to the schedule. So now it’s 17 games, even with one week off, that is a tremendous taxing of the body. And even though not every player plays every play, right, there is a lot of contact. So you’re right, I think a lot of behind the scenes is with the physical therapy, the recovery, a lot of the sensors that are used are actually used in practice and when the players are being for conditioning, so when they’re running or lifting weights, and there is a tendency throughout the season to, you don’t want to overtrain or you definitely don’t want to have an injury when you’re not actually on the football field if you can not have that happen or to occur.

And then also the other piece of this is the recovery from injuries. So if somebody has an ankle sprain or an injury, a shoulder injury earlier in the year, then how fast is that recovery occurring? So that I think in the sports medicine piece, that has been a huge help over the years and it continues to increase, I believe pretty much every team at the professional level. And that’s in all sports, not only football, that are using analytics in that way right now.

Ashley Kilgore:

So looking ahead to the future, what do you think is going to be the next big application for analytics in football? Or maybe not even the next biggest application, but just one you’re really excited to see developing?

Walt DeGrange:

So when you talk about applications of analytics, the first thing you have to ask yourself is, there are three different applications and they’re all really cool. And this all goes back to basic analytics 101. You have descriptive analytics, it’s the pyramid, you have descriptive at the bottom, which that has traditionally been like how good a team’s offense, how good a team’s defense is, yards given up or yards gained on the ground or passing. And that’s been around for years. You could go back to the 1940s and probably see that in a newspaper. Then there’s predictive analytics, who’s going to win the Super Bowl or that’s especially prevalent in betting in Las Vegas or who are the odds on favorites or who do you want to bet on? And then prescriptive is where we see, what am I going to do? What play should I call?

Should a player play today if they’ve had an injury or what is the best alignment? What is the best game plan, game strategy to defeat another team? And I think that’s where we’ve seen the most growth in advanced analytics is that prescriptive. And once you get there, the question becomes how do you help a team increase its probability of win? And for most people in analytics, they’re going to say, well make a better model. And in fact, that’s probably not the right answer. It’s how do you create a model that’s going to help the people that that’s their job to do win? So you don’t want to potentially make the team un-defeatable, you want to make the coach do their job better and that will lead to more success with that team.

So that’s I think where analytics professionals in general, and this is the same in business, you don’t necessarily want to make a model that increases the stock price for an organization. You want to help the people that that’s their job to do that. So we’re not trying to replace anybody. A lot of people have the joke about AI taking over everybody’s job. I think the best use of AI is augmenting decisions and then making the people that, that’s their actual job, making them perform better, whatever role that is.

Ashley Kilgore:

So as I alluded to earlier, knowledge of sports is not my strongest skillset. So in preparing for this podcast, I enlisted my husband who is a huge football fan. In fact, I heard him turn the TV down downstairs, so I think he’s listening in on us right now. And he reached out to his network, specifically the Big Blue View, which is an SB nation blog. And here are some questions that he and his friends came up with. So time to throw, is this the most misunderstood stat in football and if so, why?

Walt DeGrange:

So it’s amazing for those of us who have done or been around sports analytics for years that whenever you’re out at a bar or hanging out with friends watching a game, there are questions like this that come up all the time. It’s like, what about this statistic or what about this statistic? And analytics in general is about using multiple data sources to create a different picture. And this time to throw, it’s important, however, there’s a context around it. So just because a quarterback has a quick release doesn’t necessarily mean they’ll be successful. They could just throw it really fast and get intercepted every play, which that would be a failure and they wouldn’t be around very long. It depends on many things such as how does that quarterback need to release the ball? Is that offense designed to run quick passes? How fast can the quarterback actually read the defense? And that even happens pre-snap. So even before the ball of snap the quarterback and other players and coaches and everything are looking at the defense.

So yeah, I think it’s if there’s an entire game plan that is around a quarterback releasing the ball very quickly, which there are a few around, then that’s very important. But if it is an entire offense that perhaps isn’t dependent upon quick reads and such, and there’s a lot of examples of quarterbacks that haven’t had the quickest release such as Tom Brady and have had wonderful careers. So I think, most of the answers to these questions are, it’s a combination of the coaching, the game plan through the players that are around that quarterback. So you’re absolutely right. I mean, you could have a quarterback that could release the ball very quickly and have wide receivers and tight ends and such that just can’t run the routes that fast. So it’s a combination of things. I think that’s what makes football super interesting is, you’ve got 11 players versus 11 players, then you call two plays and the plays may change at any time and then the ball is hyped, they’re snapped, and then all the players have to run and do things independently.

They may trip, may fall, things may change. So yes, I think it’s important and it may give a quarterback more opportunities to run different plays. But yeah, I don’t think it’s like the stat that will make a quarterback super successful. And the other side of this is that if you get rid of the ball, you’re not going to be taking a lot of sacks. That’s true, but a lot of the injuries occur with the quarterback, and I’m not saying an illegal hit that causes a penalty, but they may trip over another player or twist an ankle even on their own as they’re scrambling around. So yeah, it may reduce the injuries over a career, but there are so many other ways that football players can injure themselves while performing great feats of athleticism on the football field.

Ashley Kilgore:

Which are the most valuable stats when evaluating the performance or potential of football players and how do you measure these?

Walt DeGrange:

Yes, it kind of starts all at the combine when all of these people who are looking to get drafted have to perform some drills. 40 yard dash is one of the more famous ones, but a bench press and they run through these obstacle courses and such. And I believe they even take a test, although there was a written test, there was some folks or peers back that were like, well, that written test is kind of useless to measure their football knowledge. I think overall, that is one aspect of it. The other piece though that’s even more important is how do players fit into a certain scheme or game plan? And that’s a lot harder to measure and there’s like a lot of chemistry. So for instance, if your offensive line, which is very important because they protect the quarterback and they open up poles for the running backs to run through.

So very important for the offense, but you need the line to be relatively equal in skill. If you have somebody who is really big and really fast and you may be able to take advantage of that to a certain extent, but they really work as a team and that’s the piece I think coaches have to be very inventive and innovative and look for the skill sets and players that work best in their scheme and also be flexible enough to change their schemes to fit, for instance, if there’s injuries or if players have certain weaknesses, maybe they run really well to the right but don’t run really well to the left and such. That’s just a simple example.

Yeah, so I think again, that’s the fun part about football is that you do have many variables and you’re looking at many, attributes, not only physical, it’s great if somebody can run really fast, but again, if the quarterback can’t throw, if the wide receiver’s always out running the passes, then it doesn’t make a difference how fast you are or how far the quarterback… If the quarterback can throw, there’s a lot of examples of quarterbacks being able to throw 70 or 80 yards, but there’s not enough time for any wide receiver to run that far, right. The quarterback will get sacked before anybody, even the fastest of wide receivers can’t run that far. Yeah, I mean there’s the reality of the field and the physics of it. So yeah, there’s the individual statistics and then there’s how it all fits together.

Ashley Kilgore:

Okay, so catch probability seems to be an impossibly complex stat. Is it something that is quantifiable?

Walt DeGrange:

Right? Well, according to Amazon, yes. So I think everybody, I’ve greatly enjoyed, I don’t know, and I don’t work for Amazon, so just to get that out front, but when Amazon Prime televises the games, there’s actually a dedicated channel to analytics and stats. So if anybody is geeky enough to listen to this podcast, next year, you need to turn on that channel. It is just absolutely incredible real time statistics and one of them is the catch probability.

I will say this, it’s like if somebody jumps up in the corner of end zone and catches a bullet pass from your favorite quarterback with one hand and gets both feet down in the end zone before being pushed out, it wouldn’t take any type of analytics to figure out that was a very, very low probability of catch, right. I think more surprising to me is that there was a playoff game last weekend between the Cincinnati Bengals and the Buffalo Bills and it occurred in Buffalo and it was snowing and there was snow on the ground and it was cold and part of the game it snowed and the snow was coming down relatively hard and Cincinnati was able to complete passes and even the easiest of passes in those conditions when it’s windy and snowy and people’s hands are cold.

I think that is where this probability of catch was just amazing because some of those routine catches, and that’s where the probability of catch has come about in prevalence is that not only does it consider the actual difficulty of the catch, but it also considers weather conditions and wind conditions. And so that, I haven’t seen the statistics yet, unfortunately that wasn’t broadcast on Amazon. However, some of those catches even in normal conditions might have been above 90%, but in those conditions it could have been 50/50 just because of the weather, the coldness and everything. And the Bengals completed a large percentage of those. I think it was well over 90%. It was just incredible. So I think in my mind that’s where that stat really shines is that when it’s impossible, everybody can figure it out. But when it’s not so impossible, then people gain an appreciation for just how good that team was or how good they performed.

Ashley Kilgore:

What goes into the formula for deciding on a fourth down when to go for it and when to punt?

Walt DeGrange:

This gets into a very interesting aspect of analytics and the statistics depending on what you consider, what your objective function is and many people a few years back, and this goes all the way back into the two thousands, there was a coach, a high school coach that never punted and would go for it on fourth down. So they gave them four downs to get a first down and so the statistics supported that. And it does as far as scoring, if that is your main goal, and then you need to teach people statistics. Statistics works over time on any one given. If we flip a coin 50% of the time it’s going to be heads, 50% of the time it’s going to be tails. But if we just flip the coin 10 times, you could get nine heads, one tail, and then you’re like, well, statistics doesn’t work.

So that is the main piece of this fourth down. If over a thousand coin tosses or a thousand fourth downs, then you basically win a lot of games or score more points, then you’re good to go. But if you’re a coach, you may not have the ability, you may be fired before you get to 1,004 downs to prove your point. And I think that’s the thing about the NFL and in college football, why that hasn’t been prevalent, is that going forward on fourth down all the time is that it is, it’s the perception. And even in the NFL today, there’s a lot more teams going for fourth down, especially closer the 50 yard line in the opponent’s territory. And I think that is supported by statistics.

However, again, if that happens in the Super Bowl and that leads to your opponent getting the game winning drive or field goal, you could have been successful 999 times out of a thousand and that’s the one time it failed. And analytically, that was the correct call but if that’s what everybody remembers, then you could potentially lose your job. And I think that’s the thing that all head football coaches are extremely sensitive to and it just goes to show you their objective, yes, is winning the Super Bowl, but they also have an alternative objective and it’s not being fired.

And we had touched on this in an earlier question, how do you help them be successful? And so I think giving them the insight that the odds are that the statistics and probabilities say that you should go for it on fourth down, it’s overwhelming. I mean I’ve never seen a model that hasn’t said, in most situations that you shouldn’t go for it on fourth down. But again, when your job’s on the line, then it becomes a matter of trust in analytics, trust in statistics. And I don’t hold anybody, I’m not going to blame anybody for wanting to protect their job. So that’s something that they can control.

Ashley Kilgore:

So as I mentioned earlier at the time of this recording, there are four teams still in the playoffs, three of whom have a quarterback on a rookie contract. Does this mean there is validity to the idea that it’s better to have a rookie quarterback so you have more money to invest in the rest of the team and ultimately a better chance of making it to the playoffs?

Walt DeGrange:

Yeah, this is a pretty interesting question on many levels, and again, I’m not an expert at the draft analytics and such, but personally I think when I’ve done some analysis on doing trading in for draft picks and such, I think that there are two things that stand out. One is like you had mentioned in the question, that may allow for more money to be spent in other areas and we’re seeing a lot of specialization in football, which means that again, that out of those 22 players and if you count the punter and the kicker and some special teams folks, you’ve even got more players that you’re going to need to spend more money on quality players. And so you’re exactly right that may allow for that. I will say another aspect is that the college offenses and the professional offenses tend to look more the same today.

I know when I grew up in the eighties, in the nineties, especially in the eighties, there was a distinct feel for college football and most all players had to come up to the professional level and learn almost a new game as far as new skills and the way the game was played. And the rules were also a lot different. And now I believe when you see teams like Alabama or Clemson or Michigan or Ohio State, they’re playing those same offenses. Like Jalen Hurts is the quarterback for the Eagles and he was the quarterback for Alabama and then transferred to Oklahoma. If you take a look at the offense the Eagles run, you’ll see a lot of the offenses that he ran when he was the quarterback for at Alabama and Oklahoma.

They’re not exactly the same, but they’re much more similar, the reads and the things that they want that quarterback to do than if you go back to 1980 and trying to figure out that college quarterback had a lot to learn and it was expected that they would basically start from almost ground zero again. So I think in general, that has to do with the team also, there are teams that don’t run offenses that are similar to the college. So a rookie quarterback on that team may not have as much success, right. Now you’re expecting that during their rookie contract to not be as successful and perhaps be three or four years before and they may have to play underneath a more seasoned quarterback before they get trusted with the system.

Ashley Kilgore:

Do you see any analytics trends that are indicators of success and lead to a better chance of creating Super Bowl contenders?

Walt DeGrange:

Yeah, I think that the use of analytics is different for every team, and that’s with anything in life, with every organization, whether it’s a business or government or military or football team has a different level of use of analytics. And I think that, again, first of all, they have to trust it. And one of the extremely exciting things that I’ve seen over the past probably six years since we’ve last talked was in 2018, Croatia was in the World Cup. The men’s team went all the way to the finals and one thing that they use analytics for was to go through the game tape and we just had a World Cup a few months ago. So the games are very close together, which is the same thing with football, especially if you play on Sunday and then have to play again on a Thursday night game.

So that’s just one application where the AI can go through all the game film, which there’s a ton of because there’s different angles and such. And the coaches can basically say, what are they looking for? They’re looking for a certain player to do a certain thing and the AI will actually go through all the game film and create those clips for them. And that’s what the Croatian men’s soccer team did too. And then the coach can basically develop a plan or coach their own players appropriately on what to do with that. And I think that’s the biggest thing again, it’s kind of enhancing the strengths of the coaches already. So you’re working with the coaches and I think getting that time to where you take all the data, in this case video data, or it could just be game data and identifying what is important to that team and then being able to make a game plan from that I think incredibly, it increases your competitiveness a lot just due to the fact that you’ve only got a week to prepare.

Now after the conference championships, they’re going to have two weeks to prepare for the Super Bowl, however, that seems like a long time, but there’s a lot to be done. And then the other thing is not only do you have to develop the game plan, then you have to communicate that to the players and then they also have to practice some of the plays and such to make sure everybody’s on the same page. So yeah, I think just the cycle time of being able to actually analyze the massive amounts of data now that are out there with all the cameras and sensors and everything, make heads or tails of it and then be able to translate that into an actionable plan that is also that you can actually practice and then be able to come out and execute it. It is just amazing, the amount of data analytics that goes into that nowadays.

Ashley Kilgore:

So now speaking of the Super Bowl, at this time there are a lot of predictions being shared and maybe even some bets being made. Activity that will only continue to increase the closer we get to the Super Bowl. What role can analytics play in helping predict the winning team and what kinds of factors are considered?

Walt DeGrange:

So this gets back to the fun part of analytics, and not that any of it is not fun, but I think a lot of folks will find the predictive nature of analytics pretty fun. And of course that’s what Las Vegas lives on, people betting on their favorite team. I think in general, and this is a really good piece of analytics also, the data that supports game planning and player analytics and if a player plays. That data, a lot of times isn’t the right data to predict who’s going to win, which just kind of blows people’s minds. I mean, you may have a data set that’s really good at doing prescriptive data, like what you should do, but it is a really poor predictor of the outcome of a game and such. And I think that a lot of people have a hard time wrapping their head around that.

So some of the predictive things that have been used in the past have been the efficiencies of the offense or defense and the styles that teams play that I think are pretty useful. The other piece of the predictive nature is injuries and who’s playing. So I know now the big question for this conference championship weekend is the Mahomes’ high ankle sprain and how much that’ll limit his mobility. So I think factors like that are incredibly interesting to throw into analytical models and then see how that affects their offense, which is highly dependent upon Patrick Mahomes moving around the pocket and outside to gain extra time to throw the ball. He does not have one of the quickest releases in football, so sometimes he has to create separation with his speed and agility to give him some more times. And he’s also very inventive on how he can throw the football, which is pretty cool too to watch.

But yeah, I think those factors, so things like injuries and then the matchups, which once you get to this part of the season, some of these teams have played a few games if they’re in the same division, which I don’t think any of these teams are. So if they’re not, maybe they’ve played against one another this year, but again, maybe not. And then people will look at common opponents and such. And I think overall, one of the overriding things since the length of the season up to this point is the injury status and who can play and at what level they’re going to play at.

Ashley Kilgore:

So now Walt, I have to ask, do you have any predictions of your own for this year’s winner?

Walt DeGrange:

Right, so I’m a huge Eagles fan, so this is a great year for me. I know 2017 was also a great year, however, this year the Eagles are really dominant, especially offensively and seem to have both the running attack and the passing attack and have really shown that they can defeat teams multiple ways. So I’m hoping that that’ll give them the edge. I will also say the 49ers are a very good team, so that should hopefully be a much better game than the Giants, Eagles game was last weekend. So yeah, I love just watching good football. It’s not so fun watching blowouts, but they occur. But yeah, so I’m hoping for a second Superbowl for the Eagles this year.

Ashley Kilgore:

So as I mentioned earlier, you published a book last year, Field Guide to Compelling Analytics, and I’d love to hear more about that.

Walt DeGrange:

Yeah, it was a great year for publishing books, at least for myself and my co-author, Lucia Darrow. Yeah, we both felt after practicing analytics for many years that there wasn’t a real guide to the implementation side of it. Most books tend to focus on the doing analytics, which is incredibly important. And we always joke, there’s only one equation in our book and it’s basically analysis plus trust, plus communication, plus experience must overcome a convince them factor, and that’s it. So if you know addition, and you can do it one equality, you’ve mastered all the math in our book and it really is about implementing analytics at any level, even if you’re just trying to convince your team members that your analysis or your analytical method is working or the data’s right, or it could be at a higher level, it could be your team lead or your C level, or it could be the President of the United States or a general or an admiral could be the World Trade Organization.

At any level, you’re going to have to have those elements. So you’re going to have to have good analysis and then people are going to have to trust you. Most people do not understand the math, so if they don’t understand the math, they’re going to have to trust Walt, right? Because they’re going to have to trust Walt understands the math, and I think that’s a commonly overlooked piece of this. The communications is obviously really important as far as you have to explain just enough. You’re not going to teach people statistics or analytics and that shouldn’t be your job, but you need to give them enough information to understand. And there’s different levels of that. Not everybody needs to understand everything, some people just need to be aware of what you’re doing. And if you have the trust, then that may take pressure off your communication strategy.

And then of course, experience is the last piece. So everybody has the analytics experience, but then there’s also subject matter expertise. And a lot of times you may have that, we’re talking about football, you may have that level of expertise up to a certain level, but if I’m trying to convince an owner of an NFL team, then I’m probably going to need to bring in somebody outside that has more experience than me, like a former coach or somebody that that owner knows football and then understands analytics. And so they have that trust and that experience and communicate with that person. So I think the main message of the book is a lot of times we tell analytics professionals, don’t brief the same to different levels.

And that’s good advice. However, it’s only part of the story. So I think we tried to fill in some of the other deficiencies such as the trust and experience piece. And again, there’s probably no analytics professional in the world that at every level can fill all those boxes. You’re going to have to collaborate and bring in team members. So it’s not only about you finding these skills, it’s about when you need to bring in help.

Ashley Kilgore:

So Walt, thank you again so much for joining me. I definitely learned a lot and I will say, have a brand new appreciation for football that I did not have before our conversation. Before we wrap up, I’d love to hear what’s next for you. Any more books in the works or other fun new projects in 2023?

Walt DeGrange:

So my company has a fun new project. So we’ve gone into eSports and for those of you who aren’t familiar with eSports, it’s basically competitive video gaming. It’s been around for quite some time. And we’re partnered with a worldwide women’s rocket league. And for those of you aren’t familiar, rocket League is it’s playing soccer, but you’re driving cars. It’s really fun. And one of the things that is very important to CANA and to myself is, trying to get women into eSports. There are many, the traditional sports, there could be reasons why you have men and women separated, right? In eSports, we feel like there shouldn’t be a reason since it’s video games and everybody can play video games. So we’re really excited to be partnering with that organization and they have, competitions in North America and Europe and hoping to roll out worldwide next year. So we’re actually providing some advanced analytics and some machine learning to try to provide that type of real time analytics during their Twitch broadcast. So that’s super exciting. And World Championships are in August.

Ashley Kilgore:

Want to learn more? Visit resoundinglyhuman.com for additional information on this week’s episode and guest. The podcast is also available for download or streaming from Apple Podcasts, Google Play, Stitcher, and Spotify. Wherever you listen, if you enjoy Resoundingly Human, please be sure to leave a review to help spread the word about the podcast. Until next time, I’m Ashley Kilgore and this is Resoundingly Human.