Inside the NFL Numbers Game with Cynthia Frelund
She’s on a crusade to bring analytics to the average football fan
Cynthia Frelund has a message for all her fellow quants (the analytically-inclined): “Numbers people need to not be a-------, so that people genuinely like it. People hear math and they plug their ears, like la la la.” In her role as the first-ever analytics expert for NFL Media Group, Frelund is on the crusade to bring analytics to the common NFL fan. She offers analytics insights on NFL Network’s GameDay Morning and NFL Fantasy Live, and will contribute to Super Bowl, combine and draft coverage. She spoke to The MMQB about the future of analytics in the NFL, why this is a breakout year for data, and the football trends she’ll be watching.
KAHLER: You studied biology as an undergrad at Boston College and then went to Northwestern’s Kellogg School of Management for an MBA in finance, entrepreneurship and innovation. How did you find your place as an NFL analytics expert?
FRELUND: I worked at the NFL, in 2008, in the CFO’s office, and I realized that—because I can never say, When I was a shutdown corner—if I wanted to be impactful in this space, I have to be really good at something that gives me a unique value. And for me, I’m a quant; my brain is a math brain. That was a place where I could provide value. I saw that opportunity when I first started in 2008. There were early indications in the analytics and evaluations of players at that time, how valuable is this player to our team or to the league? I was able to relate my love of football and film study to a unique value in numbers. While at Kellogg, I got a chance to realize that if I added the ability to code and understand predictive capacity of numbers, then I could change my career trajectory.
KAHLER: You were an analyst to the CFO at the NFL, what was that job like?
FRELUND: I worked in financial strategy directly for the CFO. At the time the CFO was Anthony Noto, who is now the CFO of Twitter. He oversaw all the big projects. At that time, it was prior to the lockout and we were looking at the rookie pay scale. We also researched if a team should move to L.A. The NFL does a good job of really looking at their product and the games that they put on the field and saying, How can we maximize what we are doing for the growth of our company? We also looked at the season inventory—how many preseason games should there be? Right now the inventory is 16 [regular season] and four [preseason]—and maybe it is better to go 18 and two. All the evaluation stuff is number strategy. It’s financial modeling and telling stories with numbers, strategic stories about numbers based on a lot of different factors.
KAHLER: What did you learn from the research about a team moving to the L.A. market?
FRELUND: I think the better question that the league always asks every year is, Where does it make the most sense to have teams? Yes, L.A. was obviously on that list. I think L.A. has been on that list since a team moved out. If an owner says that he wants to move a team, the league will look into it and do pretty comprehensive studies about potential successes in each city and look at drawback and overall impact. Much of the league’s revenues are shared, so individual-owner decisions are vetted by the league office. The other 31 owners get a chance to understand what the consequences of one owner’s moves are.
KAHLER: Before joining the NFL Media Group, you were a predictive analytics analyst on SportsCenter and covered a variety of sports. Compared to other leagues, how is the NFL unique or different when it comes to analytics?
FRELUND: In terms of analytics, baseball is obviously the one that is the most out about it ... it’s just more of an accepted practice to talk about analytics. In every sport, you’ll see some applications of analytics around contracts, around injury information, but in terms of pure strategy and which players we play in certain situations, baseball is obviously the most advanced. The NBA is very forward-looking about what they are trying to do in tracking players; they are probably taking the biggest steps forward. Apart from Paul DePodesta being hired [by the Browns], analytics in the NFL are kind of a newer thing to talk about. Many NFL teams may not call it analytics, because analytics can be a confusing word. It’s overused and misused a lot. It is not just statistics, it is using historical models to contextualize statistics. The lens you filter it through can create better models to ask, What does this mean for next year? The more data you have and the more you understand your priorities for each position, the better your analytics and evaluation can be. It is context and historical kind of modeling to help predict future outcomes, and I think that gets lost in things. Football is harder, because there are 11 people moving on each side of the ball on every snap. In baseball, a pitcher throws a ball to a batter and there are a lot of different pitches and things that can happen there, but nowhere near the amount in football—multiply that by 22.
KAHLER: Did the DePodesta hiring surprise you? Did you take it as a sign the NFL's acceptance of analytics is advancing?
FRELUND: I think many teams have an analytics person on the staff already. The job is just not called analytics. It's the director of player operations, or special quality control or something like that. There are different titles of people who are different levels of analytics and statistical analyses on every team. They just don’t label it like that. It’s not specifically titled, chief strategy officer. The thing I thought was really interesting was the fact that Paul DePodesta got hired before Hue Jackson did. On other teams, it seems like a coach has their ace in the hole, someone who they can communicate with who will give them the analytics they are looking for and who knows them very well so that they work in lockstep. A very big and important function of analytics is communication. I can give you every statistic in the world, but if I’m working for Bill Belichick and I don’t know exactly what he wants, he’s going to think I am useless. So that was interesting to me, the timing that Paul got hired first because that places an interesting emphasis.
KAHLER: How do you think DePodesta will specifically help the Browns?
FRELUND: Now, here’s the thing that I think gets conflated a lot with his hiring. There are performance analytics and then there are sports management analytics. Performance analytics are like third down running offense, and situational context with statistics. And then there are management analytics, like contract bonuses. Many teams structure contracts so that if a defensive end gets a certain number of sacks in a season, like 10 sacks, then he will get a bonus. Maybe some of the stuff that Paul DePodesta is doing is saying, How about five sacks in the first six games? Because really, a team doesn’t care about those nonsense stats when they are not playoff-eligible. I do not think there is any world where DePodesta is in Hue Jackson’s ear saying it’s third-and-7, call this play! But I do think there is a framework that he can help provide, by saying this is something that is a high probability of happening in this situation and this is something that is a low probability of happening. I don’t know if he is doing sports performance analytics or management analytics or some blend of both. But I would stake my salary on the fact that he is not calling plays.
KAHLER: It seems to me that many players are scared of analytics being used against them. Have you noticed this?
FRELUND: That’s something that I have been thinking a lot about because I think we need to have more empathetic quants. Maybe it is a personality trait; if you are attracted to numbers, maybe you don't care enough to sit there and say, Hey, what is really important is for us to talk this through to show you what it means. A lot of times I hear numbers people take shortcuts and it seems scary. If I was a player, I might be scared too if I don’t have a numbers person who is willing to sit down and say, Just so you know, this actually can help you too. Players see some stats used against them and there is this overwhelming feeling, whether it is perceived or real, that the team is trying to pay the player the least amount of money as possible. And of course, the player wants as much money as possible. I think players are afraid that because the analytics people are employed by teams, they are getting numbers to support the team’s position to pay someone less. You need to have the people who are willing to take the time and explain it, and I think players should be included in it. I would love to have players watch film with me. I would love to ask them questions about what they were supposed to do, because I think you could create a fair environment of evaluation. I don’t want to tell anyone what to do. I want to provide them with tools to do it better.
KAHLER: What are some trends in football you are excited to watch develop this season?
FRELUND: We are doing a big study on position spending by team. Not just your number one player, but your entire position group and relating that to on-field performance. That is something that I am really keeping track of this year. I went back and did it in historical years, to see what happened and it does hold true, teams who spend a lot of money on pass catchers do better, but I am curious about the impact of defense. Like these Von Miller contracts, that is going to happen to Khalil Mack in a couple of years and I’m curious to see how spending relates to on-field performance. I’m not fully through it, but I will be tracking it this season. Does spending on defense get you more wins?
KAHLER: Will we see any new football stats come into use?
FRELUND: The NFL actually has the Next Gen Stats platform that collects data from sensors tracking capabilities in shoulder pads. The idea is that, given that we want to track new things and understand more about the game, where are the cool spots to explore? I am trying to encapsulate things we should be looking for—things that we care about but aren’t captured by statistics right now. Like, what is a running back doing on a passing down? There is no statistic for that, you have to watch that on film, because there's nothing you can do to measure that. What happens in the game that we don’t have the statistics for? The trick is to solve and answer questions that people actually have. Anytime I spend time writing a code or an algorithm, it's to answer a question that someone actually cares to know. While I was making algorithms, I got a chance to ask actual coaches and players what the most important stats at each position are. What makes a quarterback great? And the answer is not the same for each coach, but there are some things we can stack up as one way to measure these kind of quarterbacks against each other and it becomes a conversation. One of the reasons I came here from ESPN, is that there are certain strategy things that I am able to get in on pretty early because I think this is the year that analytics starts to be more accepted. Obviously ProFootball Focus’ stats on the NBC broadcast shows it, the hiring of Paul DePodesta shows it.
KAHLER: Would you want to go work for a team as their "strategy" person?
FRELUND: Yes. I’ve done some projects for teams. I'm new to TV, and maybe it is a little idealistic, but I believe right now is when people are going to take the time to understand and listen to analytics, and I think I am good at telling that story. I believe I have that kind of empathy and am good at looking at things through someone’s lens. I want to be shepherd of those conversations so that people like it. I just want people to like analytics.
• Question? Comment? Story idea? Let us know at email@example.com