Access exclusive podcasts, interviews and analysis with a monthly or annual membership

Become A Member
Dublin: 10°C Friday 25 September 2020

The brains behind the NFL's brawn

Cynthia Frelund was NFL Media’s first analytics expert.

The Patriots and Falcons in Super Bowl LI.
The Patriots and Falcons in Super Bowl LI.
Image: Bob Andres/PA Images

ON A SCALE of one to Alanis Morissette, the fact that Cam Newton was making sexist comments to a Charlotte Observer reporter at the same time The42 was interviewing Cynthia Frelund, one of the smartest minds in football, falls somewhere between a black fly in your Chardonnay and a death row pardon, two minutes too late.

If you’re even a passing American football fan, you’re probably well aware of Frelund’s work. Before joining NFL Media, she was the predictive analytics analyst on ESPN’s flagship show, SportsCenter.

Since taking up her role as the organisation’s first analytics expert, Frelund’s work has appeared on everything from NFL Network to and, this season, she’s been given her own podcast alongside Matt ‘Money’ Smith, called Game Theory and Money.

The new show, which drops every Thursday, sees Frelund use advanced analytics and a proprietary algorithm of her own creation to run over 10,000 simulations per game in an effort to gain insight into player performance and game outcomes.

Given her already busy schedule, how does Frelund fit in the time to enter all the data needed to run those 10,000 simulations for the new show?

“The code I wrote automatically inputs statistics from the NFL’s Game Statistics & Information System,” Frelund says.

“Then it adds in things like personnel changes so the projection will be different for the Oakland Raiders this weekend because Derek Carr is not playing than if he was.

“I created paradigms of contextual situations after I asked a bunch of coaches what was most important to them so I came up with a bunch of statistics that may not necessarily be tracked normally, things like missed tackles that don’t appear in the box score.

“In addition to that, I watch each game and manually add in all of the other statistics the coaches have told me are important, and then I run the simulations with all of that information in mind.

Once everything’s inputted, the code is pretty quick, so it takes about an hour to simulate 16 games 10,000 times, but watching all the games and manually inputting the stats, that takes a long time.”

The result is a podcast that not only appeals to those with an interest in fantasy football and/or gambling, but any American football fan whose motto is “if everyone is thinking alike, then somebody isn’t thinking.”

This is especially true when trying to predict NFL seasons and even NFL games. The narrative often replaces analysis and, even when writers and broadcasters do look closer at the Xs and Os, they can often be looking in the wrong place says Frelund.

“I think defence gets underrated a lot, especially the influence of defensive turnovers.

Takeaways from a defence are so key because they drive the starting field position of an offence and, in a normal game, a team has between 10-11 possessions on offence and they score between two and 2.5 points per possession.

“If your defence creates a turnover, they not only give your offence another possession, often in great field position, but they’ve also taken away one of the opposition’s chances to score and I don’t think people give it enough credit for how important it is.”

NFL 2017 - Houston Texans vs Jacksonville Jaguars Malik Jackson celebrates one of 10 Jaguars sacks against Houston. Source: Erik Williams/PA Images

Conversely, Frelund says there are aspects of the game that teams and pundits consistently overreact to.

“The players we hear the most about, they’re the ones we tend to overrate the most.

For example, if we hear a lot of stories about Odell Beckham Jr during a week, or analysts keep saying ‘the Giants don’t have a run game’, we tend to overrate the importance of that narrative even when it doesn’t play out.

“Odell Beckham can have a great game, even if he doesn’t score a touchdown, because he draws defenders away from other receivers, leaving them open. But if he doesn’t score, that’s the media narrative and what people talk about.”

Frelund also has little time for inflated stats.

“You look at a team like Jacksonville that had 10 sacks in one game. Now they have 18 and are the sack leaders but, when 10 are coming in one game, that changes the narrative.

“People just kind of overrate the things they can see.”

For anyone interested in sport, particularly the outcome of sporting events, gut-feeling is often something that’s difficult to ignore, even when all the data is pointing in the other direction.

This weekend, for example, the winless New York Giants host the winless LA Chargers. Eli Manning has never beaten the Chargers, so it feels like this game is a chance for the Bolts to get their first W of the season.

However, Frelund’s prediction, based on her simulations, is that the Giants get a 23-21 win.

But how does she remove subjectivity from her analysis?

“Usually for me, with the data, I can look — using code — for associations and correlations that I might be underrating and then I ask myself why.

“Look at the LA Rams this season; their offence is scoring 21.5 points more per game but their defence is giving up significantly more yards per game than it did last season.

“Therefore, they need to have more offence and take more shots down the field. So, for me, it’s about putting everything I see, or the data that’s presented, in context and that’s the cool part.

I think, for most people, they would say it’s an advantage to have played football if you’re talking about football and I would agree with that in large part. But I think I’ve turned my disadvantage into an advantage because I’ve really listened to a lot of different voices.

“Instead I have a lot of data, and a lot of models and if there’s something that doesn’t makes sense I can go back to the people who informed my model and ask them to clarify.

“I’m not afraid to ask them, and I’m not afraid to be wrong either.”

Teams, like the Browns, have embraced analytics in many different ways says Frelund. Source: Phil Masturzo/PA Images

On the surface, especially compared to sports such as baseball, American football appears to be late in embracing analytics but Frelund says it has been around the game for a long time.

“I think the league has really embraced analytics now and the media narrative is that the turning point was the hiring of Paul DePodesta by the Cleveland Browns in 2015, because people were all ‘oh there’s the Moneyball guy coming over to football.’

“But I don’t think the practise of analytics — actually, let’s call it decision science because I think a lot of people think analytics is just statistics — in football has been around a lot longer than people give it credit for.

Each team has varying levels of investment in decision science, but it is there, and they just haven’t called it that. Take Ernie Adams, he’s been with Bill Belichick for nearly two decades and he does decision science; it’s a strategy component for the Patriots.

“And what I try to do with the podcast is, when you talk about being wrong, is bring transparency to the process. Now, I’m not trying to teach people how to code in R because I’d lose every single listener and Money would not pay attention to me, but I want more people to get into this.

“I think it’s something that can really bring our sport forward and I want people to be even better at it than I am and I want people to think it’s fun and exciting and use it however they want to use it.

“Ultimately I want people to have a better understanding of the sport and the strategy that goes into it.”

Subscribe to The42 podcasts here:

About the author:

Steve O'Rourke

Read next:


This is YOUR comments community. Stay civil, stay constructive, stay on topic. Please familiarise yourself with our comments policy here before taking part.
write a comment

    Leave a commentcancel