Eagles analytics hire created competitive advantage

The secret sauce to the Super Bowl Champion Philadelphia Eagles’ historic run is analytics, I’ve often said this. With the application of analytics still in its infant stages, the Eagles are ahead of the curve. This is evidenced by them being the only team to hire EdjFootball for the 2017 season, a division of Edj Analytics. With any new advancement in the sport there’s bound to be confusion regarding how analytics are applied. Thankfully the success of the Eagles and other teams with a soft spot for game theory has caused an uptick in discussion surrounding the topic.

Howie Roseman spoke at the Wharton People Analytics Conference earlier this year, shedding some light on how the Eagles use player tracking, sports science and the drafting of Carson Wentz. Now we have another source from which to gain a better understanding. Frank Frigo, the Founder of Edj Analytics and Chief Business Officer of Edj Sports recently spoke with the gentlemen at Wharton Moneyball. The following is the notes taken from that discussion.

One quick note not discussed in the podcast before we begin. EdjFootball runs thousands of game simulations that provided the Eagles with situational, opponent specific data. This data aided them in making in-game decisions based on the success rates provided in the simulations. It gets as detailed as providing analytics for on-field match-ups and coordinator tendencies.

Their mission statement…

“EdjFootball provides NFL & NCAA teams with powerful predictive and prescriptive tools for game planning, play calling and post-game analysis.

Leveraging more than two decades of play-by-play data and rigorous mathematical analysis, EdjFootball assesses a team’s Game Winning Chance (GWC) at any moment of the game. Bring the power of EdjFootball to your team, with understandable and customized, actionable insights. The result is more confidence in your critical decisions, with a true competitive edge that impacts the bottom line – winning.”

On bucking conventional wisdom…

“Like any area where there’s new technology.. People anchor on conventional wisdom, they anchor on conventional metrics. We came from a world where it’s all about win probability, you’re not really concerned with score difference, you’re concerned with making decisions that give you optimal win probability. So I think for a lot of coaches they traditionally look at, ‘well how often do we convert a first down, what’s our expected points or our success on this particular drive?’.. Our modeling approach is saying it’s all about win probability.”

On their model and GWC…

“There are a lot of situations where teams march down the field, they’re in fourth and short deep in opposing territory and from a risk aversion bias they’re going to take the field goal because it’s the highest expectation of getting something on the drive.. But from a win probability perspective, what the model’s picking up on is what happens when they fail, what’s the resulting ball position, how does that then result in more favorable position for them down the road and the immediate expected points?

Situations like that can be very counter-intuitive to the way coaches traditionally think about the game because they might be thinking about maximizing points on the drive but the model’s picking up that you’re getting much higher win probability by a more aggressive action. And we see those as being pretty chunky differences in win probability.. in situations like that it’s pretty common to see errors that cost teams 10% GWC.”

On researching coaches…

“When I say (coaches) were doing worse than we originally expected, I think we would have thought that coaches were giving up to about a quarter to a half game a season in aggregate GWC. We started to look at this much more closely, we saw that it was actually north of half a game, into the two-thirds to three-quarters area.. that’s on really fourth down decision making alone which is where a lot of the bigger equity differences are.”

On the Eagles Super Bowl 4^th & 1 decision at their own 45-yard line..

”We thought that was right by a lot, that was 12-13% to do that vs the punt.. if they downed the ball at the 1-yard line, it still would’ve given up 6% in GWC. You can imagine the average NFL team if they had a punt of that success level they’d be high-fiving and all excited about it, not fully realizing they had just given up a chunk of win probability.”

On fourth down decision making impacting third down…

“When you think about setting up for fourth down in terms of third down decision making, I think it’s the example that the example that we talked about with the Eagles. Once you know that you’re going to behave optimally on fourth down and you understand the value of that, it sets things up very differently on third down. If you have a third and eight and you’re only picking up five or six yards and you know you’re going to go for it on fourth down, it opens up your play options and makes you a little less predictable on some of those third down decisions.”

On the Eagles application of their model…

“I do think the Eagles were a considerable anomaly this year in terms of how they treated these kinds of decisions.. They had really a buy-in from the top of the organization down to all the way throughout I think to say, ‘if it’s mathematically defensible, let’s do it, let’s not worry about the conversation in the post-game press conference, let’s do the right thing.’”

On surprise onside kicks…

“There’s other decisions, particularly around kick-off decisions. We have felt for quite some time that onside kicks are underused. There’s more scenarios than one would expect where they could be a viable alternative and boost equity.”

On coaches in-line with their thinking…

“The obvious one is [Bill] Belichick traditionally has understood it really well. Although I will say this season we saw some things that made us sort of reassess that.. The AFC Championship Game for instance, we had the Patriots, I think there were four decisions in that Jacksonville game where they gave up an aggregate of about 20% [GWC].”

On Belichick’s infamous 4^th & 2 against Peyton Manning…

“We concurred with the decision, but the big counterargument on that was, ‘well you just can’t give Peyton Manning the ball back if you fail there.’ Of course, he was trying not to give Peyton Manning the ball, he was trying to retain possession. By punting he’s definitely giving Peyton Manning the ball, he’s just got to go a little bit further. People tend to cherry pick those arguments and our point is in the simulation, we’re account for those things. We can make it Peyton Manning, we can make it somebody who is not nearly as effective as Peyton Manning and let’s see what that means.”

According to a source, there are at least two other teams jumping in the EdjFootball pool with the Eagles. Those teams are the Tennessee Titans and the Indianapolis Colts. More are sure to follow if teams using this new technology continue to have success. It’s a copycat league, but for now the Eagles are leading the pack.