First, can you let me know how you got into the football analytics business?
Well I graduate with a civil engineering degree and worked in construction for many years. That gave me a background in math and problem-solving. But my passion was in understanding why NFL teams won and lost games – what the numbers were behind those who won more and less.
I built my own models and analysis for sports betting and predictions, for sportsbook professionals…. they became interested because I was doing a better job forecasting games than they were at times.
What made your work so successful?
I incorporated more data – play-by-play data and studied it. I realized that no other website showed that data – but Sharp Football showcased it and visualized it – showed the perspective of different things in color-coded fashion – things that you could filter by choice. For example, instead of just seeing how a quarterback did by yards per attempt, you could filter them by first or second down, first or fourth quarter… you could also sort by game charting data too – was there pressure on the play? Was it play action? What was the personnel …
After that several teams reached out to talk to me about the data and hiring me on to lead their analytics teams, but I decided to keep my options open and remain a consultant and keep sharing the data publicly.
I went to the combine and met with a number of offensive coordinators and other coaches and now consult for a number of teams. I couldn’t be happier with the way things are going now.
You’re 2020 football preview has just been released – what doe sit entail – and any fun details there as they pertain to Steelers?
One of them is lack of play action – they ran play action the least in the NFL. Across the league the numbers show play action woks. No matter what it keeps linebackers guessing. There is no way using it should hurt.
Also they have issues with predictability. When Snell was on the field they ran 78% of the time. When Samuels was on the field they passed 78% of the time. When your predictability levels near the 80% mark defenses eat that up.
Furthermore, when Snell and Holton were on the filed together they ran 92% of the time. Its an egregious level of predictability – no wonder their runs were unsuccessful. What’s frustrating is they found a way to win games despite their quarterback issues. That is impressive. But the frustrating thing was that they didn’t give their quarterbacks much help from an strategic perspective.
What should they have done?
Look at Andy Reid in Kansas City with Mahomes. Mahomes can do anything I know. We remember the big splash plays he made but that is a lot of mental recall. The thing he did most was make the easy passes with play-action and motion. Reid made it easier for his quarterback by designing a lot of plays that were easy to complete – using a lot of movement and play action. The Steelers didn’t do any of those things.
In the offseason they introduced Matt Canada to the offense. He has lots of ideas that I’ve seen work before – they could be effective in Pittsburgh. The trouble is you have an offensive coordinator in Fichtner – how much will he incorporate from Canada into the offense. I feel like Canada’s suggestions may be too foreign to Fichtner and that it could lead to power struggles like what happened in Cleveland. It’ll be up to Tomlin to have a specific set of expectations for both guys
The game used to be more about the strongest and fastest guys. – winning battles. How important is innovation to team success now versus the who wins the man-on-man challenges?
Absolutely, it’s more important now. You saw Le’Veon Bell led the team in rushing, but we found that passing the football is more important than running the ball in winning games. It’s less important t have a stud running back now. It’s more important to have a good offensive line and calling running plays at the right time- it doesn’t matter as much who the running back is.
How do the Steelers stack up in their use of data to make decisions?
I think they are falling behind others in their use of data and analytics. It’s hard to get get players and keep them. The teams that are using analytics best are helping themselves the most.
Who is using it well?
Baltimore uses it well – though it’s ironic because they run the ball against the better judgement of passing it more. But they are doing that a different way and using analytics to help be more aggressive on when to punt, when to call plays and personnel packages, and other game decisions.
Philadelphia and New England use it well too. Philly and Baltimore talk about it more openly. If you ask Belichick about his use of analytics he’ll tell you he doesn’t know what the word even means. But he uses it. He’s also blessed because he has so much he can use his intuition more than most coaches could.
What from your analysis should the Steelers be doing more to be successful?
Use play action more. Be less predictable with personnel groupings and use more pre-snap motion and design plays off of that.
I know they had quarterback and running back injuries, but they have a good offensive line and good defense. With a few intelligent changes on offense they could have a huge improvement. Lighten up on the predictability and use more pre-snap action and play action.
I also like the addition of Derek Watt – they can run more 12 and 21 personnel with him and throw out of that – he was good in that role in San Diego. The pieces with Ebron and Johnson and the rest of those guys – they are there for the Steelers. The problem is that making the kinds of changes that I was discussing this offseason is a question mark as teams don’t have a lot of time to practice. The Steelers do have an advantage over most teams in not having many new coaches or players they are counting on, but they need to make those chances to have success and get into the playoffs.
Why do you think a guy like Fichtner is reluctant to adopt a less predictable approach to the offense?
That’s a good question. It may be due to a lack of understanding of how effective those strategies are. Daniel Jones- when you look at his splits, he was bad at play action under center. But in the shotgun he was very successful. Better than when he did a straight drop-back pass.
You have to be willing to experiment a little more. The data shows it works. Maybe he thinks you have to establish the run first before you can pass or use play action. Or that play action just works a couple of times a game. But that’s not true. The data shows that play action works as well in the first quarter as it does in third or the fourth. The defender is reacting the same way throughout the game. It still gives you an edge, no matter what.
The Steelers have an easy strength of schedule – how does SOS usually play out when you look back at those teams?
I look back at my strength of schedule predictions every year- I track that. And I’m usually spot on. There is definitely a correlation between strength of schedule and the success of projections for teams. The thing is, we don’t look at win-loss percentages when determining strength of schedule. We look at teams in other ways. It’s valuable as a win predictor. The only time it isn’t is when injuries occur – usually to quarterbacks. Sometimes teams end up playing a number of teams who lost their starting quarterback, or their own got injured.
What does the data and analytics landscape in terms of NFL usage look like in 5, 10, 20, years from now?
I think you will see it will change what happens more on the field. You’ll see teams be more aggressive – more double passes and downfield laterals as analytics show those to work. Things like that. More aggressiveness and interaction between the booth analytics guys and coaches on gamedays too. Not just on when to punt or not but with in-game coaching on play calls, as well as more efficient pre-game planning.
It comes down to getting more points per drive for offenses. Every team will have different philosophies, but they will look to adopt more wrinkles to what is already working.
So a lot of talk about how analytics help offenses. What about those poor defenses?
Those poor defenses yes. They will always be behind offenses due to the rules constructs of the NFL that favor passing and points. The only edge defenses have is if an offense is too predictable or has a bad quarterback.
What I see more of is that when offenses use pre-snap movement the successful defenses are the ones using late movement when adjusting to offensive movement and who use simulated pressure more. Teams like the Steelers who use the 3-4 – that helps them and gives them an advantage at times when offenses don’t know where the pressure is coming from.
And maybe a greater emphasis on defense on taking more chances on getting turnovers since stopping offenses is increasingly more difficult?
Yes, but maybe even more so, for defenses to stress more scoring off of turnovers. The Steelers do a great job of this – I Tweeted about this not too long ago about Tomlin prepping his team about this before the Super Bowl, when Harrison ran the interception back for a touchdown. You’re going against guys not accustomed to stopping people from scoring so you have to take advantage of that opportunity.