First – how did you get started in analytics coverage?
I started off as a traditional newspaper beat writer. I covered the Jets mostly – some Giants coverage too – for the New York Daily News. I always had a quantitative interest though and paid attention to what football analysts were doing. In the press box my fellow writers would remember me as being the crazy guy who said they should always go for it on fourth downs.
In 2017 I left for ESPN. I didn’t have the technical skills for a job in analytics- that ruled out 99% of the jobs I wanted. But I was lucky to find that one-percent job that opened up at ESPN. I didn’t have to do the modeling there – I used ESPN’s metrics to write stories instead.
It was exactly what I wanted. I shifted now to covering the NFL only which was what I wanted. I’ve been able to expand my own technical abilities too. I’m not as skilled as the analytics guys but I can do some analysis for myself now.
I saw you select Derrick Harmon as your DROY prediction -why?
I don’t do a lot of prospect evaluating but I do look at pass rushing and have created a sack-rate forecast model that predicts sacks numbers for the first few seasons of a player’s career. It uses data – taking into account things like their college pressure and pass rush win-rates, quality of school they went to and played against, combine information…. Harmon graded strongly – he was sixth in his class. As a defensive tackle, getting sacks is more impressive than someone who gets the same number of sacks as an EDGE rusher, and the DROY awards pass rushers mostly. And playing next to Heyward, Watt, Herbig and Highsmith should give him some good chances to get sacks.
You have mixed thoughts on the signing of Rodgers though, correct?
I’m not very confident in the signing. He played poorly last season and played poorly in 2022 – so it wasn’t just about him recovering from his Achilles injury last year.
He’s a prayer for the Steelers – he could work out. The most likely outcome is disappointment – but there could be upside even at his age if things break right. You can talk yourself into different aspects – maybe it’s a better system for him, maybe he just needed more time to recover from his injury. With the price tag and the alternatives, I do think it was worth the try. It’s a pretty palatable move – they didn’t hurt themselves for years to come doing it. You can’t just run out the clock on Watt I guess – you have to try something.
I do think they should have taken a quarterback sooner in the draft though.
Many of the teams that have the biggest analytics staffs are not at the top of NFL – is there something they are doing wrong?
I’d push back on the correlation on staff size as a way to help determine win percentage. It’s one of many elements of building a team and doesn’t necessarily indicate team buy-in. It’s like saying having a good safety is the difference between good and bad teams. It’s one element. It generally does serve though as a barometer for analytics buy-in.
What’s the best use of analytics you see right now?
The areas where teams can gain an advantage using them are numerous. Player evaluation – now teams have metrics to evaluate players – pass rush win rates, pass blocking measures like sustaining blocks…when those came out they were novel and now teams use them regularly. Teams are measuring entire games not just individual plays. For example the ability for receivers to get open. It’s not just about separation on receptions. You can’t just evaluate them when they are targeted – you need to evaluate them on all routes. We have the data to do that now for better metrics.
Play selection is another big one. Which run plays are more likely to be successful against certain fronts. Some of them seem obvious but teams still don’t always do them. Like trading down in the draft – that is shown to be a dramatic advantage for teams, but teams often won’t do it. Even though it’s staring them in the face as an advantage.
Where can the Steelers improve on its use of analytics and how do you marry the “what you see on film” with the analytics?
It’s hard to see how teams use data and analytics – they keep that to themselves. But we do a survey every year with NFL analytics staffers on what teams are most analytically advanced. The Steelers, of 19 total votes, got four. That is tied for second least. So we do know that the NFL analytics staffers see them as near the bottom.
No one says scouting and watching tape shouldn’t be a significant part of player evaluation. I think of it more as the qualitative part of the process. You need to look at scouting calibration when you give a grade. What’s the range of outcomes when you give a certain grade to a certain position. You can quantify that over time to help determine the risks in selecting certain players with certain grades. You can see if you are better over time at drafting certain positions over others that way too. You can see if how you rated character, for example, works over time.
Basically, quantifying your qualitative measures?
Exactly.
What are the trends you see working right now in the NFL – what excites you?
What excites me most is when we learn something new – finding some piece of information that suggests that we should change the way a team should do things differently. Something that will make us look back 10 years from now and say “I can’t believe we weren’t doing that then!” Like using motion pre-snap is now something we look at that way.
What do the Steelers need to do differently as an organization, from your perspective?
Finding a better quarterback is the issue. You don’t want to be results-based. Just because Pickett didn’t work out doesn’t mean you don’t try and draft another quarterback. But it all doesn’t just come down to quarterback.
One thing I will say is I am a big Nick Herbig stan. I think he can become a monster for the Steelers.