Q&A: Previewing Super Bowl XLIX with Brian Burke of Advanced Football Analytics

When we’re done with everything about atmospheric pressure, media silence and other sources of outrage, there will be a football game to play, and it should be a good one. The Patriots and Seahawks are seen in Las Vegas on near even terms, and Super Bowl XLIX is an attractive matchup of teams that have been there before.

To break it all down, we’re joined by Brian Burke of Advanced Football Analytics, which features statistical breakdowns of the players and teams involved and will have lots of live content come 6 p.m. ET on Sunday.

Burke will also be a panelist at the 2015 Sloan Sports Analytics Conference, February 27-28 in Boston.

Some of the questions and answers have been slightly edited for length and clarity.

Q: First question. Am I allowed to do a football analytics Q&A without asking about going for it on fourth down?

A: That’s the low-hanging fruit of football analytics, probably the thing that has the biggest impact immediately. And it’s what people see themselves – it’s not like it’s some kind of salary cap analysis, it happens right on your TV every Sunday. (Laughing) Some people are sick of the topic, and I understand that.

Q: OK, let’s talk about the game. One of the things that stood out to me in your statistical breakdown is that I think a lot of people just watching TV coverage wouldn’t have known that (Patriots linebacker) Jamie Collins was the most valuable defensive player on either team, according to win probability added. What has he done that’s been so valuable to the Patriots?

A: He’s had a big impact. He’s been one of those big guys that can tip a pass or have a pass fall into his arms and he has those three interceptions, which is really unusual for somebody with a number like 91. Even if you throw away those interceptions, he still leads the team in what I call success count. (Success count is the sum of sacks, interceptions, passes defended, QB hits, forced fumbles, fumble recoveries, tackles, and assists that result in team-level successes, explained here)

So he’s a grinder. Occasionally, a guy will have a big win probability added because he got lucky and had one pick six or something, but he’s not that kind of guy – he’s definitely a difference maker.

Q: What is the Seahawks’ best option in the pass game, not necessarily a type of player, but do they need to do anything different in style against New England?

A: If I were Seattle, I wouldn’t change a thing. Their formula is kind of a low-risk type of offense, and it starts with the run and relies on Russell Wilson to make good decisions and good reads. So I wouldn’t change a thing. They don’t throw downfield a ton, but if that’s the read, do it.

Both teams run so effectively that they can stay ahead of the chains, stay ahead of the schedule and keep from being in the position where you’re going to make the big mistake.

Q: How does Seattle go about controlling Rob Gronkowski?

A: There’s no one way to contain Gronkowski. He’s going to get his catches; he’s that big. He’ll have four catches, but the key is to make him have four catches for 40 yards and not four catches for 120 yards.

What Seattle does great is that they make the tackle on first contact; they get the guy to the ground right away. They’re going to play their scheme. They’re not going to do anything special and put Kam Chancellor or anybody right on Gronkowski glued man-to-man all day. They’re going to run their scheme, but the key to stopping Gronkowski is to contain him, meaning getting him to the ground right after the catch.

Q: Let’s talk about something that a lot of people were discussing the week, the issue of the Patriots’ fumble rate. What was your take on the statistical back and forth?

A: The original stuff came from Warren Sharp. He had some very good insight, and I think it definitely warrants some attention and warrants some further digging. What it did not warrant, though, was this ad hominem attack from people with Ph.Ds who were far guiltier of the kind of exaggerations of which they were accusing Sharp. Ph.Ds should be above that sort of thing. That’s what Ph.D means -you’re kind of trusted to be a scientist about these things. The discussion turned into political advocacy research, like the kind of research about climate change or certain foods that are bad for you – some industries call it science, but it’s advocacy. Both sides of the debate are very guilty of that.

I think the original insight is worth looking at. Why deflate the balls? It helps you hold onto them better. So let’s go check if that’s true. Are the Patriots holding onto the ball better? Yeah they are. Are they a six-sigma outlier? Or are they a four-sigma outlier? It’s worth checking into.

It doesn’t prove anything. I think the standard of proof should be a lot higher than that if you’re going to accuse somebody of cheating. On its own it doesn’t prove anything, but wrapped up in the broader context of what’s going on, it’s intriguing and it warrants more digging.

Q: How much of fumbles, and fumble recovery percentage, are luck and how much can be predicted by traits that we know about how teams run their offense?

A: Sack rate really drives it. Running back-type fumbles and wide receiver-type fumbles are really, really random. A helmet hits a ball at a bad angle. Recoveries are almost completely random – the closest thing you can get to a coin flip in sports outside of the actual coin flip itself. It’s 50-50. It kind of depends on where it happens, whether it’s in the backfield or downfield, but that’s a tiny difference.

What really drives fumble rate is No. 1, luck and No. 2, how many sacks you give up. It’s the strip sacks. I was misfortunate enough to be a Ravens fan for a long time when they had Kyle Boller at quarterback, and the guy just fumbled all the time. A guy like Joe Flacco is much better at ball security and gets rid of the ball quicker, so that drives it as well.

I’ve looked at the numbers and when you account for sacks, New England is still No. 1 in the league in whatever time period you want to look at. They’re the best at ball security for whatever reason, even accounting for all these different things.

Q: What have you thought of Pete Carroll as an NFL coach tactically? Has there been an evolution with him between his first tenure and now?

A: He doesn’t stand out either way to me, tactically. Both him and Bill Belichick are guys on their second lives and both probably learned a lot. Given a second chance, you say ‘I’d do this differently; I’d do that differently,’ and they got that second chance.

He doesn’t stand out, but he’s avoided making big blunders, which you can’t say for everybody. Even Bill Belichick in a Super Bowl (this one), he went for it on 4th-and-13 from the Giants’ 31. It was just a big head-scratcher. Maybe his kicker was hurt and he didn’t tell anybody. But there are some head-scratchers out there and so far, Carroll has avoided those.

Q: I remember as the baseball playoffs went along, much of the analytics community was rooting against the Royals, thinking they were a team that got lucky and was winning despite poor strategy. Is there anything like that in football, teams that more analytically minded people want to see having success or failure?

A: The community is not nearly as large as the baseball community, so I would hesitate to say that there really is much of a community. I could sense it; I’m a Baltimore kid, so I was amazed at how lucky the Royals really were.

I’ve never really thought about it. I think we’re rooting for the teams that are dipping their toes in the water with analytics – the Jacksonvilles and the Baltimores. New England has always had a good reputation; they do everything very secretly, so it’s hard to tell what goes on behind closed doors. And of course, we always kind of boo the coaches that pooh-pooh numbers and statistics when asked in their press conferences.

Q: Finally, what’s your prediction for the game?

A: My numbers have Seattle as slight favorites. Actually pretty fair favorites – 61 percent. But the model that I use has been under-predicting New England all year, so I think it’s probably halfway between that and 50-50. Maybe a 55-45 game for Seattle.