Writers of Pro Football Prospectus 2008

02 Feb 2018

FiveThirtyEight: Sneaky Stats That Could Decide Super Bowl LII

At FiveThirtyEight, we looked at a few more stats to keep in mind for this Super Bowl, including New England's No. 1 ranking in weighted DVOA, its No. 31 ranking in defensive DVOA for the season, how Tom Brady has been dealing with pressure as he's gotten older, and how the game script has gotten so familiar in New England Super Bowls.

I did a very similar article last season for Super Bowl LI for ESPN Insider that looked at Matt Ryan taking too many sacks (he took five in the big game), the impact of receiving backs (James White had Super Bowl-record 14 catches), and how this could be the biggest comeback in Super Bowl history with the way the Falcons blow leads (28-3). I think this year's matchup is harder to predict with the Nick Foles factor and what to trust with New England's defense (first month vs. last 14 games), but it should be a good game.

Posted by: Scott Kacsmar on 02 Feb 2018

7 comments, Last at 05 Feb 2018, 5:06pm by ChrisS


by nat :: Fri, 02/02/2018 - 5:17pm

To make matters worse, Philadelphia’s worst defensive quarter this season was the fourth quarter, in which its DVOA fell to 14th after ranking in the top five in each of the first three quarters.

Great. Just great. Now I'm predicting an Eagles win.

I trust predictions based on fourth quarter DVOA less than any other quarter. Clock management, when it matters at all, means sacrificing VOA (i.e. the goal of most plays) for increased win probability.

by The Ninjalectual :: Sat, 02/03/2018 - 2:04am

Doesn't DVOA account for that?

by Kulko :: Sat, 02/03/2018 - 2:49am

No DVOA does not consider the clock on the necessary detail. So teams with a big lead running out the game will be penalized.

by nat :: Sat, 02/03/2018 - 7:23am

Yes and no.
DVOA compares you to average. But if you do clock management more than other teams, by which I mean you sacrifice VOA for increased win probability when other teams wouldn't, then you look worse in the stat.
Also, if your clock management schemes are more average than your usual scheme, but still increase your win probability, it hurts your overall DVOA, but still is better than the alternative.
All this gets confused by the possibility that you do clock management schemes too early, and don't get an increase in WP for your pains.
That's why I trust 4Q least, rather than saying it's just wrong. It's untrustworthy because all 4Q stats are noisy for predictive purposes.

by The Ninjalectual :: Sat, 02/03/2018 - 11:11am

I thought http://fivethirtyeight.com/features/how-tom-brady-became-the-king-of-com... was more interesting. It suggests that the Brady Comeback Express is a recent development, which means to me that it's a small sample size, and the Pats have merely been hot recently

by The Ninjalectual :: Sat, 02/03/2018 - 11:15am

And https://fivethirtyeight.com/features/the-patriots-defense-needs-30-minut... gives Pederson a framework what to expect from the Pats halftime adjustments. It means Pederson merely has to come up with two or more winning gameplans

by ChrisS :: Mon, 02/05/2018 - 5:06pm

Terrible headline, Stats do not decide games, it seems cliick-baitish. Stats indicate what strategies could be determinative or where one team has an advantage.