2019 Pre-Game Thread: Week 4 at Bills

[PedroKsBambino]

I do 100% agree that they have people that study this stuff. At the very least, it would be useful to know so that when they face teams that are analytically-inclined, it helps predict what they would do in those situations.

I think our dispute is how heavily they weigh it. They say they don't, others say they do, my read is "not very much."

I think it's also important to note that unlike baseball, where thorough public record-keeping dates back more than 100 years, most of the interesting data and information for football has been kept internal until very recently. All-22 has only been exposed since 2012, for instance. Belichick has 40+ years in the sport with access to all the information that insiders have. When new analytics studies come out, much of the time he probably has better information that contradicts or complicates some of it; probably some of the time he agrees with the studies but has figured out their conclusions years if not decades back.

It always bears saying when we talk about football analytics that NFL sample sizes are for the most part tiny and so most of it is either a) not analytically rigorous in a way that would hold up to academic scrutiny or b) has to capture so much to generate large sample sizes that it excludes basically all context and thus isn't actionable to someone like Belichick in any meaningful way.

Just saw an article on the Athletic speaking to above and figured I'd add---even though of course not really about the Bills!

A few years back, before the kickoff rule changed, one of those guys wanted to hit the New England duo with something they absolutely had never considered. That member of the analytics community decided to figure out how much win probability would be added if a team signed a soccer player to boot the ball out of the back of the end zone on every kickoff. No accuracy. No field goals. Just there to generate guaranteed touchbacks. After having met with Belichick and Adams a few times before, he was certain it was the one, the obscure study that was going to take them by surprise. So, he laid out the scenario and then asked them how much win probability they thought would be added if a team could be certain of a touchback on every single kick.

Belichick paused for a beat before answering with the exact number.

The Patriots had already done the same study.

This is consistent with reporting of other internal analytics work done in the past---it is not a one-off.

Edj Analytics, which works with several NFL teams to help them review data and make informed decisions, has worked to create a customizable model that allows for more specific analysis. What Edj’s models show is Belichick’s analysis of situations and subsequent decision-making is typically on point. Using a three-year average from 2016-2018, the Patriots coach ranked fifth among coaches, according to the company’s models. Payton, Baltimore’s John Harbaugh, Seattle’s Pete Carroll and Minnesota’s Mike Zimmer own the top four spots.

https://theathletic.com/1352879/2019/11/06/outsmarting-bill-belichick-good-luck-with-that-analytics-experts-find-pats-well-prepared/
The article then goes through some individual choices and how they compare to a particular analytic model. Of course, at the end of the day BB is making his own choices based on a number of factors---the Sashi Brown straw-man remains a silly one---but the alignment between his choices and analytics overall only reinforces the reporting that they are regularly doing analytical studies of their own in telling a pretty clear picture here.

 

[DJnVa]

I like that the analytics company said BB was like 0 for 5 in the Super Bowl win against the Rams.