6 replies to this topic

[BSLChrisStoner]

The Spread: The Sorry State of Football Analytics

http://thespread.us/sorry-state.html

[DJ MC]

It comes down to knowing how to develop analytics for the unique nature of the sport, and having the motivation to do so. Right now, it doesn't seem like either applies to the NFL.

[BSLMikeLowe]

Without reading it, I think the team-dependent nature of football vs other sports makes analytics much more challenging. There are 1-on-1 performances you can try to grade, but there's still variables there that only the players and coaches know (maybe an O-lineman gets beat on a play where he was supposed to have help that he didn't get, or same for a DB).

 

Also, and this is just my opinion but football coaches, with few exceptions, seem much more risk averse than their counterparts in other sports. They seem as though they would much rather have to explain why they failed using the old tried-and-true way of doing things than explain why they failed doing something more unconventional....even if the numbers show the unconventional way had better odds of success.

[Mike in STL]

The thing is with football that differs from other sports, is the teams come out in different looks on every play. In baseball, other than infield shifts, there are still 3 OFs and 4 IFs not counting Ps and Cs of course. You plan the matchups ahead of time, and every play starts the same exact way. Then the direction the ball travels off the bat is almost random.

In football, you can face 4th and 1 against a four man front, five, six, seven, eight, nine man fronts on the goal line. You can have five blockers on a spread look. Six if a TE or FB stays home. Seven with two tights staying home. The player has complete control over his direction.

People just say its converted at x% so go for it, not considering all of the variables involved. Or the game situation. Its not shocking that football guys keep discounting these math geniuses.

[Bmore Irish]

I mentioned this in the NFL Week 3 thread, but it seems appropriate to put here. Rodgers performance against the Chiefs last night supposedly got him a -0.8 grade on PFF. Seems to defy logic to me, and you can kind of understand how some coaches would be skeptical with results like this.

I think NFL coaches do tend to be averse to innovation, but there don't seem to be a ton of credible or reliable advanced stats out there right now.

[bnickle]

Yeah, it's ridiculous. Now someone could throw that back in my face because I just used PFF grades on Wagner to back my point in another thread. I would just say that I don't think he was nearly the RT PFF says he was last year. I want to at least give some credibility to these grades but how can you when they have Rodgers as a negative last night. Just no freaking way. He made so many plays in the face of pressure. I'm not sure any other QB even comes close to being able to do what he did last night.

[Mackus]

That's a bad article, IMO.

 

"Football coaches aren't using analytics because they think we're nerds" seems to be the thesis.

 

I do think there are some areas where coaches should trust the macro numbers more than the micro risk, and I've been very vocal about how I strongly prefer being more aggressive on 4th down when you're across the 50 and would never kick a FG under 20 yards in the first half unless it was right before the half (and you lose the silver lining value of backing up the offense if you fail).  I think teams would score more points overall if they went for it more in these situations, although it depends on which team.  I wouldn't suggest all teams do this, and in fact a team that typically has a mediocre offense, excellent special teams, and a good defense like the Ravens would benefit less from aggressive calling here than a team that has a good offense and a bad punter, for example.

 

I think football coaches across the board should use these types of weights and situational success rates in their decision-making.  I imagine that many do even if they don't realize it.  Ignoring the numbers intentional is doing your team a disservice, though.

 

However, when you talk about individual analytics of PFF grades and such, I don't think that is particularly useful yet.