Go for it on 4th down

Submitted by river-z on

The article at the link makes 4 points about the hazards of decision-making in football.  The study used data from the NFL, but it's definitely relevant to the college game.  

In particular, it seems relevant discussion on MGoBlog about:

1. going for it on 4th down (do it!)

2. whether it helps the team to fire the coach (not really, but there needs to be a GERG exception, IMO)

3. winning with someone else's talent (WOO! go Brady, Borges, and Mattison)

http://www.theatlantic.com/business/archive/2011/11/4-things-the-nfl-ca…

2plankr

November 16th, 2011 at 6:16 PM ^

Its a synopsis, but he does include

"If a team was on its own side of the field and needed four yards or less for a first down, going for it was the right move. At the opponent's five yard line, the chance of going for it and scoring a touchdown was more valuable than the guaranteed points from a field goal."

bo_lives

November 16th, 2011 at 6:42 PM ^

I really don't think having a good grasp of statsitics is a pre-requisite to coaching football. Les Miles doesn't really seem like a guy who took Stats 400 at U-M.

But this is an old point regarding a very tired old topic. Nothing new in this article at all.

2plankr

November 16th, 2011 at 7:33 PM ^

I tend to disagree.  He may not know a thing about statistics, but his decisions tend to maximize expected value.

Now he may not know how to read a digital clock, but however he got his "feel" for when to go for it, it aligns pretty well with the statistics

Likewise I have no idea how much Hoke understands the numbers behind it, but he clearly seems to know what works in this regard

umchicago

November 16th, 2011 at 7:14 PM ^

in general, i like being aggressive on 4th down.  in the falcons case, though, it was sudden death OT and missing that conversion puts the opposition in FG range already.  if Atl was another 10 yds downfield, i wouldn't have a problem with that decision, as the opposition would need to get a first down to get into field goal range.

BraveWolverine730

November 16th, 2011 at 7:22 PM ^

How many times can you punt to Drew Brees with a below averge pass defense and expect to hold them? Especially since you have an above average rushing attack against a poor rush defese.  I feel statistically it would be backed up as close to 50-50, but the intangibles of the scenario in my opinion push it to going for it,. 

treetown

November 16th, 2011 at 11:25 PM ^

It is always good to see smarter and better stats used in sports, but game context has to matter.

Why not analyze going for 4th down in the same "red zone" fashion as the regular offense?

The risk of turning over the ball on 4th down when there is at least 50 yards to your own goal line probably makes sense if your defense is pretty good. Most teams can't reliably drive 50 yards every time. Game context - time remaining, score and situation (OT) clearly all can affect these calculations.

ATLWolverine

November 16th, 2011 at 9:44 PM ^

It was risky, but given Brees' skill and perhaps other things we are not privy to (Smith feeling his defense was sagging, having used that same power play to success earlier or throughout the season, etc.) I don't see what the uproar is about.

Again, if Falcons convert, he's a genius-- but he didn't, so every forum-poster and sportswriter snipes at him until the next game or succesful gutsy call. Football!

turd ferguson

November 16th, 2011 at 6:49 PM ^

I think it tends to be misapplied around here.  The Romer paper - which, I think has problems, but that's for another day - is very good for telling us whether, across the board, coaches are too conservative, aggressive, etc.  It's not very useful in determining whether a particular decision in a particular game was right or wrong.

Around here, people will run some numbers and then declare a Michigan decision "right" or "wrong."  That's an inappropriate use of statistics.  The circumstances of the situation have massive effects on the relevant probabilities.  For example, quality of offense, quality of defense, weather, quality of plays available, injury/fatigue, and so many other context-specific things make it almost impossible to declare a decision right or wrong when it's relatively close in the expected value numbers.

I thought Mike Smith made the right move in that Falcons-Saints game.  If the same situation had arisen in the LSU-Alabama game, I probably would have felt differently.

AAB

November 16th, 2011 at 6:52 PM ^

Coaches are departing so far from what EV would dictate that it would take a fairly extreme situation (say, Penn State against Alabama on a 3rd and 8 from its own 42, or something) for me to conclude that conventional wisdom is more correct than Romer.  

turd ferguson

November 16th, 2011 at 6:57 PM ^

Our views are probably pretty close on this.  Here's where I'm at:

The starting point for what's the right decision should be something like what's in the Romer paper.  In other words, if you have a median offense against a median defense in median weather with median fatigue... go with Romer.  I think the starting point that people use (conventional wisdom) is much too conservative.  Still, there's enormous context-specific variation that needs to be accounted for in order to intelligently play the expected value game in a particular situation.  I think that people around here tend to be very good at understanding the starting point and not as good at understanding the relevance of context.  (Note that that means that "people around here" are just as likely to be too conservative, in my view, as too aggressive.)

I'm definitely not arguing that conventional wisdom > Romer.

AAB

November 16th, 2011 at 7:03 PM ^

but here's where I'm at:

For the most part, people (including Brian) aren't really complaining about marginal decisions where, according to Romer, going for it on 4th and fairly long from your own 38 or whatever might be slightly +EV.  People are like "oh my god stop punting on 4th and 2 from your opponent's 45, especially when you're down a touchdown in the 4th quarter."  

To use a poker analogy, coaches, by and large, aren't being criticized for failing to make marginal but probably +EV calls with Q7 when their opponent is shoving all in with a really huge range of hands.  It's more like the vast majority of coaches are folding pocket Jacks when they have 10 big blinds.  There's basically no set of circumstances where that decision is the correct one.  

alanmfrench

November 16th, 2011 at 8:21 PM ^

love to see a div 1 program adopt a go for it on every 4th down no matter what philosophy. i think it would make for some really interesting games. youre offense would have to be good though or it would just be ridiculous.

ATLWolverine

November 16th, 2011 at 9:38 PM ^

http://sportsillustrated.cnn.com/2011/writers/scorecasting/09/15/kelley.pulaski/index.html

He also refuses to return punts because the risk of getting stripped (in HS at least) is apparently too high to justify the sparse return yards generated.

Gus Malzahan started as a zany offensive-minded HS coach once  upon a time and a lot of his spread precepts are gaining traction in Div. 1. Who knows, your wish could come true sooner than you think

LB

November 16th, 2011 at 9:16 PM ^

understood and related the MGoLiveBlog's ability to will interceptions. Any stats that do not take that into account can't possibly be taken seriously.