Sunday, September 05, 2010


Inevitably when I bring up the subject of statistical analysis in football and how I think many false conclusions are drawn through the misuse of statistics, two things happen; the first is that most readers switch off mentally, preferring to eat out there own intestines rather than be forced to sit through what amounts to a lecture in applied mathematics. The second is that I wake up with an e-mail inbox overflowing with literally a couple of e-mails telling me that I'm a simpleton and that I should spend more time studying at Football Outsiders or AdvancedNFLstats. (I'm sure some of you are already preparing to send your pre-written hate mail, so for those that cant wait the address is here). Anyway, stopping by the aforementioned Advanced NFL stats I came across an article that includes a run down of an article from their seemingly hated rivals at Football Outsiders. Let the graph wars begin!! But there is a serious point to be brought up. Football number crunchers, for want of a better title, have a serious thing against the run in football. It's like the hatred NFL coaches had for the pass in the '60s... but in reverse. It would appear that most football number crunchers also seem to share a trait with the vast majority of high school coaches who frequent various Internet message boards, (no names mentioned), in that they all seem to assume that anything said by any analyst on TV must be inherently wrong in some way; doubly so if that analyst works for ESPN. The classic central point and counter-point between number crunchers and TV analysts (an argument held on the analysts behalf in absentia) surrounds the issue of whether running the ball and stopping the run is important in winning football games. The analysts believe that it is. The Number crunchers do not. I'm with the analysts on this point (hold the hate mail for a minute!) and I'd like to try and explain why. I'd like to draw on a combination of my own coaching experience, the experiences shared with me by others, what I've absorbed over the years from the idiot box, and what I've absorbed from various coaching materials (books, coaches forums, etc). First we have to highlight a point that the number crunchers/analysts/everyone seems to be at a consensus on; the greater the distance to go on third down, the harder it is to make the yards and get a first down. This is important because I believe that when analysts (even those on ESPN) talk about running the football and stopping the run, they're referring to three very specific circumstances; first and second downs (that counts as one), goal line & short yardage. I do not believe that they are suggesting for a second that a dominant run defense and a chronic pass defense will breed success (or ditto for a dominant rush/poor pass offense). So why is stopping the run such an issue on first and second down? Well again we have to throw in a caveat. When I hear "you've got to stop the run!" I understand that to mean "... with your front 7, without the need to drop a safety into the box". And it comes back to that first and second down argument. If you can stuff the run on first or second down, you improve the chances of forcing the defense into long yardage third down situations. Which means one thing; Le Passe! (I'm not entirely sure that's actual French). The defense sits back a little. Substitutions can be made, bringing in third down linebackers and linemen who specialise in either pass coverage or pass rushing. You can remove that clunky, run stopping DT (Terrence Cody anyone?) and insert a more lithe and agile individual, or at least as lithe and agile as defensive linemen get. You can tell your corners to sit back a little. Let them have the short stuff. If the offense dumps it off with a short pass then you come racing down and nail them short of the first down marker. This is what I've always understood as the primary advantage to being strong against the run. It's not about the fear of an 80 yard dash, it's the fear of the 5 or 6 yard scamper. That run that allows the offense to obtain a fresh set of downs. It's the knowledge that a stuffed run plus an incomplete or short pass puts the offense in a tricky and somewhat predictable situation. This is why we call third down the defenses down. It's the one down where they have the best possible chance of predicting what the offense will do, while having the best chance to make a play. Obviously stopping short yardage runs and goal line runs is important as well. Forcing the offense into third and 3 is of no use if they then run it straight over you. Similarly in and around the goal line it is necessary to keep the RB out and force the defense to throw in the much restricted area of the end zone. I'm not expecting this to start some kind of revolution. My only hope is that the number crunchers among you will see that sometimes there is more to football than the numbers. The underlying strategy and the many inter-weaving factors that go into football sometimes bring up unexpected goals and challenges. Things that go beyond just the statistics and the analysis. Tomorrow I'm going to do the stupid and attempt to predict the play off teams for this season. I can hear the sound of crashing and burning already.... Have a great day everyone.


Ian said...

"Things that go beyond just the statistics and the analysis."

As a stats guy, I would have to disagree with you. I can see why there would be so much resistance to stats from both the romantics (who believe football cannot be measured) and the coaches (who don't want a bunch of economists coming in telling them how to do their job).

Take your point. Essentially it boils down to 'good first and second down rush defense will win you games'. I did a very quick plot using the 2005 play-by-play data (the only season I have it for) and it is interesting to see that you are onto something. First and second down run defense correlates better with wins (r-squared 0.19) than overall rush defense (r-squared 0.15).

It's not huge, but it's a start. Perhaps it's the case that the stats guy need to use a little more subtlety when it comes to the running game.

Chris said...

Thank you Ian for taking the time to drop by and leave a comment.

I'm not completely averse to the statistical analysis of football, I just feel that if it's to be done then it has to be done properly.

When I say "Things that go beyond just the statistics and the analysis," I'm pointing at variables that are not immediately apparent when you look at a bunch of numbers, like the change in defense on third and long situations.

We also have to consider the human factor. Overall I think studying the numbers often fails to account for things like mental performance, such as tiredness or nervousness (in rookies for example).

So how would I change the statistical analysis of football? I would like to see the focus become more narrow. Over the course of a season many, many little variables (injuries, arguments, change of tactics etc) can effect the outcome. I would like to see analysis focus on things like specific down and distance situations, or on specific groups of plays.

For example, I'm not sure yet if there has been any study looking at first half rushing attempts and the corresponding success (or failure) of play-action passes?

I'd also like to see a push by the analysts to force more useful individual player stats into the mainstream such as;

catching percentage for wide receivers (and then an adjustment somehow to QB accuracy scores based on this); the number of interceptions by a DB compared to passes he has a chance to intercept; the number of passes defended by a DB versus how many passes thrown his way, etc.

Ian said...

I agree on that point - of course stats that are more suited to each situation would be preferable. As ever, the final driver of statistical quality is sample size. Too small and it's meaningless, too large and it's too general.

I always find 4th down studies fascinating. The stats show that pretty much anywhere on the field you should go for it on 4th and 1, but even armed with that information and a high-powered offense I'd struggle to make that call.

The one thing I think that the stats have shown pretty comprehensively though is that just about every NFL coach is too conservative with their play calling, and there are very few, if any, overly aggressive coaches.

Chris said...

A big problem faced by coaches on fourth down is survival, namely theirs.

Get it wrong on a fourth down call and you can guarantee that you will be in the dog house with both your owner and the press. And unless that drive ends with a score, it'll be seen as unnecessary risk taking.

Going back to the run defense on first and second down, I'd be interested to know how pass defense on third down, combined with run defense on first and second down correlates with winning.