Comments on Advanced NFL Stats: What Makes Teams Win? 4

Good analysis as far as it goes, but I think there...

2011-12-19T13:02:33.486-05:00

Good analysis as far as it goes, but I think there are a couple of stats that aren't measured here. One would be "big plays" meaning long gains and another might be "trick plays". I have a feeling that teams that attempt more trick plays have an edge. Lots of games are close, so trick plays could boost offensive scoring a bit.

Brian: Have you ever comprehensively reconsidered ...

2011-10-21T11:39:07.012-04:00

Brian: Have you ever comprehensively reconsidered this analysis? I know you have since observed (or at least speculated) that "Run Success Rate" is a better predictive factor than "Run Yds/Att." I wonder how that substitution changes (improves?) the overall regression model.

If you have revised the analysis, a short note and link at the top and/or bottom would be helpful.

This is super simplistic, but couldn't one say...

2011-01-18T21:58:16.316-05:00

This is super simplistic, but couldn't one say that in order to win one must score, and that in order to score, one must move the football toward the other guys endzone? If you add up total yards gained through the air in 2010 = 113,400. Total yards gained on the ground = 58,600. 113,400/58,600 = more or less 2. So they move the ball twice as much toward the other guys endzone passing than rushing. When you look at Brian's z coefficients, passing is roughly 2.5 times more explanatory of winning. similar. even for a non statistician dumb truck like me it seems pretty straight forward that passing will be more important than rushing.

Very good article! Yards per pass attempt is a gr...

2009-11-29T12:18:21.720-05:00

Very good article! Yards per pass attempt is a great statistic, as it takes into account yards, attempts, completion percentage, and sacks. It's a shame we constantly hear offenses (and QBs) judged on completion percentage or TDs. You don't have to be Tom Brady to complete 80% of your passes - if they're almost all 2 yard routes or screens to Welker.

It's also nice to see a statistical basis for the argument that running does not "set up the pass" as much as people think. I do think there's some effect there, but it's a fairly insignificant factor when compared to your team's actual skill in running or passing.

Analysts love to go on about how the defense was "ready for that play", or was "expecting a run" and that's why they stopped it. I think it doesn't actually matter all that much, and it's mostly just a case of the defender beating the offensive guy man-on-man. Defensive front sevens are taught to play the run first. DBs play the pass first. You can't just "guess" as a defense, on a consistent basis. This would help explain the lack of correlation between run and pass efficiency.

I just found your site and over the last 2-3 days ...

2008-12-09T15:06:00.000-05:00

I just found your site and over the last 2-3 days I have enjoyed reading it.

Your assumption that running would be more important in the first half will, I think, turn out to be mistaken. "You have to run and stop the run" is 4th quarter maximum. Clock management to increase or reduce the number of possessions as needed at the end of the game is an important strategic element.

In the same way you divided the field at the 34 yardline based upon when coaches kick the FG, go for it or punt, there is a time in the game, based upon the score, when it should be the right thing to run the ball to shorten the game. The obvious and extreme example is taking a knee at the end of the game. At that time, a change in game theory driven play calling seems to be in order. "Time becomes a factor" to cite a cliche, and the run/pass ratio (per down/distance and field position) changes. Until that time, I think your recognition that the pass is more important is correct.

I recall Cowher saying that he strove for 50/50 run/pass on 1st and 2nd downs until the game was "tilting" one way or the other. 3rd down was dictated by "to go" and field positions and after the game was "tilted" run or pass as needed.

To extend the point, I wonder if coaches' abilities in game theory can be extracted from the "luck" component you cite in your analysis of ability vs luck? Strategy has to play a role.

Sorry about the length and lack of focus in the post, but after reading most of the recent articles, this was on my mind.

Also, you might want to limit it to first and 10 i...

2008-11-18T08:51:00.000-05:00

Also, you might want to limit it to first and 10 in the first half.

I imagine that the runs stats are biased downward by, 2nd and 3rd and short runs, runs late in the game whose goal is to run out clock as much as gain yardage, and even kneel downs (don't know how they are counted). Seems there are a lot of strategic considerations that may be holding run efficiency down.

Also I wonder how the distributions of runs and passes play into the passing premium observed. Certainly passes have a higher variance in yards and have relatively fatter tails, you seldom lose 10 yds on a run but you do so quite often on a pass, you also are less likely to gain 30+ yds on a run but can do so more often on a pass. I am not sure how this affects the utility of runs vs passes.

The reason I wonder these things is that the passing premium paradox seems too large to be explained by coaching orthodoxy. Many coaches certainly aren't afraid to pass and have been passing more and more, yet it still persist and persist in college as well. While I like to believe the numbers are correct and the quants are on to something the coaches don't see, I also take seriously that the guys making the decisions have been doing this for most of their lives and really want to win. Being the son of a coach (and also an economist) I can't shake the feeling that the coaches know or feel something that is not showing up in the numbers.

I always enjoy bouncing these ideas off of my father who is a 30+ year high school coach and a math and statistics teacher. He can understand the math but also can understand where you might be missing something.

That's a great question, but I haven't looked into...

2008-11-17T20:47:00.000-05:00

That's a great question, but I haven't looked into it yet. I had been pondering a similar question--how does relative team strength (measured by record or point spread or something) affect the weights of the variables.

That's a great idea to limit things to the first half. I would almost guarantee running would be more important than it is for the game as a whole simply because there is more of it in the first half.

Brian, I really like your site and the analysis.
2008-11-17T20:34:00.000-05:00
Brian, I really like your site and the analysis.

I have one question that maybe you have looked into before.

How much does game situation influence the relative run pass efficiencies.

If a team is up they tend to run the ball to run out the clock and the other team knows this so you would expect run efficiencies to be low in that situation also they usually quit passing or pass less in those situations.

Does the difference between running and passing still exist when you only look at the first half of close score situations.

That's an interesting way of looking at it, and I ...

2007-12-09T17:03:00.000-05:00

That's an interesting way of looking at it, and I think it's a valid way of comparing the importance of defense and offense.

But is 30 pts per game the offensive equivalent of holding opponents to 0 pts per game? I would think it would be more like 42 pts per game or so.

A shutout, by definition, is holding your opponents to a point value your own team could never get below (awkward English, I know). That is, your own team could never score less than 0. So wouldn't the offensive equivalent be scoring at least as many points as an opponent could ever score.

I think you raise an important point. Defensive performance is generally bounded--it's practically impossible to do any better than zero (yards, pts, 1st downs, etc.) But offensive performance is unbounded--there is no practical maximum, which could explain why offense appears more important. But that also could explain why it is more important, if it really is.

Actually, I would guess that defense is just a bit...

2007-12-09T11:31:00.000-05:00

Actually, I would guess that defense is just a bit more important, and here's why... In today's NFL, if you allow 0 pts per game and have an average offense, you will likely go 16-0. On the other hand, if you score 30 points per game, and have an average defense, I would imagine your record would be more like 14-2. Both are great looking records, but of course, one is just a bit better.