Comments on Advanced NFL Stats: Game Theory and Great Running Backs

This is an excellent post. It explains the footba...

2011-11-15T14:22:16.867-05:00

This is an excellent post. It explains the football cliche that you need to make them respect the run. passes are better than runs but if you pass every time they will just run a pass defense every time and you are SOL.

well put it like this adrian needs to get the ball...

2008-09-23T20:09:00.000-04:00

well put it like this adrian needs to get the ball more on 3rd and jackson needs ta scramble more

Luft-No I haven't run the actual numbers. There ar...

2008-07-08T10:16:00.000-04:00

Luft-No I haven't run the actual numbers. There are some issues to solve first. You'd have to have a lot of good data of rushes when the defense knows the play will be a rush, and of passes when the defense knows it will be a pass. Plus, we'd have to have a separate analysis for each down and distance situation. Lastly, we'd also need a valid utility function, because yards do not = utility.

You state that the NFL's normal rush to pass ratio...

2008-07-08T08:34:00.000-04:00

You state that the NFL's normal rush to pass ratio is about 45% rush, but in your example you use 63% and qualify it by stating it is, "an example for demonstrating how game theory would solve the issue using notional numbers."

Have you run the real numbers, in particular for the Vikings and do they come out closer to the league average of 45% or the actual percentage that coach Childress/Bevell called running plays?

Nice blog. I enjoy the science of the game and stats analysis breaks that down. Advanced NFL Stats has just been added to my reading list.

This is great stuff, both this post and the previo...

2008-06-30T10:40:00.000-04:00

This is great stuff, both this post and the previous one. (As a side note I am now linking up your blog to mine. Hopefully I post with more frequency.)

On this topic I'm trying to think of what might happen if you had teams that were, in some abstract sense, already heavily skewed. I'm thinking of say a wishbone type team or a pass-happy college team like Texas Tech. I suppose that would just change the initial equilibrium because the payout for throwing against a run D might be higher (say 15 instead of 9, but the run payout against a pass D might be lower). I don't have my number crunchers in front of me, but is the upshot here that a team has a rough initial equilibrium but would move little after that, without a wholesale change? I hope I'm not misreading it.

In any event, I just wanted to mention a point I've been thinking about for some time, highlighted by Alamar's and Rockerbie's papers. I think strict average yardage maximization breaks down in some football situations.

For example, most 3rd downs. This might be simplistic, but I would chart (almost) all 3rd downs as binary: you either got it or you didn't. Now obviously sometimes on 3rd and 1 from midfield the offense breaks it for a long TD run or even runs a play action pass going for it, knowing that their percentage of success isn't great but that if they do it is a TD, but it will skew your statistics to factor in all the 3rd and short and 2nd and 2 runs into the same yardage totals problems. (No one thinks taking a knee should skew you back to the passing idea.)

One of the difficulties is you need to practically be on an NFL staff to gather this kind of info, but it's interesting because as significant as this work is (and it is), it's worth keeping in mind these other scenarios. That said, the vast majority of football plays take place on 1st and 10, and the analysis obviously holds true then. (Or should!)

I just want to clarify that I'm not claiming that ...

2008-06-27T09:19:00.000-04:00

I just want to clarify that I'm not claiming that the actual optimum ratio is 63% run/37% pass. It's just an example for demonstrating how game theory would solve the issue using notional numbers.

The actual solution would be more complex, and would there would be different solutions to different down and distance situations. The optimum ratio on 1st and 10 would obviously be different than on 3rd and 8.

"With rare exceptions, teams stick very closely to...

2008-06-23T16:40:00.000-04:00

"With rare exceptions, teams stick very closely to the NFL average 45% run proportion regardless of how their running game compares to their passing game." That's great info, and I think you're right on the money as to why that is (whether or not coaches are consciously aware of it). If only this was the type of analysis we see on Fox NFL Sunday.