It's Not A Mystery

[NorthCoast]

....if all you know about a running play in the NFL is the approximate field position of a team and the number of defenders near the line of scrimmage, you’re able to predict the leaguewide yards per carry with an extraordinarily high degree of accuracy: 96 percent of yards-per-carry totals are explained by the offense’s field position and the number of men the opponent has in the box. How many defenders are in the box is almost certainly the most important factor in determining rushing success in football, so it follows that we should try to account for it.

Notice that in some cases more than a yd/att difference between 6 and 8 defenders in the box. Spread the defense out and make them defend beyond a 5 yd box.

[Northern_Blitz]

Notice that in some cases more than a yd/att difference between 6 and 8 defenders in the box. Spread the defense out and make them defend beyond a 5 yd box.

It seems like this is the kind of thing that a team with a good analytics department would try to understand and exploit.

But I bet a lot of this is just doing conventional wisdom.

Or maybe it's about risk minimization. Teams don't like to throw in the shadow of their own end zone because a pick is pretty likely to be a pick-6. And that means you're pretty likely to lose.

I've always thought that a team that doesn't reflexively fire coaches should have a better chance of exploiting (or at least exploring) ideas like this because you'd think they'd be less risk averse.

It's also the reason I think it's stupid to put a coach "on the hot seat". If you're going to fire someone, just do it. Don't make them live with the sword of Damocles hanging over their head so they're too afraid to make calls that increase win rate but have a higher chance of failure.

[feltdizz]

It seems like this is the kind of thing that a team with a good analytics department would try to understand and exploit.

But I bet a lot of this is just doing conventional wisdom.

Or maybe it's about risk minimization. Teams don't like to throw in the shadow of their own end zone because a pick is pretty likely to be a pick-6. And that means you're pretty likely to lose.

I've always thought that a team that doesn't reflexively fire coaches should have a better chance of exploiting (or at least exploring) ideas like this because you'd think they'd be less risk averse.

It's also the reason I think it's stupid to put a coach "on the hot seat". If you're going to fire someone, just do it. Don't make them live with the sword of Damocles hanging over their head so they're too afraid to make calls that increase win rate but have a higher chance of failure.

I wonder what the percentage is for pass is run when inside your own 5 yard line.

Maybe I’m misremembering but it sure seems like teams pass on first down a lot in those situations.

[flippy]

So if you run every play with a 3-4 WR set, on average you should get inside your opponents 10 yard line.

With 3-4 WRs, teams can’t have 8 guys in the box if you spread things out.

So by averages, you should be able to convert 1st downs the majority of series. I’d even take it a step further and play the percentages of never punting.

[NorthCoast]

So if you run every play with a 3-4 WR set, on average you should get inside your opponents 10 yard line.

With 3-4 WRs, teams can’t have 8 guys in the box if you spread things out.


So by averages, you should be able to convert 1st downs the majority of series. I’d even take it a step further and play the percentages of never punting.

I get what you are saying flippy.
Here's the Steelers breakdown (in 20 yd increments)

Yds Away	yds/att rushing
90	5.5
70	4.0
50	3.3
30	5.4* (2.7)
10	0.8

*this number is skewed because two runs went for total 22 yds. the number in parentheses is minus those runs

[Captain Lemming]

I get what you are saying flippy.
Here's the Steelers breakdown (in 20 yd increments)

Yds Away	yds/att rushing
90	5.5
70	4.0
50	3.3
30	5.4* (2.7)
10	0.8

*this number is skewed because two runs went for total 22 yds. the number in parentheses is minus those runs

The "skewed" number recalls anothor factor here.
The further from the GL opens the possibility of longer runs period.

If a back breaks free on a run starting 90 yards away his 90 yard run contributes to the average. Same run 20 yards away have a capped potential of 20 yards max. Therefore you can have EXACTLY the same breakaway run success and the runs from 90 yards away will average more yards.

Think about this from 10 yards in. Many of those runs are FROM the one or two yard line making it IMPOSSIBLE to have a longer run, thus skewing the numbers.

Let me add that you the "in the box" stat is skewed as near EVERYBODY is in the box as you near the goal line. If you convert ONE HUNDRED percent of your one yard line attempts you will average ONE YARD. You WILL go against teams with 8 in the box on EVERY ONE of those attempts as you average one yard while being 100 percent successful.

I do think the stats have merit, but not to the degree it looks like.

[Northern_Blitz]

The "skewed" number recalls anothor factor here.
The further from the GL opens the possibility of longer runs period.

If a back breaks free on a run starting 90 yards away his 90 yard run contributes to the average. Same run 20 yards away have a capped potential of 20 yards max. Therefore you can have EXACTLY the same breakaway run success and the runs from 90 yards away will average more yards.

Think about this from 10 yards in. Many of those runs are FROM the one or two yard line making it IMPOSSIBLE to have a longer run, thus skewing the numbers.

Let me add that you the "in the box" stat is skewed as near EVERYBODY is in the box as you near the goal line. If you convert ONE HUNDRED percent of your one yard line attempts you will average ONE YARD. You WILL go against teams with 8 in the box on EVERY ONE of those attempts as you average one yard while being 100 percent successful.

I do think the stats have merit, but not to the degree it looks like.

Good point.

Would be nice if they had something like explosive play rate (replaced by TD rate when inside the 20) to try to show mow much the long tail of big runs impacts the average.