Outside Rushing: a key to success?

A) TwoMinuteWarning has been around since 1997, pioneering much of the new wave of NFL statistics
B) There's a ton of data kept that isn't always shown on the main parts of the site

Today I am going to focus on what has become a very neglected idea of the original TMW data charting efforts: directional rushing. These numbers used to be included in the layout of the game pages (a one page summary sheet for each game):

Directional Rushing Stats

Minnesota -- Projected Stats

EFR	93	67	64	95	65
Big	19 %	9 %	6 %	13 %	12 %
PSR	48 %	37 %	38 %	52 %	43 %
Yds	4.7	3.8	3.7	4.3	4.0

Wide Left

Left

Middle

Right

Wide Right

The rushing plays were classified as one of five directions, tallied up and kept for offense and defense. Now this looks like some interesting and potentially valuable stats that maybe no other spread analyst or linemaker has or is inclined to use in predicting scores. Cha-ching! When you come across a novel piece of data then it's worth a look since finding some way to use it will give you a leg up in beating the spread.

As far as I know there was never an attempt publicized about how to use these stats successfully. So this sparked my curiosity and as you might guess I have found something I think is worth knowing!

To me rushing really breaks down two ways, not five: inside and outside. So I chose to look at things in those terms. I also feel like the basic yards per carry number is the easiest to work with and that became my mainstay method for testing. So in the end I got two ratings per matchup:

Inside rushing edge = [away inside yds + home defensive inside yards allowed] - [home inside yds + away defensive inside yards]
Outside rushing edge = [away outside yds + home defensive outside yards] - [home outside yds + away defensive outside yards]

With data on directional rushing from 1998 to 2007 I have TEN years of data sample. Arbitrarily I looked at weeks four through seventeen since you need a few games I figure to get started.

INSIDE edge predictor results
Nothing too promising. Overall a strictly 50% spread predictor and even using some filters (minimum edge difference, only dogs, etc) there was no area I could focus on with value.

OUTSIDE edge predictor results
While playing every single team with any edge whatsoever was 52.6%, the real news came when I used an arbitrary 3.5 minimum edge cutoff:

Longtime TMW visitors will be familiar with this table structure. It's the basic display from the standard testing module we use. And the news is good!

Overall: 54%
Favorites: 50%
Underdogs: 59%

That's a lively method right there!

Play any underdog with a 3.5 or higher "outside rushing yards edge" in week four and on of the regular season (59% long term performance, with 30-40 plays per season on average).

You might also notice from the odds splits:
7.5+ point underdogs: 40-16 (71%)
3.5 to 7 point underdogs: 73-53 (58%)
0 to 3 point underdogs: 80-64 (56%)

So it's even better with the underdogs getting over a touchdown on the spread.

Looking at it year by year shows:

Season 2007: 2006: 2005: 2004: 2003: 2002: 2001: 2000: 1999: 1998:

Underdogs w/3.5+ edge 8-13 26-17 19-10 16-8 24-16 21-17 18-6 17-15 24-12 20-19

Underdogs getting 3.5+ points 3-7 16-10 10-3 10-3 15-11 14-4 8-4 7-9 16-6 14-12

So very seldom has this method had a poor season. Unfortunately the worst season came in 2007. I am fully expecting a HUGE bounce back to winning ways in 2008! As a result there will be a new DIRECTIONAL RUSHING feature as part of the Investor Subscription for NFL 2008.

You can even take a look at some of the small home favorites under this method as well: they are 46-31 (60%) over the ten years (home favorites with 3.5+ outside rushing edge, laying three points or less).

Find out more:
What the press says about TwoMinuteWarning
Information on subscribing
Frequently Asked Questions