UM v. NU and YPPA

Submitted by CR on

Bud Goode (now deceased) might have been the first stats guy to look at data and the NFL. He dates to 1959. He believed (eventually) that “The single most meaningful independent statistic in pro football is Yards Per Pass Attempt Differential.” I have been looking at this for over a decade and believe that net YPPA provides the best sort of team quality. Many others disagree, and use a variety of data in their models. The thing is, when I look at these sorts at the end of a season, net YPPA still tends to evaluate teams “better” (holistic, to some extent, I agree) than any single or multiple algorithms. Well, believe or not, here is the current taxonomy, with data taken from Ed Feng’s The Power Rank. The NCAA (erroneously) lumps sacks in with rushing yards and Ed puts them in the right places. I note that Ed doesn’t agree with me on this (he is a PHD on point, I am just a second rate lawyer). Caveat.

1. OSU +2.83

2. UM +2.72

3. Northwestern +2.30

4. Iowa +1.83

5. Wiscy +1.69

6. MSU +1.43

7. PSU +1.22

8. Rutgers +0.50

9. Indy +0.48

10. Illini +0.40

11. Minny +0.33

12. Nebraska -0.50

13. Purdue -0.65

14. MD (20/20) -0.97

Other warnings. Limited data to this point in time. But, there are enough trials to begin to push relevance. A bigger issue is SOS. I have never done it but I suppose one could use Sagarin (or other measure) to adjust these numbers.

Over the years a multiplier of 5 tends to fit with turnover neutral outcomes and not adjusting for home fields. That is, a multiplier of “5” has tended to fit the measure to points.

So, let’s look at UM and NU. UM is plus .42. Adjusting (via Sagarin) for SOS (UM’s is 4% stronger) so call the delta .4386. Times 5 is 2.19 points. Home field? I might say about 4 in this one, unless we get the O’Neill crew (all bets off). I don’t know if anyone has studied college HF advantages, other than assigning an arbitrary “3”. So, in my mind, in a TO neutral game I like UM by 6.

Tuebor

October 9th, 2015 at 1:04 PM ^

If Sacks come off passing yards then shouldn't QB scrambles count towards passing yards since they are yards gained during a passing attempt?

 

Based on what we've seen this year those rankings look pretty good.  I'm hesitant to put MSU so low though, but their style doesn't get them alot of passing yards and their secondary is a major weakness so perhaps it is right.  I think it would be more interesting to look back at 2013 or 2014 and see how this matches up with the final standings.

 

And Sagarins current Home advantage is 2.71 points.  So it looks like Michigan is favored around 5 points which seems more or less right.

CR

October 9th, 2015 at 4:49 PM ^

A part of the problem is that from a box score you can't distinguish a read option from a scramble. And there wil be team-specific outcomes. Michael Vick's team won't look anyhting like John Navarre's team, once you add scrambles into the equation. Maybe someone can solve this part of the equation, but I have no answer.

I can say, year after year, net YPPA tends to match up with outcomes pretty seamlessly.

Tuebor

October 12th, 2015 at 1:04 PM ^

I think if you had scorers who were looking at plays from a high level they could distinguish a read option from a scramble and score the read option a rushing play and the scramble a passing play.  But just leaving sacks as runs is a pretty simple way to do it too.

 

The only reason I mention it is because having YPPA leave off scramble yardage nullifies the effect a guy like Russel Wilson can have.  Cover everything downfield and still give up 10 yards because you didn't account for the QB.

CR

October 9th, 2015 at 6:00 PM ^

Here it is. Net Yards Per Pass Attempt from 2011. No Bowl Games

                  net     wins

Wisconsin  3.83   10

MSU          2.91    10

Michigan    2.05    10

PSU           1.24      9

Nebraska   0.66      9

Iowa           0.34      7

Illinois         0.19      6

NU            -0.59      6

Purdue      -0.66      6

OSU          -0.73      6

Minny        -0.91      3

Indiana      -0.92      1

BornInAA

October 9th, 2015 at 1:29 PM ^

The loss says to me that I think there will be no undefeated teams this year.

All the SEC, Big Ten, ACC and PAC will lose at least 1.

Baylor, OK and TCU have to play each other so they will end up with 1 loss too.

The selection committee for the playoff will be interesting. One of the conferences will get screwed probably the ACC.

El Jeffe

October 9th, 2015 at 1:26 PM ^

I went back and watched the highlights of our three home games. Rudock was really pretty good in those. He was really pretty awful in our two away games.

I think UM wins this one by two touchdowns. Or since Craig likes numbers:

Non-sad ghost Rudock + UM defense + awful NU offense = +14.

Avon Barksdale

October 9th, 2015 at 1:38 PM ^

We play the same type of football: meh offense, solid defense, no mistakes on special teams, good coaching. Michigan playing at home is the key for me. The Big House should be rocking at 3:30!



Michigan 20 | Northwestern 7

LSAClassOf2000

October 9th, 2015 at 2:14 PM ^

Interesting stuff and an intriguing analysis, OP. 

Another one that I tend to look at which actually is reasonably predictive of overall success is third down conversion differentials, a battle which Michigan is currently winning by a wide margin. Over the course of the season, we're converting 42.1% compared to 19.4% for our opponents for an overall differential of +22.7%. Tracking that over the last few years, teams that can maintain that tend to go to nice bowls - I will put it like that. 

 

rockydude

October 9th, 2015 at 2:59 PM ^

I don't have numbers for you (because I'm lazy), but remember the bad old days? 1st down, bad guys try a run. No dice. 2nd down, they pass. Incomplete. 3rd down and long, they get aggressive and torch us for 20-25 yards. That sucked, but it sure isn't the case now. This defense knows how to finish.

CR

October 9th, 2015 at 4:45 PM ^

...but i don't know the answer right away. I will look at it. I can say, historically, a team at plus 3.5 is in the running to win all the marbles. A team at plus 2.5 is usually in the top 15.

But, let me look at it.

CR

October 9th, 2015 at 5:26 PM ^

shows UM to be about 14th nationally, with 8/9 teams (including OSU) bunched together. Baylor is off the charts but their schedule is off the charts weak, # 122 right now according to Sagarin. A primary problem with a broad comparison right now is that schedules are quite variant. BC is in the top 15 (they are 3-2) but they have played Howard (a 76-0 win) and Maine (also not FBS) so you can't really take their numbers seriosuly at this point.