Writers of Pro Football Prospectus 2008

Most Recent FO Features

HarvinPer09.jpg

» Impact of the NFL's Kickoff Rule Change

After three NFL seasons of kicking off from the 35-yard line, what has been the impact on touchbacks, returns, field position, scoring and injuries? Also, is this rule responsible for a record number of big comebacks?

13 Apr 2011

Varsity Numbers: Adj. POE Revisited

by Bill Connelly

The NFL Draft is right around the corner, which means it's probably time to revisit last year's POE and Draftability column. Today, we'll look at the Adjusted POE and Highlight Yardage rankings for 2010. Later this week, we'll look at the draftability portion of the equation.

First, here's a quick refresher.

POE stands for "Points Over Expected." The idea for POE is simple: It compares a runner's production (in terms of EqPts) to the production that would have been expected of an average back given the same carries against the same opponents. A runner with, say, a plus-6.0 Adj. POE produced the equivalent of a touchdown more value than the average FBS running back would have with the same carries.

POE = EqPts - Expected EqPts.

Last year, we added an adjustment to account for the quality of a runner's offensive line (based on the team's Adj. Line Yards ratings). So below, you will see a raw POE figure and the more comprehensive Adj. POE figure. Backs are ranked by Adj. POE.

2010 Adjusted POE

For the second consecutive year, mid-major backs dominated the Adj. POE rankings. Last year's Top 5 consisted of North Texas' Lance Dunbar, Northern Illinois' Chad Spann, Memphis' Curtis Steele, Boise State's Doug Martin and Oregon State's Jacquizz Rodgers. This year, there was an improvement, at least -- two major conference backs cracked the Top 5.

Top 40 Collegiate Running Backs According to Adj. POE, 2010
Player School Year POE Rk Adj.
POE
Rk Rush.
Yds. Rk
Alex Green Hawaii Senior +40.9 3 +55.0 1 27
Ronnie Hillman San Diego State Freshman +15.4 39 +26.9 2 11
LaMichael James Oregon Sophomore +30.6 6 +26.4 3 2
Brandon Bolden Ole Miss Junior +21.6 17 +24.5 6 54
Alvester Alexander Wyoming Sophomore +4.1 197 +22.6 7 92
Andre Ellington Clemson Sophomore +19.4 20 +21.1 8 115
Damaris Johnson Tulsa Junior +23.7 12 +21.0 9 152
Thomas Merriweather Miami (Ohio) Senior +7.8 117 +20.8 10 62
Daniel Thomas Kansas State Senior +22.0 13 +19.5 11 8
Tevin Drake Western Michigan Freshman +14.7 40 +18.3 12 222
Player School Year POE Rk Adj.
POE
Rk Rush.
Yds. Rk
Chad Spann Northern Illinois Senior +25.6 11 +18.1 13 15
Cyrus Gray Texas A&M Junior +20.6 18 +17.8 14 36
Knile Davis Arkansas Sophomore +30.4 7 +16.5 16 16
James White Wisconsin Freshman +31.5 4 +16.1 18 44
Phillip Tanner Middle Tennessee Senior +9.4 88 +16.1 19 57
Jeremy Eddington Rice Freshman +10.7 76 +15.6 20 249
Latavius Murray Central Florida Junior +15.9 35 +15.3 21 126
Robbie Rouse Fresno State Sophomore +5.6 161 +15.3 22 37
Randall Cobb Kentucky Junior +18.3 24 +15.1 23 211
Vick Ballard Mississippi State Junior +21.7 16 +15.0 24 53
Player School Year POE Rk Adj.
POE
Rk Rush.
Yds. Rk
Edwin Baker Michigan State Sophomore +16.9 29 +14.1 27 26
Onterio McCalebb Auburn Sophomore +28.1 10 +13.7 28 89
Josh Harris Wake Forest Freshman +12.6 59 +13.6 29 103
Montee Ball Wiscosin Sophomore +28.9 9 +13.4 30 52
Orwin Smith Georgia Tech Sophomore +16.6 31 +13.1 31 170
T.Y. Hilton Florida International Junior +13.6 51 +13.0 33 315
William Powell Kansas State Senior +13.4 53 +12.9 34 350
Doug Martin Boise State Junior +17.2 27 +12.3 36 20
Jerrel Jernigan (WR) Troy Senior +10.3 81 +11.9 39 269
Tauren Poole Tennessee Junior +7.2 126 +11.9 40 45
Player School Year POE Rk Adj.
POE
Rk Rush.
Yds. Rk
DuJuan Harris Troy Senior +9.5 87 +11.9 41 140
Le'Veon Bell Michigan State Freshman +12.1 66 +11.6 42 139
Bilal Powell Louisville Senior +16.8 30 +11.6 43 14
Kerwynn Williams Utah State Sophomore +9.2 90 +11.6 44 199
Marcus Lattimore South Carolina Freshman +21.8 15 +11.4 46 28
Jordan Lynch Northern Illinois Freshman +12.9 56 +11.3 47 251
Roy Helu, Jr. Nebraska Senior +16.2 34 +11.2 48 21
Chris Thompson Florida State Sophomore +17.3 26 +11.0 49 82
Jeremy Avery Boise State Senior +13.5 52 +10.9 50 177
Kendrick Hardy Southern Miss Freshman +8.7 101 +10.5 52 67

Judging by Adj. POE, Alex Green (who you potentially had not heard of before this column) registered the best overall rushing season of anybody in the last six seasons. His output more than doubled that of the second-place finisher, San Diego State's stud freshman, Ronnie Hillman.

Top Five Seasons According to Adj. POE (2005-10)
1. Alex Green, Hawaii, 2010 (55.0)
2. Jahvid Best, California, 2008 (51.3)
3. Colin Kaepernick, Nevada, 2008 (48.6)
4. Yonus Davis, San Jose State, 2008 (45.8)
5. Pat White, West Virginia, 2006 (41.2)

One thing becomes very clear in looking at Adj. POE through the years: There is significant variability from year to year. Just because you were great in terms of Adj. POE one season, you are not guaranteed the same the next year, and vice versa. Last year's champ Lance Dunbar, for instance, went from plus-34.9 in 2009 to plus-6.6 in 2010. West Virginia's Noel Devine went from plus-0.8 in 2008, to plus-20.0 in 2009 ... to an incredible minus-11.7 in 2010. How this ties to the other topic at hand -- draftability -- is unclear.

Other Backs of Note

  • Mikel Leshoure (Illinois): 1,700 yards (third in the country), 0.5 Adj. POE
  • Jordan Todman (Connecticut): 1,695 (fourth), 4.1 Adj. POE
  • Bobby Rainey (Western Kentucky): 1,648 (fifth), minus-6.5 Adj. POE
  • Vai Taua (Nevada): 1,610 yards (seventh), minus-6.1 Adj. POE
  • Lance Dunbar (North Texas): 1,561 yards (ninth), 6.6 Adj. POE
  • Kendall Hunter (Oklahoma State): 1,548 yards (10th), 6.5 Adj. POE
  • Mark Ingram (Alabama): 875 yards (74th), 8.3 Adj. POE

As we'll see below, Mikel Leshoure generated quite a bit from a Highlight Yards perspective, and he was one of my favorite backs of 2010, but credit for a lot of his raw POE of plus-12.3 was assigned to his offensive line. Post-adjustment, his Adj. POE was right at the national average. Vai Taua was the recipient of even stronger offensive line support, rating a touchdown lower than average post-adjustment.

Former Heisman Trophy winner Mark Ingram, meanwhile, saw his Adj. POE numbers fall just like his overall rushing yardage. In 2009, he gained 1,452 yards and posted an Adj. POE of 18.2. In 2010, those numbers fell to 875 yards and 8.3 Adj. POE.

Now let's move to quarterbacks.

Top 20 Collegiate Quarterbacks According to Adj. POE, 2010
Player School Year POE Overall
Rk
Adj.
POE
Overall
Rk
Colin Kaepernick Nevada Senior +46.7 1 +25.8 4
Taylor Martinez Nebraska Freshman +29.4 8 +24.6 5
Austyn Carta-Samuels Wyoming Sophomore +7.2 124 +17.6 15
Cameron Newton Auburn Junior* +46.4 2 +16.5 17
Russell Wilson N.C. State Junior +19.3 21 +14.2 25
Andrew Luck Stanford Sophomore +16.9 29 +14.1 26
Brandon Connette Duke Freshman +9.8 85 +13.0 32
Bryant Moniz Hawaii Junior +8.9 97 +12.5 35
Ryan Aplin Arkansas State Sophomore +10.3 82 +12.1 37
Denard Robinson Michigan Sophomore +31.1 5 +12.1 38
Player School Year POE Overall
Rk
Adj.
POE
Overall
Rk
Collin Klein Kansas State Sophomore +12.2 63 +11.4 45
Terrelle Pryor Ohio State Junior +18.5 23 +10.6 51
Alex Carder Western Michigan Sophomore +6.3 142 +10.4 54
Trey Burton Florida Freshman +12.0 68 +10.2 55
Kriss Proctor Navy Junior +12.1 66 +10.0 56
Austin Danton Toledo Sophomore +9.4 90 +9.6 62
Chandler Harnish Northern Illinois Junior +13.2 56 +9.4 64
Tim Jefferson, Jr. Air Force Junior +16.6 32 +9.4 67
Blaine Gautier UL-Lafayette Sophomore +3.7 207 +9.0 70
Tyrod Taylor Virginia Tech Senior +14.2 45 +8.9 72
* Declared early for the NFL draft.

How bad was Wyoming's offensive line in 2010? Bad enough to convince the numbers that quarterback Austyn Carta-Samuels and running back Alvester Alexander could have been better than Cameron Newton and Andre Ellington, respectively, given actual blocking.

Carta-Samuels aside, the rest of the list takes shape about how you'd have expected. The numbers think Newton got significant help from a fantastic offensive line, but having Kaepernick, Martinez, Newton, and Wilson in the top five seems about right.

Now, a moment of silence for Navy's Ricky Dobbs, a sentimental Heisman candidate for 2010 who just couldn't quite rekindle the 2009 magic. He rushed for 1,033 yards, but it took him some inefficient carries to do it. His final Adj. POE for 2010: minus-18.4. His backup, Kriss Proctor, produced an Adj. POE 28.4 points higher in 232 fewer carries.

Highlight Yards

Now let's expand on another topic we unveiled in last year's POE column: Highlight Yards.

A few months ago, FO introduced the idea of "Second Level Yards" and "Open Field Yards" for NFL running backs. These were the remaining yards that were left after each break in the baselines for Adjusted Line Yards. I've done the same thing here for college backs, with two differences. First, we're adding together both "Second Level" Yards (5-10 past the line) and "Open Field" Yards (11-plus past the line). Second, we're counting only half the Second Level Yards, just as the line gets half credit for these yards. We'll call this stat "Highlight Yards," because these longer runs are the ones that show up on the highlight shows. A three-yard run gets zero Highlight Yards. A 70-yard run gets 63 Highlight Yards. The more Highlight Yards, the more explosive the runner was, and the less his overall yardage and POE totals were due to the offensive line blocking for him.

Basically, Highlight Yards are the yards credited to the running back and not the blocking. Again, we rank them below in terms of their full-season accumulation instead of their per-carry average.

Top 20 Rushers According to 2010 Highlight Yards
Hlt. Yds.
Rk
Name School Rushes Yards Hlt. Yds. Hlt. Yds./
Carry
Per Carry
Rk
1 LaMichael James Oregon 294 1,730 829.8 2.82 45
2 Mikel Leshoure Illinois 282 1,700 789.3 2.80 47
3 Denard Robinson Michigan 249 1,741 779.1 3.13 24
4 Lance Dunbar North Texas 274 1,561 755.2 2.76 48
5 Ronnie Hillman San Diego State 262 1,532 695.1 2.65 58
6 Cameron Newton Auburn 241 1,625 692.9 2.88 40
7 Alex Green Hawaii 146 1,199 692.7 4.74 5
8 Taylor Martinez Nebraska 136 1,143 691.4 5.08 3
9 Bilal Powell Louisville 229 1,405 688.6 3.01 29
10 Jordan Todman Connecticut 334 1,695 654.6 1.96 144
Hlt. Yds.
Rk
Name School Rushes Yards Hlt. Yds. Hlt. Yds./
Carry
Per Carry
Rk
11 Colin Kaepernick Nevada 165 1,265 629.3 3.81 8
12 Daniel Thomas Kansas State 297 1,583 611.2 2.06 130
13 Roy Helu, Jr. Nebraska 188 1,245 600.4 3.19 22
14 Bobby Rainey Western Kentucky 340 1,648 599.4 1.76 181
15 Knile Davis Arkansas 204 1,322 599.0 2.94 35
16 Chad Spann Northern Illinois 258 1,388 593.2 2.30 97
17 Chris Polk Washington 260 1,413 587.5 2.26 101
18 Jay Finley Baylor 195 1,218 582.4 2.99 31
19 Vai Taua Nevada 284 1,610 573.1 2.02 136
20 Zach Line SMU 244 1,494 569.8 2.34 90

LaMichael James clearly takes the title of most accomplished major conference running back for 2010. His per-carry average was not quite in the top echelon, but his combination of durability and explosiveness were very impressive. And North Texas' Dunbar still showed upper-level quality as well.

For more full POE and Highlight Yardage rankings, stay tuned to my side blog, Football Study Hall, later today. By the way, thanks to the work of Marty Couvillan at cfbstats.com, we now have a boatload of pass targeting data through which to sort. Once the legwork is done on that, we should have a POE figure for receivers (and a pass-catching portion for running backs) as well. Good news.

Posted by: Bill Connelly on 13 Apr 2011

20 comments, Last at 18 Aug 2011, 1:52pm by vincemullins

Comments

1
by BlueStarDude :: Wed, 04/13/2011 - 12:46pm

Todman finishing 10th in highlight yards is pretty impressive given that the Huskies had nothing else that even resembled an offensive threat so every defense he faced was keyed on stopping him. Then again, those defenses weren't exactly stellar were they?

Alex Green had a good week of practice before the Shrine game, real good hands for a guy his size. I wouldn't mind seeing Jerry Jones use a late round pick on him to replace Barber in the RB rotation.

2
by Bill Connelly :: Wed, 04/13/2011 - 1:01pm

Interesting question via Twitter: "You say high variability year 2 year. Shouldnt that indicate flaw in statistic? In other words, if an individual stat has high variability, maybe what you're measuring is not under control of individual"

Thoughts on this? Since it deals with running backs, and since it seems RBs are some of the most high-variability/high-interchangeability players in football, I've been willing to accept wild swings from year to year. Is that being too dismissive?

5
by DSMok1 (not verified) :: Wed, 04/13/2011 - 3:47pm

Yes, you should consider it a significant issue to be investigated further. There are ways to try to deduce the sources of fluctuations like this statistically, but they will be tough to implement in such a scenario.

In general, if there is 0 year-to-year correlation, then there is likely either
a) No skill to it (it's random)
b) Other influences that are larger in magnitude are drowning the skill out
c) The skill is changing a ton (and not just getting better)

I would guess b) is correct here. So many things can change; there are so many variables that lead to the POE being good or bad.

If your year-to-year correlation is very low, it's not likely reflecting the true talent level of the tailback very closely.

9
by Aaron Brooks' Good Twin (not verified) :: Wed, 04/13/2011 - 6:41pm

Well, the elephant in the room is option 0: the stat is crap.

6
by Kibbles :: Wed, 04/13/2011 - 4:53pm

I'm not a baseball guy, so some Sabermetrician will undoubtedly be along shortly to correct me on this, but I believe Bill James said that a statistic was useful if it produced a list that was 9 parts exactly what you expected and 1 part complete surprise. The idea was that even the horribly flawed conventional wisdom should still be able to recognize true greatness (or true awfulness) more often than not. If he created a stat and generated a top-10 list, and 9 of the 10 entrants were head-scratchers, he assumed that there was something wrong with his stat, not that every baseball observer ever was an idiot.

How well do the lists you're generating adhere to that goal? To be honest, I don't watch enough college football to know for sure (I love college football, but unless they're playing for or against my Gators, I simply don't see them play). Perhaps someone else could weigh in on the subject.

7
by Bill Connelly :: Wed, 04/13/2011 - 5:31pm

Conventional wisdom is a little more difficult with college football because there are so many games going on that you just don't see. And success at the RB position is so seemingly random (in terms of which schools produce the great backs) that I somewhat expect to be surprised by the POE list. When I do an overall POE list (rushing + passing) for QBs, it tends to surprise me much less, but I'm convinced enough that RB is a crapshoot that I don't mind the list looking like it does. It's not trying to predict anything, and really, it's one of the most straight-forward measures I use. Your carries against these opponents should have produced X EqPts based on the quality of those opponents; you produced Y EqPts. Y - X = POE. The only intricate part is the offensive line adjustment.

10
by Kal :: Wed, 04/13/2011 - 8:16pm

I think that's being too dismissive. Anecdotally we know that while good RBs may have some meh seasons here and there, they'll usually be pretty strong overall. While you should expect some variability if there's really no correlation between seasons chances are that there's something wrong with the metric - or more likely, the metric is just not telling you that much.

One way to measure the interchangeability of it is to combine all rushes from all running backs for a given team and see, year to year, how variable that is. Your hypothesis would then be that the correlation between overall running back play year to year is strong, but individual play is not. I'd expect there to be a similar lack of correlation thus demonstrating the metric needs work.

12
by Bill Connelly :: Wed, 04/13/2011 - 9:55pm

I think it typically views good RBs as good no matter what ... but the extremes at the top change from year to year. (Noel Devine is an outlier here, but ... well, Noel Devine was a shell of his former self this year thanks to injuries, so that doesn't surprise me.)

3
by Jon L. (not verified) :: Wed, 04/13/2011 - 2:41pm

This is an interesting rating system. It's different to see players from non BCS schools up there like Alex Green from Hawaii.

4
by Aaron Brooks' Good Twin (not verified) :: Wed, 04/13/2011 - 3:02pm

Are you controlling for down and distance?

Alex Green's numbers come in an offense that almost always passes the ball. (225 non-QB rushes versus 618 pass attempts and 83 QB rushes.) Are we seeing a very efficient rusher, or just a guy who gets a lot of change of pace run plays against dime defenses?

He also generated 635 yards on 49 carries against Utah State, UNLV, and New Mexico State, who were, to put it charitably, god-awful. I see APOE adjusts for your line, but does it adjust for the defense's line?

8
by Joseph :: Wed, 04/13/2011 - 5:46pm

Bill, I think ABGT has a good point. Since CFB schedules are extremely different, this may have some effect. I'd also say that after one good year, DC's are likely to game-plan against that good RB--esp. in the non-BCS schools that generally don't have a lot of star skill-position players (e.g., like the North Texas RB). I would say that if Alex Green's #s were Defense-Adjusted, they would be much lower. Since BCS schools tend to play 2 to 3 cupcakes per year, being able to take advantage of these games may explain at least SOME of the variability. Also, since in the SEC, ACC, Big Ten, and Big 12, they don't play every conference opponent yearly, that could explain some more variance.
I think, like FO did for some other recent stat (for WR's??) in an ESPN column, you might want to have a conference adjustment. Something tells me that Alex Green might not have had those #s playing at Alabama instead of Mark Ingram. [Note--I haven't seen him play--he might be super fast, and be able to run away from some DB's in the 2nd level that catch Ingram. I just think that Ingram faced a LOT harder schedule of defenses. For anecdotal proof, I offer up the Sugar bowl of a couple of years ago when unbeaten Hawaii got slaughtered by the Georgia Bulldogs, mostly because their offensive line got whipped.]

Other variance explanations: injury (or lack thereof), O-line changes [I know this stat tries to get rid of them--but a better lineman may sustain his block a half-second longer, allowing the RB to get 10 yds vs. 5--or it might be 20 yds instead of 5. Without watching tape of everyone, it's hard to tell.]; overall team quality, reflected in the scoreboard (in other words, you won't be racking up rushing yds, attempts, and highlight yards if you're behind); usage (2nd stringer becomes the main RB); OC/head coach (Mike Leach & June Jones don't give their RB's 20+ carries per game).

Sorry--defense adjustments could prob. make the stat better and more predictive, but some variance will be there, because CFB is highly variable. But any stat that has LaMichael James near the top is probably doing OK.

11
by Bill Connelly :: Wed, 04/13/2011 - 9:52pm

It does not adjust for the defense's line -- it adjusts for the defense, period. That's where the "expected" part comes from. And I think the 49-for-635 thing still shows great things about Green. Those defenses were awful, and he torched them ... but he still torched the more than anybody else did.

14
by Aaron Brooks' Good Twin (not verified) :: Thu, 04/14/2011 - 10:13am

Sure, but his usage rate also peaked against his lesser competition. The only good team he did anything against was Boise, and Boise won 42-7. I suspect that was a game where Boise played nothing but DBs for most of the 2nd half, and was content to let Green run for 8 yards a pop, knowing Hawaii needed to score more quickly than that. Green's numbers are weird -- most game in blowouts, one way or the other.

20
by vincemullins :: Thu, 08/18/2011 - 1:52pm

watched that game, and Boise ABSOLUTELY went into dime sets and let Green get chunks of yardage.

Our work has shown that heavy backs tend to outperform smaller ones in the run-and-shoot offenses - so when Ilaoa or Green types pop up getting carries at Hawaii, or maybe Keola Antolin at Arizona this season, we should take notice.

Vince Mullins, Fantasy College Football Analyst
FantasyCollegeBlitz.com

13
by Samson151 :: Thu, 04/14/2011 - 9:58am

I gotta admit, these two metrics leave me cold. Perhaps because they bear so little relationship to something a coaching staff might use to evaluate or (hopefully) improve a team's performance.

15
by Bill Connelly :: Thu, 04/14/2011 - 10:24am

Everything else we do is team-related -- this is really the first individual measure we've pursued to any great degree.

16
by dryheat :: Thu, 04/14/2011 - 11:53am

I really don't understand attributing Open Field yards entirely to the running back. I mean, if the offensive line annihilates 8 players 0-7 yards from the line of scrimmage and the RB goes for 25 yards untouched before a cornerback can drag him down, shouldn't the O-line get most of the credit for all of those yards? Or is this meant to be a glorified speed/agility score?

17
by Joseph :: Thu, 04/14/2011 - 5:10pm

Because if 5 O-linemen are annihilating the front 7 AND the SS, then probably a BCS school is practicing against your high school's JV team. ;)

Really, I think it boils down to the fact that you will almost never see a lineman on a running play more than 5 yds downfield. So, after that, yds gained is almost a 100% result of the speed, shiftiness, ability to break tackles, stiff-arm, etc. of the RB. (Obviously, MINUS the blocking ability of any WR blocking downfield on a DB.)

18
by Aaron Brooks' Good Twin (not verified) :: Fri, 04/15/2011 - 10:23am

This also assumes Deion Sanders isn't the DB.

NFL.com had a fun glitch where one year they credited Deion with 7 INTs and 0 tackles. It took me a couple of minutes to realize that probably wasn't true.

19
by bbeckgtw (not verified) :: Sun, 04/17/2011 - 1:36pm

Bill,
If you're going to "add an adjustment to account for the quality of a runner's offensive line (based on the team's Adj. Line Yards ratings)", you should equally measure the quality of the defenses the offensive line and runner are competing against; especially if you are trying to establish a correlation to potential success at the next level. Another variable that would possibly help explain the results you've attained of "mid-major backs" dominating the Adj. POE rankings, would be including the style of offense the back is playing in, and specifically, the formation each run was executed from. It seems that the spread offenses are better suited to providing a runner with an opportunity to accumulate Highlight Yardage than a standard pro set or wishbone variation. Lastly, how are you factoring in field position to the overall results? You list "Second Level" Yards (5-10 past the line) and "Open Field" Yards (11-plus past the line) in the measurement; how are you factoring runs that fall inside those yardages?