Writers of Pro Football Prospectus 2008

Most Recent FO Features

BrownMal15.jpg

» Futures: Texas RB Malcolm Brown

DeMarco Murray is the toast of the NFL, but injury and team issues clouded some observers' view of his talent. Texas RB Malcolm Brown might have the same problem this winter. 

14 Jul 2003

Stacey Mack Daddy

by Aaron Schatz

If you learned that a running back with fewer than 100 carries ranked as one of the NFL's top offensive players of 2002, you would probably want to know who he is.  If you learned that this same running back was an unrestricted free agent, available to any team that wanted him this offseason, you might wonder why your favorite team's coach didn't know who he is.  Unless your favorite coach is Dom Capers.

That's because this surprise rushing star is Stacey Mack, four-year veteran of the Jacksonville Jaguars and new running back for the Houston Texans.  Long known as the caddy who filled in during Fred Taylor's inevitable injuries, Mack contributed more to his team's offense in 2002 than most of the league's better-known backs including Deuce McAllister, LaDanian Tomlinson, and maybe even Ricky Williams.

The explanation lies with a new statistical system called VOA, or Value Over Average, that breaks down the NFL season play by play to see how much success offensive players achieved compared to the league average in each specific situation. 

The expectation for a back's success is different on third and one than it is on first and ten.  A back will average more yards when the play begins at his own 20 than when the play begins at the opponent's 5.  If his team has a big lead, a good running back might get fewer yards per carry because he's running out the clock instead of trying to score, while another good back on a bad team might have fewer yards in a game because his team gave up on him in the third quarter, down by three touchdowns.

Worst of all, one back might have a schedule that includes two meetings with Tampa Bay and two with Carolina, both top run defenses, while another back gets two meetings with Seattle and two with Arizona, neither of which will be compared to the 1985 Bears any time soon.  All of these issues create biases in a player's statistics.  To complicate things, football has two objectives that get you closer to scoring: yards, and first downs, and they need to be balanced to determine a player's value.  All the yards in the world aren't useful if they all come in eight-yard chunks on third-and-tens.

VOA attempts to sort these biases aside to find a player's true value for the season.  The system is based on the work done by Pete Palmer and Bob Carroll in their seminal book, The Hidden Game of Football.  The book features a system to determine the success of individual offensive plays that considers context rather than just counting yards.  On first down, a play is considered a success if it gains 40% of needed yards; on second down, a play needs to gain 60% of needed yards; on third or fourth down, only gaining a new first down is considered success.

Palmer and Carroll then expanded upon this with a numerical system: 1 point for successful plays and 0 for unsuccessful plays, with extra points for gaining 10, 20 or 40 yards in one play, as well as --1 for losing more than two yards and --4 for a turnover.

VOA takes a database of every single play run in the NFL in 2002 and gives each play a success number based on the Palmer/Carroll system, though it has been somewhat modified to give fractional points between the different "success markers."  Then, the success of each play is compared to the average success of plays in similar situations for the entire season, adjusted for a number of variables.  These include down and distance, field location, time remaining in game, and current scoring lead or deficit.  VOA also changes the rating for turnovers to -8, which is more accurate based on tests so far.

Once we have all our adjustments, we can add the differences between this player's success and the expected success of an average running back in the same situation to get V+, a number that represents that back's number of successful plays over an average back.  We divide V+ by the number of plays to get VOA, or Value Over Average.  V+ is to VOA what hits are to batting average.

Of course, the biggest variable in football is the fact that each team plays a different schedule.  By adjusting each play based on the defense's average success in stopping running backs in the same situation over the course of 2002, we get DVOA, or Defense-adjusted Value Over Average.  This number also takes into account what direction a play went -- for example, the 2002 champion Tampa Bay Buccaneers were much better at stopping running plays to the left or up the middle than they were to the right.

This method, like the others used by FootballOutsiders, is explained here.  The main problem with VOA, so far, is that we still don't have a way to separate offensive line play from running back play.  We attempt to do that with a different statistic, Directional Rushing.

OK, now that we've gotten the geeky explanations out of the way, let's get back to our good friend Stacey Mack.  The list of the most successful running backs of 2002 is topped by a very familiar face, but you'll find Mack and two other unexpected names in the top ten.

Here are the top ten running backs in the NFL in 2002, based on the number of successful plays over average.  They are ranked with DV+, which is V+ adjusted for the defenses faced, but I'll include V+ so you can see how much each back is gifted or penalized by his team's schedule:

Carries
Yards
DV+
V+
Priest Holmes
313
1618
65.7
86.1
Clinton Portis
273
1508
46.9
73.0
Charlie Garner
182
962
46.0
51.6
Moe Williams
85
414
34.8
33.8
Kenny Watson
116
534
26.8
28.7
Stacey Mack
97
436
23.8
29.9
Garrison Hearst
215
972
23.5
30.7
Corey Dillon
313
1310
21.1
13.9
Michael Bennett
254
1297
19.7
21.1
Ahman Green
286
1241
19.4
  7.6

It's no surprise that Priest Holmes tops the list; despite missing two games at the end of the season, he was by far the best running back in the league this year even without taking his receiving abilities into account.  DV+ also shows that Charlie Garner is a much better runner than given credit for, his statistics limited by the pass-heavy Oakland offense, and that Corey Dillon had a very good season diminished by a difficult schedule and very little opportunity to run the ball late in games because his team was almost always far behind.

More importantly, for the purpose of evaluating the 2003 NFL free agent class, DV+ ranks Stacey Mack among the elite running backs of 2002 -- and, shockingly, it ranks fellow free agent Moe Williams even higher.  And that's with limited usage; if we used the rate stat, DVOA, instead of the total stat, DV+, they would be even higher on the list.

Stacey Mack?  Moe Williams?  Yes, these two backs delivered more yards and more first downs to their teams in 2002 than almost every other back in the NFL when compared to how an average back would do given the same opportunities.

Deuce McAllister rushed for 1381 yards to Mack's 436 yards.  Given the same opportunities, however, facing the same defenses as Mack did, in the same down-and-distance situations at the same points in the game, McAllister's 2002 record says he would have far less success in gaining yards and moving the chains.  Over 326 rushing attempts, McAllister's DVOA was 0.9%, meaning that he was less than one percent better than an average back in 2002.  Mack, on the other hand, had a DVOA of 23.4%.  Moe Williams -- shocking, but true -- was 43.6% better than average, tops among the 50 backs with the most rushes in 2002.  He was worth less than Holmes or Portis, however, because he was used far less frequently.

Stacey Mack ran 97 times in 2002, but on 38 of those carries had a first down or touchdown.  61% of his carries counted as "successful" under the terms put forward by Palmer and Carroll in Hidden Game, while the NFL average was 43%.  Williams had success on 58% of his plays, though his DVOA is higher than Mack's because he faced tougher defenses and had more plays that counted as "extra successful" by gaining over 80% of needed yardage.

The common dismissal of Mack and Williams is that they are limited backs, specialists who are only useful in a limited number of situations.  To be honest, nothing in this analysis contradicts the conventional wisdom that Williams cannot be an every-down back.  Williams certainly ran on third down far more often than an average back, but Mack's usage was only a bit different from the NFL's elite:

Player 1st Down 2nd Down 3rd/4th Down
Holmes 58% 33% 9%
Garner 59% 37% 4%
Portis 56% 34% 10%
Williams 43% 26% 31%
Mack 42% 35% 23%
NFL Average 54% 35% 11%

Mack had 13 runs of double-digit yards in 2002, which certainly contradicts the idea that he's a short-yardage specialist.  Even Moe Williams, despite being used in many third-and-short situations, had 12 double-digit runs in only 84 carries.

What about that third unexpected name among the NFL's top ten, Kenny Watson?  Clearly, DVOA believes that next year's battle to be the starting Redskins running back isn't really a battle as all.  Watson was significantly better than an average running back in 2002, while Stephen Davis and Ladell Betts come out around average.  Trung Canidate, acquired from St. Louis in trade, comes out as astonishingly poor, albeit in a very limited number of opportunities.

Watson actually gets penalized by DVOA because he played the easiest defensive schedule of any of the three Washington running backs.  Defensive adjustments are based on individual plays rather than individual games or a team's entire schedule, so two running backs from the same team may have wildly different adjustments for defense.  Stephen Davis played a very difficult set of defenses, while Kenny Watson played a more average set (hello, Seahawks).

Compare Stacey Mack to another big-name free agent of 2003.  Emmitt Smith had more than twice as many yards as Mack, rushing for 974 yards in 254 carries.  But for all those yards, Smith had -19.3 DV+, and -8.4% DVOA.  It's still early in the usage of these stats to make definitive judgments, but I don't think it is going out on a limb to say that Mack was a much better signing than Smith.  DVOA can't tell you how durable a player is, or whether he can carry a full-time load over 16 games, but at 6-1, 238, Stacey Mack doesn't seem like a man who will break down easily.

The wrench in the machine came when Mack went and signed with a second-year expansion team with a below-average offensive line.  My secret fantasy football strategy -- riding Stacey Mack to the title with whatever team he joined -- got tossed when he headed to Houston.  Mack's performance in 2003 will be a very important test to determine how much of a running game is based on the skill of the back and how much is based on the skill of the line.  Nonetheless, I think David Carr is going to enjoy a much-improved running attack, even if the offensive line's problems drag down Mack's 2002 performance.  As for Williams, he's sticking around Minnesota another year, and Watson will try to win the Washington starting job.  Which he should, easily, but there's no explaining Steve Spurrier when it comes to moving the ball on the ground.

Posted by: Aaron Schatz on 14 Jul 2003

comments