370 Carries Revisited
by Aaron Schatz
All season long, when talking about Shaun Alexander's problems, you've seen us refer to the "Curse of 370." You've also seen that term as we tracked Larry Johnson's high number of carries all season long. Now that Johnson has set the NFL record with 416 carries in the regular season, the "Curse of 370" is showing up in everything written by Football Outsiders.
We wanted to help those who were new to our site understand the Curse of 370, which is something we've been writing about for three years. The first article about 370 carries appeared in the book Pro Football Forecast 2004, and was also published on our site when Ricky Williams retired in July 2004. You'll find that article here. A sequel article appeared in the Seattle chapter of Pro Football Prospectus 2006, and it is republished in its entirety below.
Just so people understand, there's nothing magical about carry number 370 that makes a running back blow out his ACL, any more than there is something special about pitch 100 that makes a pitcher's arm fall off. It's simply a useful shorthand to represent the fact that overworking your running back with too many carries is a bad thing. The punishment gets worse and worse with more carries, and 370 is a close approximation of the tipping point.
Since we began doing football research a few years ago, we've developed a number of axioms that tend to come up over and over again, both on our Web site and in Pro Football Prospectus. With both Shaun Alexander and Edgerrin James entering free agency this off-season, one precept in particular got a lot of attention: the 370-carry theory.
The 370-carry theory is generally summarized as follows: "A running back with 370 or more carries during the regular season will usually suffer either a major injury or loss of effectiveness the following year, unless he is named Eric Dickerson."
Some recent examples of the 370-carry theory in action include Curtis Martin's collapse last year [i.e. 2005], Jamal Lewis's struggles for the last two years, and Ricky Williams falling from 4.8 yards per carry in 2002 to 3.5 yards per carry in 2003. Terrell Davis, Jamal Anderson, and James each tore his ACL a year after going over 370 carries. The 370-carry barrier helped destroy the careers of Earl Campbell, Barry Foster, and Gerald Riggs.
But when looking at the question of how much to pay Alexander and James this off-season, the 370-carry theory seemed a bit limited. Why should a running back like James, who is used so much in the passing game, have his workload be defined solely by carries without considering receptions? And why should our count end before the playoffs, when Alexander added another 20 carries per game through the Super Bowl?
Let's examine each question in turn. The first issue is whether "touches" are a better measure of workload than carries, with touches equaling the total of each back's carries and receptions for a given year. From 1978 through 2004, 60 running backs had seasons of at least 340 carries. Comparing the number of carries for each running back with the number of yards he gained the following year gives us a correlation coefficient of -.24. In other words, as players carry the ball more, they are less likely to run for as many yards the following year, due to a mixture of lost effectiveness and injury. If we want to measure only effectiveness (yards per carry) or playing time (total carries) the correlations are similar.
If we take the same 60 running backs and compare touches to yards the following year, the correlations are roughly half as large. This suggests that carries are a better indicator of workload than touches. Compare just receptions to rushing yards the following year, not even considering carries, and it is clear why: the correlation between receptions and yards the following year is actually positive, albeit tiny. If more receptions indicate anything, it is that a player will gain more yards the following year, in particular more yards per carry.
|Correlation between running back usage and
year-to-year improvement or decline, 1978-2005
Minimum 340 carries (60 players)
|Year-to-year change in...||Yards||Yd/Car||Carries|
Minimum 300 carries and 25 receptions (121 players)
|Year-to-year change in...||Yards||Yd/Car||Carries|
Using a larger sample of players not only gives the same result, but makes the correlation between receptions and improvement in rushing yardage more significant. 121 players between 1978 and 2004 had 300 or more carries and 25 or more receptions. For these running backs, the correlation between carries and rushing yardage the next year was -.17, but the correlation between receptions and rushing yardage the next year was .20. That's two relationships of similar strength in opposite directions.
(The reaction of most statisticians at this point would be that these correlations are extremely small. This is simply a fact of life when talking about the NFL, where so many factors contribute to a player or team's performance -- many of them intangible -- that no one factor will have a particularly large impact on its own.)
So if more receptions don't mean a greater chance of breakdown the following year, what about playoff carries?
This is where we have bad news for Seattle fans. It does look like postseason carries matter, with 390 carries total forming a barrier equivalent to 370 regular-season carries. Above that line, a number of players were either injured or lost effectiveness. And while Shaun Alexander just barely touched the 370-carry barrier, he flew past 390 carries once the Seahawks got into the postseason, ending the playoffs with 430 carries total.
Not counting Alexander, there have been 14 players who did not reach 370 carries in the regular season, but surpassed 390 carries during the postseason. Though some of these players continued to play well the next season and even afterwards, a number of them had major difficulties.
Terrell Davis (1997): 369 carries, 481 including the postseason. The latter total is an NFL record. Davis was spectacular again in 1998, but that season's total of 392 regular-season carries basically ended his career.
Eddie George (1999): 320 carries, 428 including the postseason. He fell from 4.1 to 3.7 yards per carry in 2000, and after 403 regular-season carries he was never again an effective player.
Curtis Martin (1998): 369 carries, 418 including the postseason. Martin saw no ill effects; 1998 was actually his worst year until 2005.
Thurman Thomas (1993): 355 carries, 418 including the postseason. Thomas continued to play well but never again was able to carry the ball 300 times in a season.
Joe Morris (1986): 341 carries, 414 including the postseason. The following year Morris plummeted from 4.4 to 3.4 yards per carry. By 1989, his career was over due to nerve damage and broken bones in his feet, except for a short-lived comeback with the 1991 Browns.
Jamal Lewis (2000): 309 carries, 412 including the postseason. 103 postseason carries is the third-highest total in history, and Lewis tore his ACL the next year.
Corey Dillon (2005): 345 carries, 410 including the postseason. Fell from 4.7 to 3.5 yards per carry and only managed 12 games due to injuries.
Emmitt Smith (1991): 365 carries, 406 including the postseason. No ill effects.
Ahman Green (2003): 355 carries, 403 including the postseason. Dropped from 5.3 yards per carry in 2003 to 4.5 yards per carry in 2004 and then 3.3 yards per carry in 2005, when he missed 11 games because of injuries.
Earl Campbell (1979): 368 carries, 401 including the postseason. Improved in 1980, when another heavy workload cost him most of his effectiveness in 1981 and beyond.
Natrone Means (1994): 343 carries, 400 including the postseason. Only played 10 games the following year due to injuries, never again played a full season, and retired in 2000 at the age of 28.
Dorsey Levens (1997): 329 carries, 400 including the postseason. Only played seven games in 1998, never again had 100 carries in a season after 1999, didn't average four yards per carry again until 2002.
Curt Warner (1983): 335 carries, 395 including the postseason. Blew out his knee during the first game of 1984 and was out for the season.
Emmitt Smith (1994): 368 carries, 395 including the postseason. Had his best season in 1995, then declined after that.
To summarize, eight of these 14 players were injured or lost effectiveness the following season. A ninth, Thomas, was still effective and healthy, but lost stamina. Three players, Davis, Campbell, and Smith (1994) had problems two seasons later, after another year of overuse. Only two of these players, Martin and Smith (1991) seemed to have no ill effects for multiple seasons afterwards.
On average, running backs with 300 to 369 carries who do not play in the postseason will see total yards drop by 15 percent the following year, and yards per carry by just two percent. But the 14 players listed above averaged a 27 percent drop in total yards, and a 10 percent drop in yards per carry.
All players with 390 or more carries, no matter how these carries were split between the regular season and the postseason, averaged a 33 percent drop in total yards, and an 11 percent drop in yards per carry.
Alexander's total of 430 carries between the regular season and the postseason ranks sixth in NFL history, and no other running back last year reached 390 carries combined. Only three other running backs were above 350 carries combined: Clinton Portis (385), Edgerrin James (373), and Tiki Barber (370).
So chalk up another reason to believe that Alexander will decline this year, to go with the Madden Curse, the departure of Steve Hutchinson, and plain old regression to the mean. The Seahawks can live with a little decline -- because Alexander played at such a high level last year, a little decline would still leave one of the top running backs in football. What the Seahawks have to worry about is that other problem with overuse: increased chance of injury. If Alexander is hurting and Maurice Morris is starting come midseason, that new contract with $15.1 million in guaranteed bonus and salary for 2006 won't look like such a good decision.