The Curse of 2000 Carries (1 Viewer)

[TheDirtyWord]

I had begun to do a Player Spotlight analysis on Clinton Portis (who I'm bearish) on in 2009. And a large part of my argument was going to center around workload. One thing led to another and I wound up delving into if there was a benchmark you could predict when an RB's production would start to decline.

There have been 30 RB's with 2000 or more career carries. 14 of them have played in the last 10 years, but for the purposes of this exercise only 11 were used because 1) they had yet to accumulate enough post 2000 carry data (Clinton Portis & Ahman Green) or 2) including their post 2000 carry data was very insignificant (Shaun Alexander).

So the 11 RB's evaluated were:

Emmitt Smith

Curtis Martin

Jerome Bettis

Edgerrin James

Marshall Faulk

Corey Dillon

Fred Taylor

Warrick Dunn

Eddie George

LaDainian Tomlinson

Jamal Lewis

The basis of the evaluation was how did these RB's produce after their 2000th carry. And that was further defined by evaluating their season after they acquired their 2000th carry. So if an RB got his 2000th carry in 2004 and finished with 2100 career carries that season, the split would be made between the 2004 and 2005 seasons. The only distinction was made for Warrick Dunn where I used is 1970 carry split after the 2005 season.

Pre-2000 Carry Seasons

170 Games

169 Games Started

3429 Rushing Attempts

15211 Rushing Yards

128 Rushing TD's

4.4 YPC

470 Receptions

3695 Receiving Yards

20 Receiving TD's.

Post-2000 Carry Seasons

168 Games

162 Games Started

3219 Rushing Attempts

12886 Rushing Yards

96 Rushing TD's

4.0 YPC

403 Receptions

2737 Receiving Yards

12 Receiving TD's

So how does this translate when evaluating the average season of a RB crossing this threshold when simply taking the season they acheived 200 career carries and the season after?

Pre-2000 Average Season: 1718.8 Yards From Scrimmage, 13.5 RRTD's.

Post-2000 Average Season: 1420.3 Yards From Scrimmage, 9.3 RRTD's

From a YFS standpoint, that is 17.36% percentage point drop and the an RRTD perspective, the drop is event steeper (27.03%).

Now 2000 carries is alot of carries, so it's no crime for an RB to start running out of gas at this point in his career. Just be careful when sniffing around the likes of Clinton Portis and Thomas Jones (who is close to the 2000 number - 1949).

 

[CalBear]

Is there a particular reason you chose 2000 carries? Because this is a perfectly normal populational regression to the mean, and I'd expect you'd find the same thing at 1500, or 2500 carries.

 

[TheDirtyWord]

Is there a particular reason you chose 2000 carries? Because this is a perfectly normal populational regression to the mean, and I'd expect you'd find the same thing at 1500, or 2500 carries.

Beside being a round number, it also represented a fairly solid barometer of a significant career. 1500 would actually not add too many other players to the pool believe it or not (Thomas Jones, Ahman Green and Ricky Williams). I don't know if you would find close to 20% drops in production at that level. if anything, you might see an increase. At 2500 the drop is event further, but the sample size is smaller. Tomlinson would be interesting to see through this lens since he's at 2657. But only Dunn, James, George and Faulk would be comparison points.

 

[Couch Potato]

I believe it is age, not carries, that matter.

Find RBs at age 28 (Portis' age to begin the season) who have 2,000 carries and declined and then I'll be interested. Most RBs decline after 30-31, and it's coincidental that it's also when they reach the 2,000 carry total. RBs who reach 31 typically decline without the 2,000 carries too, so I don't think carries are the factor to focus on. I think Portis will be fine for the next 2-3 years..

 

[dgreen]

Five of those backs had better fantasy rushing points in year n+1 than they did in year n. Six had worse numbers.

Of the six who performed worse in year n+1, four were older than Portis (27 going on 28). Faulk was 28 going on 29, Lewis 28/29, Dillon 30/31, and Dunn 30/31. Edge and Tomlinson had worse years in year n+1. However, Edge went from a powerhouse offense in Indianapolis to the Cardinals and Tomlinson was coming off one of the best rushing seasons ever (1815 yds and 28 TDs; he had nowhere to go but down).

Of the five who got better, Martin and Bettis were of similar age to Portis (again, 27/28). Emmitt was a year younger, 26 going on 27. Eddie George was 28/29 and coming off what appears to be an absolutely horrendous season where he only posted 939/5 in year n. And, the fifth is Fred Taylor who posted a negligible increase.

 

[CalBear]

Is there a particular reason you chose 2000 carries? Because this is a perfectly normal populational regression to the mean, and I'd expect you'd find the same thing at 1500, or 2500 carries.

Beside being a round number, it also represented a fairly solid barometer of a significant career. 1500 would actually not add too many other players to the pool believe it or not (Thomas Jones, Ahman Green and Ricky Williams). I don't know if you would find close to 20% drops in production at that level. if anything, you might see an increase.

Uh, no. Ricky Williams reached 1500 carries in 2003; up until that point he'd averaged 1270 rushing yards and 9TD per year. Since then he's averaged 472 yards and 3.6 TDs, if you leave out his two years out of the league. Green averaged 1025/8.9, since then he's averaged 467/2.75. Thomas Jones has just one season since reaching 1500 carries; it was a good one, but if you give him a couple more years, his average will also crater.

 

[switz]

I believe it is age, not carries, that matter.

There is no fixed determinant. Age plays a role, as does workload. Workload at certain ages appears to impact players differently.For instance, an RB that gets 362.3+ carries at age 27 is 83% more likely to be injured the following season than one who receives the same number of carries at 22 years old. Now if the 27 year old RB had already carried the ball 1702 times in his career than that likelihood to get injured increases. Of course the situation is that the more carries you look at per season, and per career, the smaller a sample size you have. Nonetheless, a combination of age, carries over career, and carries a season do seem to indicate an inclination to wear down at certain points.

 

[TheDirtyWord]

Is there a particular reason you chose 2000 carries? Because this is a perfectly normal populational regression to the mean, and I'd expect you'd find the same thing at 1500, or 2500 carries.

Beside being a round number, it also represented a fairly solid barometer of a significant career. 1500 would actually not add too many other players to the pool believe it or not (Thomas Jones, Ahman Green and Ricky Williams). I don't know if you would find close to 20% drops in production at that level. if anything, you might see an increase.

Uh, no. Ricky Williams reached 1500 carries in 2003; up until that point he'd averaged 1270 rushing yards and 9TD per year. Since then he's averaged 472 yards and 3.6 TDs, if you leave out his two years out of the league. Green averaged 1025/8.9, since then he's averaged 467/2.75. Thomas Jones has just one season since reaching 1500 carries; it was a good one, but if you give him a couple more years, his average will also crater.

But you're not including the 2000 carry club in your evaluation.

 

[CalBear]

Is there a particular reason you chose 2000 carries? Because this is a perfectly normal populational regression to the mean, and I'd expect you'd find the same thing at 1500, or 2500 carries.

Beside being a round number, it also represented a fairly solid barometer of a significant career. 1500 would actually not add too many other players to the pool believe it or not (Thomas Jones, Ahman Green and Ricky Williams). I don't know if you would find close to 20% drops in production at that level. if anything, you might see an increase.

Uh, no. Ricky Williams reached 1500 carries in 2003; up until that point he'd averaged 1270 rushing yards and 9TD per year. Since then he's averaged 472 yards and 3.6 TDs, if you leave out his two years out of the league. Green averaged 1025/8.9, since then he's averaged 467/2.75. Thomas Jones has just one season since reaching 1500 carries; it was a good one, but if you give him a couple more years, his average will also crater.

But you're not including the 2000 carry club in your evaluation.

For the people who had 1500 and not 2000 carries, the dropoff is greater than 20%, therefore, when you add them to the 2000 carry group, the relative performance will get even worse. So 1500 is a better dividing line than 2000 carries, but that's really not the point--the point is that the idea of a dividing line is fallacious.All you're measuring is the fact that RBs (and other players) tend to have weaker seasons towards the end of their careers. You're not measuring anything about 2000 carries as a dividing line, and you're definitely not doing anything that has any relevance to Clinton Portis.

 

[Michael Fox Fan]

Is there a particular reason you chose 2000 carries? Because this is a perfectly normal populational regression to the mean, and I'd expect you'd find the same thing at 1500, or 2500 carries.

Beside being a round number, it also represented a fairly solid barometer of a significant career. 1500 would actually not add too many other players to the pool believe it or not (Thomas Jones, Ahman Green and Ricky Williams). I don't know if you would find close to 20% drops in production at that level. if anything, you might see an increase.

Uh, no. Ricky Williams reached 1500 carries in 2003; up until that point he'd averaged 1270 rushing yards and 9TD per year. Since then he's averaged 472 yards and 3.6 TDs, if you leave out his two years out of the league. Green averaged 1025/8.9, since then he's averaged 467/2.75. Thomas Jones has just one season since reaching 1500 carries; it was a good one, but if you give him a couple more years, his average will also crater.

But you're not including the 2000 carry club in your evaluation.

For the people who had 1500 and not 2000 carries, the dropoff is greater than 20%, therefore, when you add them to the 2000 carry group, the relative performance will get even worse. So 1500 is a better dividing line than 2000 carries, but that's really not the point--the point is that the idea of a dividing line is fallacious.All you're measuring is the fact that RBs (and other players) tend to have weaker seasons towards the end of their careers. You're not measuring anything about 2000 carries as a dividing line, and you're definitely not doing anything that has any relevance to Clinton Portis.

Exactly. This "analysis" is borderline meaningless.

 

[Moniker]

Exactly. This "analysis" is borderline meaningless.

Most are. Every player is different and players today are in better overall shape than players were ten years ago.

 

[TheDirtyWord]

Is there a particular reason you chose 2000 carries? Because this is a perfectly normal populational regression to the mean, and I'd expect you'd find the same thing at 1500, or 2500 carries.

Beside being a round number, it also represented a fairly solid barometer of a significant career. 1500 would actually not add too many other players to the pool believe it or not (Thomas Jones, Ahman Green and Ricky Williams). I don't know if you would find close to 20% drops in production at that level. if anything, you might see an increase.

Uh, no. Ricky Williams reached 1500 carries in 2003; up until that point he'd averaged 1270 rushing yards and 9TD per year. Since then he's averaged 472 yards and 3.6 TDs, if you leave out his two years out of the league. Green averaged 1025/8.9, since then he's averaged 467/2.75. Thomas Jones has just one season since reaching 1500 carries; it was a good one, but if you give him a couple more years, his average will also crater.

But you're not including the 2000 carry club in your evaluation.

For the people who had 1500 and not 2000 carries, the dropoff is greater than 20%, therefore, when you add them to the 2000 carry group, the relative performance will get even worse. So 1500 is a better dividing line than 2000 carries, but that's really not the point--the point is that the idea of a dividing line is fallacious.All you're measuring is the fact that RBs (and other players) tend to have weaker seasons towards the end of their careers. You're not measuring anything about 2000 carries as a dividing line, and you're definitely not doing anything that has any relevance to Clinton Portis.

This is incorrect. You are evaluating the 2000 carry club at their 2000th carry, not their 1500th.I'll include Thomas Jones and Ahman Green at their 1500th carry, but let's agree that after his 1500th carry, Ricky Williams 'retired' to smoke pot. Taking his stats after he restarted his career (and then got suspended again shortly thereafter) isn't really apples to apples.

Pre-1500 Carry Seasons

199 Games

197 Games Started

4033 Carries

14291 Rushing Yards

108 Rushing TD's

4.1 YPC

558 Receptions

4473 Receiving Yards

14 Receiving TD's

Post-1500 Carry Seasons

187 Games

183 Games Started

3535 Carries

15172 Rushing Yards

131 Rushing TD's

4.3 YPC

504 Receptions

3936 Receiving Yards

16 Receiving TD's

Pre-1500 Carry Season: 1619 Yards From Scrimmage; 9.4 RRTD's

Post-1500 Carry Season: 1470 Yards From Scrimmage; 11.3 RRTD's

Also consider that the the average post-1500 carry season averaged out to about one less game per season and the YFS difference is negligible (it's 90% of the pre-1500 carry number as it is) and still scored more TD's. The difference between the performance at the 2000 carry level is a much more pronounced decline. The stats here show that there is really very little performance difference on either side of 1500.

 

[perry147]

I like statistical analysis for many things in fantasy football, but IMHO you are correlating global warming with pirate populations.

In order words, the age of the RB, his workload (amount of carries per year) and injury history have far more to do with the ability to predict future production than the lie in the sand 2000 carry mark.

Just my 2 cents.

 

[TheDirtyWord]

I like statistical analysis for many things in fantasy football, but IMHO you are correlating global warming with pirate populations.In order words, the age of the RB, his workload (amount of carries per year) and injury history have far more to do with the ability to predict future production than the lie in the sand 2000 carry mark. Just my 2 cents.

Why is age any less arbitrary than career carries? LaDainian Tomlinson and Brian Westbrook are 2 months apart (both will be 30 upon the start of the season). One has 2657 career carries, the other 1247. I know you mentioned workload from a carries/year standpoint, but that's simply another derivative of the original theory. So it shouldn't be just age, in fact age may be a deceiving factor in an RB's career. What this exercise pointed out was that 2000 was a fairly solid turning point to start being cautious when evaluating RB's. Portis and Thomas Jones are the targets this year, but perhaps Westbrook is being undervalued because while his age is entering danger zone, his workload isn't.

 

[switz]

I like statistical analysis for many things in fantasy football, but IMHO you are correlating global warming with pirate populations.

In order words, the age of the RB, his workload (amount of carries per year) and injury history have far more to do with the ability to predict future production than the lie in the sand 2000 carry mark.

Just my 2 cents.

Why is age any less arbitrary than career carries? LaDainian Tomlinson and Brian Westbrook are 2 months apart (both will be 30 upon the start of the season). One has 2657 career carries, the other 1247. I know you mentioned workload from a carries/year standpoint, but that's simply another derivative of the original theory. So it shouldn't be just age, in fact age may be a deceiving factor in an RB's career. What this exercise pointed out was that 2000 was a fairly solid turning point to start being cautious when evaluating RB's. Portis and Thomas Jones are the targets this year, but perhaps Westbrook is being undervalued because while his age is entering danger zone, his workload isn't.

Ummm.... I don't think perry147 said age was alone the factor. However, age does player a more significant role than you want to think IMO.

 

[perry147]

I like statistical analysis for many things in fantasy football, but IMHO you are correlating global warming with pirate populations.In order words, the age of the RB, his workload (amount of carries per year) and injury history have far more to do with the ability to predict future production than the lie in the sand 2000 carry mark. Just my 2 cents.

Why is age any less arbitrary than career carries? LaDainian Tomlinson and Brian Westbrook are 2 months apart (both will be 30 upon the start of the season). One has 2657 career carries, the other 1247. I know you mentioned workload from a carries/year standpoint, but that's simply another derivative of the original theory. So it shouldn't be just age, in fact age may be a deceiving factor in an RB's career. What this exercise pointed out was that 2000 was a fairly solid turning point to start being cautious when evaluating RB's. Portis and Thomas Jones are the targets this year, but perhaps Westbrook is being undervalued because while his age is entering danger zone, his workload isn't.

Westbrook's injury with regards to his knee and ankle are more worrisome than his career carries. He might not even be able to start the season and at the least will miss some training camp.

 

[TheDirtyWord]

I like statistical analysis for many things in fantasy football, but IMHO you are correlating global warming with pirate populations.

In order words, the age of the RB, his workload (amount of carries per year) and injury history have far more to do with the ability to predict future production than the lie in the sand 2000 carry mark.

Just my 2 cents.

Why is age any less arbitrary than career carries? LaDainian Tomlinson and Brian Westbrook are 2 months apart (both will be 30 upon the start of the season). One has 2657 career carries, the other 1247. I know you mentioned workload from a carries/year standpoint, but that's simply another derivative of the original theory. So it shouldn't be just age, in fact age may be a deceiving factor in an RB's career. What this exercise pointed out was that 2000 was a fairly solid turning point to start being cautious when evaluating RB's. Portis and Thomas Jones are the targets this year, but perhaps Westbrook is being undervalued because while his age is entering danger zone, his workload isn't.

Ummm.... I don't think perry147 said age was alone the factor. However, age does player a more significant role than you want to think IMO.

I recognize that age HAS to be a component because in order to accumulate 2000 career carries, you have to be around for awhile. But what age? 28? 29? 31? We can identify an RB is getting old, but can we really narrow down at what age it begins to affect his performance in a negative way?Nothing is absolute. I recognize that there will be outliers and exceptions. But we look for odds that sway our thinking one way or the other. This is one way to look at and try and predict when that decreased performance will hit that I've not seen brought up before.

 

[switz]

I posted this in the Portis thread where TDW brought up his theory... this should put an end to this theory pretty easily:

Barry Sanders reach 2000 carries during week 14 of his 7th season, he was 27. He averaged 4.9 YPC and 1453 yards per season through 2000 carries. Sanders played three more seasons after his 2000th carry, in which he averaged 5.2 YPC and 1699 yards per season. In fact, Sanders BEST season 2,053 yards and 11 TDs came AFTER his 200th carry. Oh, and his second best season came in the season he reached 1,500 carries, so that's bunk too.

LTs best season came after 1,500 carries, and during his 2,000 carry season, he was also 27. The game after he reached 2,000 carries he had 123 yards at 5.6 YPC, better than his season average that year. The following season, he averaged 4.7 YPC, better than his career average at that point, and put up 1474 yards rushing, only 55 yards less than his career season average, and had a better season than 3 of his prior 6 seasons.

Interestingly, Emmitt also reached his 2,000th carry at age 27. However, his decline after that season had nothing to do with how many carries he had accumulated, but bad coaching, loss of Jay Novacek, and general loss of team talent over 2 years without Jimmy Johnson running the draft.

 

[perry147]

I like statistical analysis for many things in fantasy football, but IMHO you are correlating global warming with pirate populations.

In order words, the age of the RB, his workload (amount of carries per year) and injury history have far more to do with the ability to predict future production than the lie in the sand 2000 carry mark.

Just my 2 cents.

Why is age any less arbitrary than career carries? LaDainian Tomlinson and Brian Westbrook are 2 months apart (both will be 30 upon the start of the season). One has 2657 career carries, the other 1247. I know you mentioned workload from a carries/year standpoint, but that's simply another derivative of the original theory. So it shouldn't be just age, in fact age may be a deceiving factor in an RB's career. What this exercise pointed out was that 2000 was a fairly solid turning point to start being cautious when evaluating RB's. Portis and Thomas Jones are the targets this year, but perhaps Westbrook is being undervalued because while his age is entering danger zone, his workload isn't.

Ummm.... I don't think perry147 said age was alone the factor. However, age does player a more significant role than you want to think IMO.

I recognize that age HAS to be a component because in order to accumulate 2000 career carries, you have to be around for awhile. But what age? 28? 29? 31? We can identify an RB is getting old, but can we really narrow down at what age it begins to affect his performance in a negative way?Nothing is absolute. I recognize that there will be outliers and exceptions. But we look for odds that sway our thinking one way or the other. This is one way to look at and try and predict when that decreased performance will hit that I've not seen brought up before.

The point of my post is that you can't use statistical analysis of amount of carries as a good predictor of future performance in regards to the NFL RBs; there are vastly more factors involved.