ConstruxBoy Football Thoughts

Does College Running Back workload relate to NFL Success?

2009-05-28T12:28:00.000-05:00

I. Introduction

When considering the future NFL success of a college running back, there is usually a consideration of his workload in college. Sometimes there are thoughts that too much work will lead to injuries in the NFL. Other times it could be that too little work means a back is not ready for a full time NFL workload. Or maybe a large workload shows that the back can handle a large workload in the NFL? I decided to see if I could find any correlation between college workload and NFL success in terms of fantasy points or games missed.

II. Parameters of the Study

I started by looking at all the running backs drafted in the first 3 rounds since 2000. This gives a dataset of 54 running backs. For each running back I found the following values for their last season in college: Rushes, Catches, Touches, Games and Touches per Game.

Next I looked at their NFL careers. For each running back I calculated their fantasy points (pt/10 yards rush or rec, 1 ppr, 6 pts for TDs) in their first year in the league as well as for their career. I also found the number of career games they have played and compared it to a full NFL season of 16 games to find a career missing game percentage.

Finally, I calculated 3 sets of Pearson Correlations to see if any of the 4 college workload factors related to 1^st year fantasy points, career fantasy points or career missing percentage.

NOTES: It likely goes without saying that running backs drafted in the last few years may not show accurate career values yet, but I think it was worth using them to get a sample size of 54 backs. Also, missing games may be because the player is not ready yet or stuck behind a veteran back, not just due to injury. However I saw no way to separate those out and feel like the effect should be similar.

III. Results

In a nutshell, the results showed almost no correlation between the workload of a college running back in their last collegiate season and their success and lack of missed games in the NFL. You can find the spreadsheet here. The first tab shows the Raw Data. The second tab shows the Correlations. The third tab shows the collegiate workloads of 14 top RBs in the 2008 draft.

Here is the Correlation table:

Variables	Pearson Correlation
Rushes to 1st FP/G	0.09649
Catches to 1st FP/G	-0.09311
Touches to 1st FP/G	0.07342
T/G to 1st FP/G	0.11392
Rushes to Car FP/G	0.17081
Catches to Car FP/G	0.15522
Touches to Car FP/G	0.19064
T/G to Car FP/G	0.18586
Rushes to Car Miss Pct	-0.03248
Catches to Car Miss Pct	-0.00551
Touches to Car Miss Pct	-0.03170
T/G to Car Miss Pct	-0.01238

The highest number in the set is .19 for the Touches to Career FP/G. This doesn’t show much of a relation, but it is interesting that all of the Career comparisons have the highest Pearson numbers. They are also all positive, so if we can take anything from the study, it would seem to be that more touches in a college running backs final season should lead to greater NFL success. However that is obviously a weak assertion.

It’s interesting to see such small values next to the Career Missed Games Percentage comparison. As discussed above, there are reasons besides injury for missing games. However I still would have expected high workload RBs to miss more games in the NFL.

Looking at the raw data is very interesting. The three backs who averaged over 30 touches a game are LaDainian Tomlinson, Brian Calhoun and Steven Jackson. Two of those 3 are very good NFL running backs and the 3^rd has been injured in his short career. Of course, right after those three guys are some backs who have been less successful in their NFL careers: Chris Perry, Cedric Benson, William Green and DeShaun Foster. So you can see by looking at the raw data that merely ranking players by touches per game will not lead you to the very best NFL backs.

Looking at the lower end of the spectrum, players like Brian Leonard, Greg Jones, Eric Shelton and Lorenzo Booker are at the bottom of the list. None have showed much in the NFL so far although Leonard and Booker are still young. However right above those four are some very successful running backs that had light collegiate workloads like Maurice Jones-Drew, Ronnie Brown and Joseph Addai. Again, it’s easy to see that a low number of touches are not an indicator of poor future performance.

Finally, it’s interesting to compare this year’s crop of college running backs to those from the past. The most interesting thing to note is that the previous record for touches in a season in this data set was 401 by Brian Calhoun. That record is shattered this season with Kevin Smith’s 474 touches for Central Florida. Ray Rice also beat Calhoun with 404 touches. In fact, adding in Matt Forte’s 393 touches puts 3 of the top 5 workloads in this year’s class, with Steven Jackson just ahead of Forte with 394 for 4^th place. It’s probably just a quirk, but it is interesting to see. Smith, Rice and Forte also all averaged over 30 touches a game, which had only been done by Tomlinson, Calhoun and Jackson among the past 54 RBs. I’m sure fantasy owners will be happy if this year’s trio has the same level of NFL success as those three.

On the low level, there are two interesting players to check. The first is Felix Jones, who at 11.46 touches per game would be the second lowest in that category of the data set. That puts him just above Brian Leonard and just below Greg Jones and Eric Shelton. The other player to watch is Ryan Torain, who got hurt after 6 games last season. His 117 touches in those games are the 2007 low, but he had a respectable 19.50 touches per game. The only player with a lower number of touches in his final season was Tony Hollings, who had 95. However, Hollings got those touches in only 4 games for a 23.75 per game average, good enough for 22^nd out of the 54 RBs in the data set.

On a side note, let’s hear it for LaDainian Tomlinson. He is ranked 1^st in rushes, 4^th in touches, 1^st in touches/game, 4^th in first year fantasy points per game, 1^st in career fantasy points per game and has only missed one game out of 112. That is a stud.

IV. Conclusions

Looking at the results of this study, it appears that there is not a correlation between college running back workload and NFL success nor is there a relationship between heavy workload and more games missed in the NFL. However, I think that showing that there is not a relationship or correlation is useful information for evaluating the NFL prospect of college running backs. The ideas that a heavy workload is bad due to greater injury propensity in the NFL or that a light workload means that a running back can’t handle a pounding in the NFL should be taken with a grain of salt after reviewing this study.

In retrospect, a way to make the study more accurate might be to include the workload over a running back’s entire college career instead of only their last season. We could also use more than the first 3 rounds of drafted running backs to get a larger data set than 54 runners. However, even with these minor quibbles, I feel confident in excluding these worries for the most part when I evaluate future college running backs.

Fantasy Points by Draft Position

2009-03-25T15:12:00.001-05:00

Fantasy Point Distribution by NFL Draft Round
By John Powderly

I: Introduction

I was recently thinking about how excited people get about fantasy football sleepers. Someone who drafted Marques Colston last season was estatic and likely making sure their fellow owners knew all about their sleeper-finding skills. That got me wondering how often a lower round NFL draft pick really makes a mark in fantasy football. I have always assumed that most of the top fantasy football scorers are higher draft picks, but I had no way of knowing how true that is and whether there is any sort of pattern to the fantasy point distribution among NFL draft rounds. So I decided to try to find out.

II: Parameters of the Study

I decided to take the top 30 fantasy point performers at QB, RB, WR and TE over the last 5 years and find out in what round each player was drafted in the NFL Draft. To address positional inequalities in fantasy football lineups, I also compared an average 12 team starting lineup of 1 QB, 2 RB, 3 WR and 1 TE (ie. Top 12 QBs, Top 24 RBs, etc) over the last 5 years. I was going to compare both total fantasy point distribution, as well as draft round distribution. In other words, how many of the top 30 QBs last season were drafted in the 1st round? How many of the Top 30 from 2002 were drafted in the 3rd round? I believe that examining these data points may help me determine how often a lower round draft pick contributes to a fantasy football team and whether there are any lessons to be learned that could help me in the future.

A trickier part of the study was the fact that it wasn't possible to know exactly which years of NFL draft data would perfectly coincide with 5 years of fantasy point data. Someone who was a top 24 RB in 2003 could have been drafted back in 1991. But using the entire 1991 draft as a data point for fantasy points in 2002 through 2006 was probably unfair because of the short duration of the average football player's career. There is no way to find a perfect match, but for this study I decided to go back through 10 years of NFL draft data. So I will examine drafts from 1997 through 2006 and compare to fantasy points from 2002 through 2006. This should give me a large enough dataset to make some reasonable conclusions.

III: Results of the Study

The data set is available in spreadsheet form here. Let's look at the results by position.

Quarterback:

Here are the Draft Slot and Fantasy Point distributions for the Top 30 QBs from 2002 to 2006 (or 150 QBs in total)

Round 1 - 68 or 45.3% of the Top 150 QBs were drafted in Round 1; 47.07% of fantasy points were from Round 1 draftees

Round 2 - 17 or 11.3% of the slots; 11.89% of fantasy points

Round 3 - 6 or 4% of the slots; 3.02% of fantasy points

Round 4 - 7 or 4.7% of the slots; 4.96% of fantasy points

Round 5 - 3 or 2% of the slots; 1.7% of fantasy points

Round 6 - 20 or 13.3% of the slots; 13.33% of fantasy points

Round 7 - 2 or 1.3% of the slots; 1.02% of fantasy points

Round 8 - 4 or 2.7% of the slots; 3.41% of fantasy points

Round 9 - 4 or 2.7% of the slots; 2.36% of fantasy points

UDFA (UnDraftedFreeAgent) - 19 or 12.7% of the slots; 11.24% of fantasy points

In examining these numbers, the first thing that jumps out is the outlier of the round 6 QBs. The second most Top 30 slots are from 6th round Quarterbacks. Everyone knows about Tom Brady, but Marc Bulger and Matt Hasselbeck also contributed to the 6th round QB successes.

Another noticable result is that the draft slot percentage and fantasy points percentage are pretty closely tied. I expected an even distribution, but wanted to see if there was a case where, besides having the most slots, a particular round had a much higher percentage of fantasy points within those slots. That doesn't seem to be the case here.

The final observation from these results is that the most draft slots and fantasy points came from QBs drafted in the 1st round. In fact, I added up the first three rounds to come up with 60.6% of the Top 30 scorers over the last 5 years have been first day draftees.

Next, I did the same examination for the Top 12 QBs each year (60 QBs total), or those that should be starting in a 12 team, Start 1 QB fantasy league. Here are the results:

Round 1 - 26 slots or 43.33%

Round 2 - 9 slots or 15%

Round 3 - 0 slots

Round 4 - 4 slots or 6.67%

Round 5 - 0 slots

Round 6 - 10 slots or 16.67%

Round 7 - 0 slots

Round 8 - 4 slots or 6.67%

Round 9 - 1 slot or 1.67%

UDFA - 6 slots or 10%

So we again see that Round 1 has the most Top 12 fantasy points finishes and Round 6 still does surprisingly well.

Running Back:
Here are the Draft Slot and Fantasy Point distributions for the Top 30 RBs from 2002 to 2006 (or 150 RBs in total)

Round 1 - 71 or 47.33% of the Top 150 RBs were drafted in Round 1; 48.69% of fantasy points were from Round 1 draftees

Round 2 - 26 or 17.33% of the slots; 18.29% of fantasy points

Round 3 - 25 or 16.67% of the slots; 15.43% of fantasy points

Round 4 - 15 or 10% of the slots; 8.9% of fantasy points

Round 5 - 1 or 0.67% of the slots; 0.44% of fantasy points

Round 6 - 2 or 1.33% of the slots; 1.28% of fantasy points

UDFA (UnDraftedFreeAgent) - 10 or 6.67% of the slots; 6.96% of fantasy points

In examining these numbers, the first thing that jumps out is the expected downward trend of Top 30 slots with draft round. The first round has the most, followed by the 2nd, followed by the 3rd, etc. This seems to indicate to me that, in general, NFL teams knew what they were doing when drafting running backs. The ones drafted in the first round had a greater chance of contributing in terms of fantasy points than those drafted in the 2nd round, and so on. Of course, there also may be more pressure for the coach to play a first round RB, thus increasing his points. Either way, as a fantasy owner, you win.

Again notice that the draft slot percentage and fantasy points percentage are pretty closely tied. This mirrors the QB position and lends more weight to the conclusion that there is an even distribution of rounds within the Top 30.

A final observation is the much greater percentage of First Day draftees on the list than at the QB position. The First Day RBs account for 81.33% of the Top 30 fantasy scorers over those 5 years. This is far greater than the 60.6% for QBs. One possible explaination for this is that running backs are easier to evaluate than quarterbacks, so it is far less often that all teams "miss" on a great player until the later rounds. Priest Holmes and Willie Parker are really uncommon exceptions.

Next, I did the same examination for the Top 24 RBs each year (120 RBs total), or those that should be starting in a 12 team, Start 2 RB fantasy league. Here are the results:

Round 1 - 57 slots or 47.5%

Round 2 - 24 slots or 20%

Round 3 - 19 slots or 15.83%

Round 4 - 11 slots or 9.16%

Round 5 - 0 slots

Round 6 - 2 slots or 1.67%

UDFA - 7 slots or 5.83%

So we again see that Round 1 has the most Top 24 fantasy points finishes and that the first day is a very high 83.33%.

Wide Receiver:
Here are the Draft Slot and Fantasy Point distributions for the Top 30 WRs from 2002 to 2006 (or 150 WRs in total)

Round 1 - 69 or 46% of the Top 150 WRs were drafted in Round 1; 45.75% of fantasy points were from Round 1 draftees

Round 2 - 33 or 22% of the slots; 21.46% of fantasy points

Round 3 - 21 or 14% of the slots; 15.18% of fantasy points

Round 4 - 7 or 4.67% of the slots; 4.32% of fantasy points

Round 5 - 3 or 2% of the slots; 2.28% of fantasy points

Round 6 - 0 slots

Round 7 - 7 or 4.67% of the slots; 4.85% of fantasy points

Round 12 - 2 or 1.33% of the slots; 1.31% of fantasy points

UDFA (UnDraftedFreeAgent) - 8 or 5.33% of the slots; 4.85% of fantasy points

In examining these numbers, we see the same downward trend as we did with the running backs. So again, it seems like the NFL knows how to evaluate WRs and that generally the higher a WR is drafted, the more likely he is to be a Top 30 performer in fantasy points.

As with the Running Backs, the First Day percentage on WRs is very high. 82% of the Top 30 WRs from 2002 through 2006 were drafted on the first day of the draft.

You have to love 12th rounder Keenan McCardell finishing in the Top 30 twice. In fact, he finished 10th and 17th. I guess he showed those scouts.

Keenan highlights an interesting trend: Several of the lower round or UDFA WRs are older WRs who were drafted awhile ago, like McCardell, Rod Smith, Wayne Chrebet and Donald Driver. This could mean that as the years go by, NFL teams get better and better at scouting talent and fewer players slip through the cracks. However, there is not enough data to come to a solid conclusion about this trend, as Colston has proven.

Next, I did the same examination for the Top 36 WRs each year (180 WRs total), or those that should be starting in a 12 team, Start 3 WR fantasy league. Here are the results:

Round 1 - 77 slots or 42.78%

Round 2 - 39 slots or 21.67%

Round 3 - 27 slots or 15%

Round 4 - 10 slots or 5.56%

Round 5 - 3 slots or 1.67%

Round 6 - 2 slots or 1.11%

Round 7 - 8 slots or 4.44%

Round 12 - 2 slots or 1.11%

UDFA - 12 slots or 6.67%

So we again see that Round 1 has the most Top 36 fantasy points finishes.

Tight End:
Here are the Draft Slot and Fantasy Point distributions for the Top 30 TEs from 2002 to 2006 (or 150 TEs in total)

Round 1 - 39 or 26% of the Top 150 TEs were drafted in Round 1; 31.45% of fantasy points were from Round 1 draftees

Round 2 - 32 or 21.33% of the slots; 19.22% of fantasy points

Round 3 - 20 or 13.33% of the slots; 12.87% of fantasy points

Round 4 - 10 or 6.67% of the slots; 6.44% of fantasy points

Round 5 - 3 or 2% of the slots; 1.14% of fantasy points

Round 6 - 12 or 8% of the slots; 6.47% of fantasy points

Round 7 - 11 or 7.33% of the slots; 6.92% of fantasy points

UDFA (UnDraftedFreeAgent) - 23 or 15.33% of the slots; 15.48% of fantasy points

In examining these numbers, the first thing that jumps out is the more even distribution of slots by round. The TE position looks much more like the QB position than the RB or WR positions. Look at the high values for rounds 6, 7 and UDFAs. Only the 6th round QBs have a higher number of 6, 7 or UDFA slots taken than the TEs. Clearly this is a position that NFL teams have a hard time projecting with certainty.

Another interesting result is the large difference between the Round 1 slot percentage and the Round 1 fantasy point percentage. Every other round at other positions have a 0 to 2 percent difference. But here that difference is doubled, at 5.45%. This leads me to believe that the first round TEs that are successful enough to be in the Top 30 list are very successful within that list. Tight Ends like Jeremy Shockey, Tony Gonzalez and Todd Heap are high up on the list every year. Possibly NFL teams do greater due diligence when scouting a TE that they would consider taking with their first pick.

As expected based on the first comment, TEs have only a 60.66% First Day percentage, almost exactly the same rate as the QBs. This reinforces the idea that TE is a difficult position to scout.

Next, I did the same examination for the Top 12 TEs each year (60 TEs total), or those that should be starting in a 12 team, Start 1 TE fantasy league. Here are the results:

Round 1 - 24 slots or 40%

Round 2 - 12 slots or 20%

Round 3 - 5 slots or 8.33%

Round 4 - 5 slots or 8.33%

Round 5 - 0 slots

Round 6 - 3 slots or 5%

Round 7 - 4 slots or 6.67%

UDFA - 7 slots or 11.67%

Notice the higher percentage for Top 12 TEs in the first round, 40%, compared to the Top 30's 26%. This is another data point that shows that the first round TEs that make it into the Top 30 also likely make it into the Top 12.

IV: Fuzzy Math

As mentioned in section II, I also wanted to try to find out what percentage of total players drafted by position and round made up the Top 30 fantasy point list. In other words, if you consider making the Top 30 fantasy point list in a given season a "success", then what percentage of players drafted were "successful"?

There are a couple of issues with this part of the study. First of all, as mentioned in section II, there is not a way to completely map which drafts produced most/all of the Top 30 lists from 2002 to 2006. I decided to look back 10 years to 1997 and examine drafts from 1997 to 2006.

The second issue is that it may not be reasonable at the QB and TE position to assume that a player ranking 30th in fantasy points was successful. To combat this issue, I am going to use the fantasy starting lineup list (ie. Top 12 QBs, Top 24 RBs, etc) to define success. Of course, this shortchanges the good backup players that likely had to start in your lineup due to an injury or bye week, but I would rather underestimate than overestimate.

So let's look at the data. Here are the number of players drafted at each position in each round from 1997 to 2006:

QB:

Rd 1 - 27

Rd 2 - 8

Rd 3 - 14

Rd 4 - 13

Rd 5 - 14

Rd 6 - 21

Rd 7 - 27

RB:

Rd 1 - 30

Rd 2 - 21

Rd 3 - 25

Rd 4 - 34

Rd 5 - 17

Rd 6 - 15

Rd 7 - 29

WR:

Rd 1 - 41

Rd 2 - 41

Rd 3 - 37

Rd 4 - 42

Rd 5 - 40

Rd 6 - 50

Rd 7 - 64

TE:

Rd 1 - 14

Rd 2 - 18

Rd 3 - 20

Rd 4 - 14

Rd 5 - 26

Rd 6 - 21

Rd 7 - 31

Comment: Interesting that 2nd round quarterbacks are the most rare. This probably shows the pressure on NFL teams to pick a QB in the first round, even if they are graded out as a second rounder (coughcoughJPLosmancoughcough).

If we assume that each player drafted has 5 chances or 5 seasons to be in one of the 5 Top 12 lists from 2002 to 2006, then we can multiple all these numbers by 5 to get a total number of seasons that each position/draft round combination could be in the list. So if we then divide the number of players making the Top 12, 24, 36 or 12 list at each position for each round by these seasons, we MAY be able to infer the general success rate for being a fantasy football "starter".

QB: Top 12 TotSeasons PctSuccess

Rd 1 - 26 135 19.25%

Rd 2 - 9 40 22.50%

Rd 3 - 0 70 0.00%

Rd 4 - 4 65 6.15%

Rd 5 - 0 70 0.00%

Rd 6 - 10 105 9.52%

Rd 7 - 0 135 0.00%

Comments: One interesting thing we see here is that although there were almost 3 times as many Top 12 QBs drafted in round 1 as round 2, there was a higher percentage of QBs drafted in round 2 that made the Top 12 because fewer were drafted. Second rounders like Favre, Brees and Plummer really helped increase the success rate.
Another observation is that although we have seen the large number of successful 6th round QBs like Brady, Bulger and Hasselbeck, so many have been drafted in the 6th round that the success rate is not that impressive.
It's also interesting that the 3rd round, in which there should still be some talented QBs available, has a 0% success rate. Former young guns Brian Griese, Charlie Frye, Chris Simms and Josh McCown were all 3rd round picks.

RB: Top 24 TotSeasons PctSuccess

Rd 1 - 57 150 38.00%

Rd 2 - 24 105 22.85%

Rd 3 - 19 125 15.20%

Rd 4 - 11 170 6.47%

Rd 5 - 0 85 0.00%

Rd 6 - 2 75 2.67%

Rd 7 - 0 0 0.00%

Comments: Easy to see again that at running back, the first day picks are the most likely to have success and the rate drops off after that. What is really interesting to me is the number of RBs drafted (seasons) in rounds 2 and 4. For all the talk about being able to find RBs anywhere for some offenses, there sure is a large drop off from round 1 to round 2. That tells me that NFL teams are making sure to draft a RB with their 1st rounder if they want a player because they value the position highly. There is also a huge jump in the 4th round, which is when teams have re-adjusted their draft boards for the 2nd day. Combine these two things with the fact that there are less lower round or UDFA sleepers on the Top 24 list and I think it's clear that running back is a position highly valued and highly scouted by NFL teams. Maybe that is going to change as more teams go to a RBBC approach, but that hasn't happened over the last 5 years.

Statistical oddity: There are no Top 12/24 performers from the 5th and 7th rounds for both QBs and RBs. I guess those numbers are unlucky.

WR: Top 36 TotSeasons PctSuccess

Rd 1 - 77 205 37.56%

Rd 2 - 39 205 19.02%

Rd 3 - 27 185 14.59%

Rd 4 - 10 210 4.76%

Rd 5 - 3 200 1.50%

Rd 6 - 2 250 0.80%

Rd 7 - 8 320 2.50%

Comments: Similar to the RB position, Wide receivers that are drafted earlier in the draft seem to be more successful, with the a pretty severe fall off after the first day of the draft. What is interesting to me is the large number of WRs drafted in the 6th and 7th rounds. Those rounds have the most WRs drafted for the last 10 years. Apparently all you need is a pulse to get drafted at the end of the second day if you are a WR. As excited as the Saints were about Colston last year, he has several high performance years ahead to match 7th rounders Donald Driver (4 Top 36 finishes) and T.J. Houshmandzadeh (3). "Championship" indeed.

TE: Top 12 TotSeasons PctSuccess

Rd 1 - 24 70 34.28%

Rd 2 - 12 90 13.33%

Rd 3 - 5 100 5.00%

Rd 4 - 5 70 7.14%

Rd 5 - 0 130 0.00%

Rd 6 - 3 105 2.85%

Rd 7 - 4 155 2.58%

Comments: This data looks similar to the QB data. Some good success in the first round, but random success after that. The most TEs drafted in the last 10 years have been in the 6th and 7th rounds.

Statistical oddity Part 2: At 3 of our 4 positions (QB, RB and TE) there have been no players drafted in the 5th round that finished the season as a fantasy starter. That is very strange. I guess the 5th round is for slackers. And hats off to the man responsible for the 3 fantasy starter seasons from the 5th round at wide receiver, Joe Horn. You're no slacker, Joe.

V. What does it all mean, Basil?

So what conclusions can we reach with this data? Well, I think it's clear that there is a lot of evidence that the earlier a player is drafted, the greater chance of fantasy football success he will have. The QB and TE positions show that there is some uncertainty about evaluating the position, as there are some top performers later in the draft or who are undrafted.

The RB and WR positions are remarkable similar in their success rates. They average approximately a 38% success rate for 1st rounders, a 20% success rate for 2nd rounders and a 15% success rate for 3rd rounders.

The lesson for fantasy football owners should probably be to heed the advice given to you by the NFL teams in form of the draft order. Sure there are players like Ryan Leaf and Tom Brady who don't perform to the level of their draft round. And maybe you are a very good judge of college talent and just know that Jerome Harrison should have been a 3rd rounder, not a 5th rounder. But do yourself a favor and draft the running backs taken in rounds 3 and 4 first. You will likely have more success that way.

Detailed Analysis of 2005 NFL Injury Reports for Fantasy Purposes

2008-01-27T21:19:00.000-05:00

This article was written about the 2005 NFL season and selected as a freelance article by Footballguys.com, but never published on the site.

Detailed Analysis of NFL Injury Reports for Fantasy Purposes

By ConstruxBoy

I. Introduction

One of the most difficult and frustrating experiences in playing fantasy football is checking the weekly injury reports and trying to decide whether or not your player will play in their game. Many a fantasy owner's ulcer has been started by coaches who do as much as possible to cloud their player's true injury status. I think it would help owners a great deal in their weekly start/bench decisions to examine how often injured players actually play and how much they contribute when they do play injured compared to their average fantasy output. I am going to attempt to shed some light on the injury report and how it can be used to its greatest utility by a fantasy football owner.

II. Parameters of the Study

Each week of the 2005 NFL season, I used the Footballguys Friday afternoon Injury Report in the Subscriber section. I entered every injury for every team to a starter at a normal fantasy football position (QB, RB, WR, TE, PK; apologies to IDP owners) for the three levels of injury status: Doubtful, Questionable and Probable. I skipped the Out status because there is never a reason to start a player who is declared out.

For each injured player, I recorded several different variables in the Raw Data section of the spreadsheet:

1) I recorded the week of the injury to see if there were any trends toward more injuries later in the year or more players playing through injuries later in the year.

2) I recorded whether the game is a non-Sunday afternoon game (Thursday, Friday, Saturday, Sunday Night, Monday Night) to see if less players play in earlier games or more play in later games.

3) I recorded the player's team to see if there were trends by team, such as Jeff Fisher not understanding the definition of the word "questionable".

4) I recorded the player's opponent and whether it was a division game or not to see if there were patterns of players playing more often in division games.

5) I recorded the player's position to see if there were any trends by position.

6) I checked the current depth chart to ensure that I was talking about "starters" when reporting and comparing this information.

7) I recorded the player's injury status to see if Probable players really do play more than Questionable players.

8) I recorded the actual injured body part to see there were any trends toward certain injuries, like the knee, making a player less likely to play.

9) After the game, I recorded whether the player played and how many fantasy points they scored on a 6/TD, .1 Rush/Rec Yds, .05 Pass Yds scale.

10) After the season, I recorded each players average fantasy points per game for the year and compared it to their points in the games they played hurt.

If it sounds like a lot of work, it was.

There are a couple of things worth clarifying with these parameters:

1) It is sometimes the case that a player may have their injury status change between Friday afternoon and game time, or that more information on a given player's chance of playing may come out on Sunday morning from an insider like Chris Mortensen. I did not account for these changes in this study.

2) To consider a player playing, I required them to score at least .1 fantasy points in the game. Obviously a player could see the field but never get a carry or catch a pass. However, tracking that through a play-by-play examination of each game is too difficult and the idea for our fantasy football purposes is that they contribute points.

3) Starters - I believe that the only accurate way to measure this information is to just track "starters", as backups are listed on injury reports but often don't play regardless of their injury. I used the Depth Chart to identify starters and keep in mind that it changes from week to week due to other injuries. Also keep in mind that I defined most WR3s, like Kevin Curtis, and RB2s, like TJ Duckett, as starters if they regularly receive work each game. So most teams will have 1 QB, 2 RB, 3 WR, 1 TE and 1 PK as starters. (One small exception was the week 15 Patriots, who listed their top 4 RBs as Questionable. Good luck figuring out the NE RB that week). This is obviously a judgment call on my part and will influence the results to a small degree.

III. Results

Overall

Now I would like to review some of the results on the study. The full data set and summarized results are available in the spreadsheet here. Overall, I tracked 562 injuries in which 374, or 66.5% (Play rate), end up with the player scoring fantasy points. The percentages based on the injury status were 5.6% playing for Doubtful, 50.4% playing for Questionable, and 89.7% playing for Probable. At a very high level, the lesson seems clear: Doubtful players seldom play and Probable players usually play. Baring any last minute game day updates, you should almost never start a Doubtful player and you should almost always start a Probable player. It's the Questionable players, appropriately, that need much further analysis.

Division Games

As you might suspect, the percentages of players playing in Divisional games is higher across the board, although not quite as high as I would have thought. The percentages are 7.1%, 56.9% and 90.6% for Doubtful, Questionable and Probable respectively. Although this wasn't originally part of the study, I calculated 562 fantasy position injuries for 256 total NFL games, or 2.2 injuries/game. I had suspected that the Divisional game rate would be higher in an attempt by some coaches at gamesmanship, listing more players as injured to throw off their most bitter opponents. This turned out to be true; 222 divisional fantasy position injuries in 96 divisional games for a 2.3 injuries/game rate. But the increase was not as high as I thought it would be. However, combining the higher injury rate for divisional games with the higher actual play rate for divisional games makes me believe that some gamesmanship is going on in these important contests. This is something to keep in mind when your player suddenly shows up on the injury list for an important divisional game.

Day of Game

Another part of the study was to examine whether players are more likely to play in later games (Sunday Night, Monday Night) or less likely to play in earlier games. With so many games on Sunday afternoons, the sample size for the other times are probably too small to be relevant. But the data does show an interesting result: Almost all the other game times (Thursday, Saturday, Sunday Night and Monday Night) show a higher overall percentage of players playing through the injury than Sunday afternoon, at play rates of 85.7%, 70.8%, 68.8% and 69.4% respectively to Sunday afternoon's 65.4%. Again the small sample size does skew the data some, but I would have expected a decrease for games before Sunday afternoon.

Week of the Season

Another way to examine the data is to see whether the players were more likely to play later in the season, when their team has completed its Bye week and might be fighting for a playoff spot. If you look at the week-by-week data in graphical form, you can't see any real pattern.

The only important trend that I see here is that week 17 shows a decrease in the percentage of Probable players who play, as well as a lower percentage in the Questionable category. This confirms the common knowledge that some players are rested in week 17 if their team has secured their playoff position.

What is a little bit more interesting is a graph of the total number of fantasy position injuries throughout the year.

Here you can see that the number of injuries grows throughout the season with a small dip down after the Bye weeks have been completed and then another upward trend leading into week 17. Players get more banged up as the year goes on, but this seems to be some further evidence that coaches increase their injury report as week 16 and 17 approach so that they can rest the starters.

Position

The positional analysis shows some interesting trends. First of all, the total percentage played results are as follows:

QB	80.5%
RB	66.7%
WR	63.0%
TE	55.7%
PK	74.3%

I was not surprised to see QB and PK at the top of the list, but a little bit surprised to see TE at the bottom. Given that the TE is generally the biggest/strongest among the RB/WR/TE players, I expected them to play through injury more than most. When further examining the detail, I see that they have the second highest percentage of probable players who play, 92.6%, second only to the QBs' 98.1%. But they have an abysmal 41.3% of questionable players who play, almost 10% worse than the second lowest percentage, the RBs' 50.6%. The only explanation I can come up with for this anomaly, other than the fact that they really don't play through injury at the same rate as other positions, is that they are more replaceable in the lineup than the other positions. Maybe if the TE hurt was the blocking specialist, a team would change their game plan to adjust to using quicker patterns or more throws to the backup TE and not rush the starting TE back into the lineup. Looking at the raw data, many of the Questionable TEs who didn't play do appear to have been replaceable. Here is a partial list:

Team	Questionable Starter	Possible backup
CLE	Aaron Shea	Steve Heiden
GB	Bubba Franks	David Martin
NE	Daniel Graham	Ben Watson
NO	Ernie Conwell	Zach Hilton
TEN	Erron Kinney	Ben Troupe
TEN	Ben Troupe	Erron Kinney

So maybe in weeks 9 and 11 when Ben Troupe was questionable with an ankle injury, Jeff Fisher just changed the game plan to match Kinney's strengths. And when Kinney was questionable with a knee injury in weeks 15 and 17, Fisher worked Troupe's strengths into the game plan.

A lesson for fantasy football owners here may be that if your TE is questionable and there are other decent TEs on his team, it may be more likely that he does not play.

The position with the lowest percentage of Probable players to play was WR, with an 83.1% play rate. This is a lower percentage of probable players playing than the Overall Percentage, Division Percentage and most Week percentages. I am unsure of whether this means that the WRs cannot play as easily with a "probable" level injury as other positions or whether they just can't play through minor injury due to pain. I can see a minor injury affecting a WRs speed and making him less likely to be effective, although I can also see some of the "Prima Donna" WRs not wanting to play through the pain. For what it's worth, Terrell Owens was listed as probable 4 weeks and Randy Moss was listed as Probable 7 weeks and both played in every week.

Team

Some very interesting results are seen at the team level. Although there has been a lot of turnover in head coaches this season, there is still some value in seeing how coaches use the injury list. A common conclusion is that Jeff Fisher of the Titans never uses Probable or Doubtful but lists all his players as Questionable to make if more difficult for the other team to prepare.

To start the examination of the team-by-team results, I'll list the teams with at least 14 reported injuries and a less than 60% overall play rate:

Team	# of Injuries Reported	Play Rate
NO	14	28.6%
SF	24	37.5%
CLE	14	50%
GB	16	50%
DET	21	52.4%
JAC	19	52.6%
HOU	20	55%
CHI	14	57.1%
STL	24	58.3%
WAS	22	59.1%

Assuming that on average there was a 66.5% overall play rate, some of these percentages may help you identify teams that don't/won't let their players play through pain or risk further injuries. NO and SF are especially low. Jim Haslett has moved on from coaching the Saints, but Mike Nolan at SF does not appear to like his injured players to play.

Two interesting notes outside of this range: Both BUF and DAL only listed 3 fantasy position starter injuries all year, with the 1 Probable Bills player playing and none of the 3 Questionable Cowboys players playing. The teams may have just been lucky or possibly there coaches do not like to list players on the injury report. That is easy to see from Bill Parcells of the Cowboys based on his statements over the years, from Terry Glenn of a decade ago to Julius Jones recently.

At the other end of the spectrum are teams that have a high play rate. Here is a list of the teams with at least 14 players listed and more than an 80% overall play rate:

Team	# of Injuries Reported	Play Rate
DEN	14	92.9%
PHI	25	92.7%
IND	16	87.5%
CIN	22	86.4%
CAR	14	85.7%

One interesting thing that I see from this data is that those are all some pretty good teams who have had a lot of recent success and have solid coaches. It seems that perhaps these coaches motivate their players to play through injury and that leads to more success on the field. On the other hand, maybe these good coaches like to list lots of players on the injury report even though they know that the player will be able to play. Either way, it's something to keep in mind for 2006.

Two interesting notes outside this range: The Giants were 100% on there 7 players listed and SD was 10 for 11 on their players listed. Both Tom Coughlin and Marty Schottenheimer seem like old school coaches that do not like to list many players on the injury report, but expect them to play when listed.

A more important view of the team injury results is checking each team's Questionable players to see if there are any trends that may help you when choosing who to start for your fantasy team. Here is a list of the Number and Play Rate for Questionable players on all teams:

Team	# of Questionable Players	Play Rate	Team	# of Questionable Players	Play Rate
ARI	2	50%	MIA	7	71.4%
ATL	6	33.3%	MIN	2	100%
BAL	4	25%	NE	31	51.6%
BUF	2	0%	NO	14	28.6%
CAR	4	75%	NYG	3	100%
CHI	7	42.9%	NYJ	3	33.3%
CIN	3	66.7%	OAK	4	100%
CLE	8	25%	PHI	5	80%
DAL	3	0%	PIT	8	37.5%
DEN	2	50%	SD	10	90%
DET	15	40%	SF	12	33.3%
GB	10	30%	SEA	2	50%
HOU	11	27.3%	STL	6	50%
IND	3	33.3%	TB	9	66.7%
JAC	6	16.7%	TEN	44	72.7%
KC	3	66.7%	WAS	5	0%

Some interesting notes from this list:

1) Jeff Fisher does win the award for most Questionable players with 44 and a 72.7% play rate shows that he is mostly just playing mind games. Bill Belichick comes in 2nd place with 31 Questionable players, but his players only play at a reasonable 51.6% rate.

2) Those 44 Questionable players are the only players on the TEN list. No fantasy position starter was ever listed as Doubtful or Probable. NO had the same oddity with a much lower play rate of 28.4%.

3) Although there were not many players listed, MIN, NYG and OAK get a thumb up for playing all of their Questionable players.

It's hard to draw a lot of conclusions from this data due to small sample sizes, but if nothing else it's instructive to see that most of the teams that list lots of players as Questionable, like NE, NO, DET, SF, HOU and GB, do not play those players at a very high rate. Other than the Patriots, those teams did not fare very well last season.

Injury

Finally, let's look at the Injury data to see if we can tell that certain injuries are less likely to result in the player playing in the game. First let's look at all the injuries with 15 or more instances and their overall play rate:

Injury	Instances	Play Rate
Knee	145	56.6%
Ankle	70	60%
Shoulder	56	75%
Calf	35	80%
Foot	34	79.4%
Hamstring	31	67.7%
Back	22	72.7%
Groin	17	94.1%
Toe	17	41.2%
Quad	15	40%

Two things that stand out from these numbers are the huge number of Knee injuries, with a poor play rate of 56.6%, and the very high play rate for Groin injuries. I would have expected Groin injuries to keep players out of the lineup more often.

Looking in more detail at the important Questionable listings, here are some of the good play rates for questionable injuries:

Injury	Instances	Play Rate
Foot	18	77.8%
Chest	8	75%
Back	14	71.4%
Elbow	8	62.5%

It seems like both Foot and Back injuries are pretty good bets to play, along with the Groin, based on the last two tables.

Now let's look at the poor play rates for Questionable listings:

Injury	Instances	Play Rate
Quad	7	14.3%
Knee	72	37.5%
Concussion	7	42.9%
Ankle	42	45.2%
Hamstring	13	46.2%
Shoulder	21	47.6%

So besides the already mentioned poor play rate on Knee injuries, it also looks like Quad and Concussion injuries are difficult to play through. Keep in mind, of course, that 7 instances of each of those Questionable injuries is a very small sample size.

Average points per Game

The secondary part of my research was to examine the number of fantasy points per game scored by players playing injured, compare that value to those player's average fantasy points for 2005, and see if there was a noticeable difference. It seems like the overall point per game average should be lower overall for injured players and likely lower as the injury status goes from Probable to Questionable to Doubtful.

However, looking at the actual data, this is only partially true. Here are the overall numbers:

Injury Status	Avg Pts when Play	Avg Pts per Game for Year	Difference
Doubtful	4.8	9.5	-4.8
Questionable	9.2	9.4	-0.2
Probable	11.1	11.3	-0.2
TOTAL	10.4	10.6	-0.2

The Doubtful values are based on only two players playing through injury (Randy Moss in week 7 and Fred Taylor in week 11), so the sample size is too small to draw a true conclusion. But look at the difference for Questionable, Probable and Total: Players who played only scored .2 points per game less than their yearly per game average. The value is so negligible that you could almost say that if a player plays through the injury, on average, they should score the same amount that they would any other week. I had expected at least a point or two difference for the Probable status and a couple of times that decrease for the Questionable status. Let's dig into more detail to see if we can identify situations where the players do score consistently less than their yearly average.

At the day of the week level, we still don't see any real statistical trend, although the game time with the largest increase in points scored versus average points scored was on Monday Night, the latest game of the week. But it was a mere .7 points per game above the average.

At the week of the season level, we do see an interesting trend if we graph the data:

Other than a small up tick in week 15, 4 of the last 5 weeks have negative values (with week 17 having the lowest value of the year), meaning the players who played hurt scored less than their average for the year. We could possibly conclude that as the year wears on, players who are playing through injury find it harder to perform at their highest level. This may be a consideration when starting an injured player near the end of the year.

At the position level there are no obvious insights. The only interesting piece of data was the fact that although TEs have the lowest play pct among positions at 55.7%, they have the highest average point difference with a 1.4 points a game increase from their yearly per game average when they do play. So you may have reason to be worried that your injured TE will actually play, but you probably shouldn't be worried that his fantasy points will drop off because of the injury.

At the team level there are a couple of small insights. The lowest average difference among teams with at least 5 players playing through injury are the -3.3 pts per game of SD and the -2.3 pts per game of MIA. My thought on this result is that maybe the coaches on these teams push their players to play through injury, even though the player really is too hurt to fully contribute. Although he is new to the NFL, Nick Saban seems like an old school coach that would press his players into action, as does the veteran coach Marty Schottenheimer. At the opposite end of the spectrum are the 3.4 pts per game increase of ATL and the 2.8 pts per game increase of DET. Possibly on these teams the players are only playing when they are truly at 100% and able to contribute as though they are not injured.

Finally, at the injury level there are some trends worth discussing. First of all, let's examine the overall average point difference for the injuries with 10 or more starts:

Injury	Starts	Overall Avg Pts Diff
Knee	82	-0.3
Shoulder	42	1.0
Ankle	42	-0.5
Calf	28	0.5
Foot	27	-0.7
Hamstring	21	-0.1
Back	16	1.7
Groin	16	-1.2

One interesting note here is the Back injury, which we showed had a pretty high play rate, also has the highest average points difference here. You should feel more comfortable starting a player with a back injury than most other injuries. Conversely, the groin injury, which had a very high 94% play rate, has the lowest average points difference. So it seems like your player with a groin injury may gut it out, but not be very productive.

At the Questionable level, we also see two interesting trends. The average points difference for Questionable Knee injuries is -1.1 in 27 starts. The difference for Questionable Shoulder injuries is 1.7 in 10 starts. This reinforces the idea that knee injuries are difficult to play through and that shoulder injuries can be managed without losing too much production.

IV. Conclusions

One of my initial ideas for fantasy football use of this data was to string together the various play rates of a given injured player in 2006 to try to find out what the percentage chance was that they would play based on 2005 data. So for example, if I was trying to figure out whether to start the OAK RB in Week 6 in a non-Divisional game on a Monday night with a Questionable Back injury, I could average out the following:

Detail	Play rate
Questionable - Overall	50.4%
Questionable - MN	57.1%
Questionable - Week 6	43.8%
Questionable - RB	50.6%
Questionable - OAK	100%
Questionable - Back	71.4%
Total Average	62.2%

But now I think that a lot of the play rates included as equal weights in the average are going to be based on such small sample sizes that the result is not statistically relevant. However, it may still help someone to look through these play rates and make a generalization about whether or not their player will play. Looking at the above data, you could see the higher than overall rate for a Monday Night game and for a Back injury and maybe decide to start your player. The 100% play rate for Oakland looks nice, but Art Shell may do things totally different than Norv Turner did last year.

Although this study did not provide very many hard and fast rules that can be used to determine whether or not to start an injured player in your league, I hope that it helped point out some general trends in injury play rates and performance in those games. Some of the theories that I had regarding injury reports, such as certain coaches using the injury report as a "smokescreen" and Divisional games being an incentive to play through injury, were confirmed with this study. Others, such as games later in the week being easier to play in and injured players almost always scoring below their average, were not proven using this data set. However, hopefully you have gained enough insight into the trends in the injury report and it's use by the different teams to make your injury start/bench decisions easier and to help dominate your league.

A Detailed Examination of Handcuffing Running Backs

2008-01-09T19:43:00.000-05:00

This article was written in 2005 concerning the 2004 NFL season. It was selected as a Freelance Article at FootballGuys.com, but never published on the site.

A Detailed Examination of Handcuffing Running Backs

John Powderly

I. Introduction

Two years ago, I wrote a freelance article for Footballguys.com titled “The Case Against Handcuffing”. You can view the article here if you are a member. I believe I made a pretty good case against handcuffing your top RB, however I wanted to do a more detailed examination of the results of handcuffing.

II. Parameters of the Study

I decided to examine the draft and end results of 15 different Phenom leagues, which are a group of great leagues run by Unlucky. I have participated in these leagues, and their precursor, the Greek Leagues, for several years and I feel confident that these leagues are a good sampling of smart, informed fantasy football owners. For each of these leagues, I went through the entire draft and noted where the same owner drafted two Running Backs from the same team. I am defining this as a Running Back Handcuff. I also noted the next three Running Backs drafted after the second same team RB. I am defining these as the Missed Opportunity Running Backs. My theory is that by taking a handcuff to your main RB you can easily miss out on better value at the RB position.

I will compare how these players do throughout the year and, week by week, examine situations is which having the handcuff RB worked out to the owner’s advantage. If a handcuff RB starts for a team during a particular week (a handcuff RB coming into the game to spell the starter does not qualify as you would not have started the handcuff RB in your league), I will compare that handcuff’s fantasy points for the game to the three Missed Opportunity RBs points for that week. I will also do a season comparison.

I hope to be able to examine the frequency of handcuffing RBs in “shark-infested” leagues, the frequency of those handcuff RBs working out, and the missed opportunities that handcuffing entails.

The scoring I will use to determine fantasy points will be the standard 1 pt/10 Rush/Rec yards and 6 pts for a TD.

NOTE - These 15 leagues represent several different scoring and roster size options to try to give a broad view of handcuffing. There are no Dynasty leagues.

III. 2004 Pre-season Running back Depth Chart Discussion

One of my main problems with handcuffing is that you are often unsure whom the handcuff for a star RB is when you draft. This is especially true if you draft early in the pre-season. I am going to examine each team’s Running Back depth chart at the end of pre-season.

Handcuff Clarity: 3 = Certain, 2 = Some Doubt, 1 = No Clue

NOTE: Only 3 teams in the NFL had no instances of handcuffing RBs in the 15 leagues examined: ATL, PHI, and SD. I will not discuss them.

Baltimore: Starter- Jamal Lewis; Backups – Musa Smith, Chester Taylor

Notes: Lewis is clearly a star RB and his impending Federal Trial made getting his handcuff a priority for some owners. The issue is that people were split on whether the main backup would be Smith or Taylor. The consensus seemed to be that Taylor was more of a 3^rd down back and that Smith would start in Lewis’ place.

Handcuff Clarity = 2

Buffalo: Starter – Travis Henry; Backup – Willis McGahee

Notes: Henry is pretty clearly the starter, although some folks thought McGahee might beat him out by season’s end.

Handcuff Clarity = 3

Cincinnati: Starter – Rudi Johnson; Backups – Chris Perry, Kenny Watson

Notes: Johnson is the starter, but is a bit of an unknown. Perry was likely drafted to be the backup and future, but good things were heard about Watson in camp. Still, a pretty clear picture of who was the handcuff.

Handcuff Clarity = 3

Cleveland: Starter = Lee Suggs; Backup – William Green

Notes: This was a murky situation, with lots of folks declaring that Suggs was much better than Green and others saying that Green had turned his life around and would win this job early in the season.

Handcuff Clarity = 2

Denver: Starter – Quentin Griffin; Backups – Tatum Bell, Garrison Hearst, Mike Anderson

Notes: This situation was pure insanity in the pre-season with everyone seeming to have a different opinion on how things would work out. By the end of preseason it was clear that Griffin was the starter, but the backup was really unknown.

Handcuff Clarity = 1

Houston: Starter – Domanick Davis; Backups – Tony Hollings, Jonathan Wells

Notes: This situation was pretty straightforward with most folks assuming that the greater "talent" in Hollings would be the main backup. Wells was being phased out.

Handcuff Clarity= 2

Indianapolis: Starter - Edgerrin James; Backup - Dominic Rhodes

Notes: This was a very clear situation with Rhodes thought of as a reasonable backup for James. Some folks were still wondering if Edge was back from his knee injury, which made Rhodes more necessary than usual.

Handcuff Clarity = 3

Jacksonville: Starter - Fred Taylor; Backups - LaBrandon Toefield, Greg Jones

Notes: When you're nickname is Fragile Fred (correct pronunciation: FRA-JEE-LAY), you're going to have folks picking up your backup. Toefield had shown more of his stuff, but was also injury prone. Jones was a rookie from FSU with an awesome smack down of a Tar Heel defender on his resume.

Handcuff Clarity = 2

Kansas City: Starter - Priest Holmes; Backups - Derrick Blaylock, Larry Johnson

Notes: Holmes had been injured, but released a workout video to show the world he was fine. Due to the outstanding OL of KC, most folks wanted his backup. Blaylock was an underrated backup who usually produced when called upon and Johnson was a 2nd year guy who needed to get his diaper off according to the coach.

Handcuff Clarity = 2

Miami: Starter - If you drafted early, Ricky Williams; Otherwise, Travis Minor; Backups - Sammy Morris, Leonard Henry, Jim Kiick

Notes: Ricky did the fantasy world no favors by retiring right as camp opened. Minor was a shaky bet to be the starter based on his prior experience in Miami. But Sammy Morris and Leonard Henry loomed.

Handcuff Clarity = 1

New England: Starter - Corey Dillon; Backups - Kevin Faulk, Cedric Cobbs

Notes: Corey Dillon finally escaped Cincinnati with something to prove. Faulk was the 3rd down back who wasn't big enough to take over for Dillon and Cobbs was a bruising 2nd day draft pick.

Handcuff Clarity = 2

New York Jets: Starter - Curtis Martin; Backup - Lamont Jordan

Notes: Nice, clear-cut, no questions asked situation.

Handcuff Clarity= 3

Oakland: Starter - Tyrone Wheatley; Backups - Justin Fargas, Amos Zereoue, Zack Crockett

Notes: The backup situation was a bit of a mystery. Fargas was a 3rd round pick with great measureables, Zereoue was coming over after a decent stint as the Steelers backup and Crockett was the TD vulture.

Handcuff Clarity = 1

Pittsburgh: Starter - Duce Staley; Backup - Jerome Bettis

Notes: A pretty clear situation with Bettis probably getting the short yardage work.

Handcuff Clarity = 3

Tennessee: Starter - Chris Brown; Backup - Antowain Smith

Notes: Another pretty clear-cut situation with some people worried about Brown's injury history going back to his college days at Colorado.

Handcuff Clarity = 3

Arizona: Starter - Emmitt Smith; Backups - Marcel Shipp, Damien Anderson

Notes: Emmitt was the starter, but his age made most people think a backup may be valuable. Shipp was the best bet, but some folks liked the sleeper, Anderson.

Handcuff Clarity = 2

Carolina: Starter - Stephen Davis; Backup - Deshaun Foster

Notes: Davis was the starter, but was getting up there in years and Foster had looked great in preseason a year before.

Handcuff Clarity = 3

Chicago: Starter - Thomas Jones; Backup - Anthony Thomas

Notes: Clear situation in which there was a worry that Thomas would steal goal line carries.

Handcuff Clarity = 3

Dallas: Starter/Backups - Eddie George, Julius Jones

Notes: This was an interesting situation in which there were different opinions on who the starter would be from week to week. Many thought that Parcells would go with a veteran, as he has done in the past, over a rookie.

Handcuff Clarity = 2

Detroit: Starter - Kevin Jones; Backup - Artose Pinner

Notes: Much to the dismay of FBG Message Board User jwvdcw, Jones was named the starter over Pinner.

Handcuff Clarity = 3

Green Bay: Starter - Ahman Green; Backups - Najeh Davenport, Tony Fisher

Notes: Davenport was the main backup while Fisher was more of the 3rd down back.

Handcuff Clarity = 3

Minnesota: Starter - Michael Bennett; Backups - Moe Williams, Onterrio Smith, Mewelde Moore

Notes: Minnesota, or as I like to call them "Denver Lite", had mass confusion in their backfield going into 2004. The pre-Whizzinator Smith was likely to be the main backup, although he had some doubters. Williams was always a good bet to catch some passes and vulture some short TDs. And Moore was a rookie who many thought should have been drafted higher. There were actually 5 different combinations of these 4 RBs taken in the 15 leagues I examined.

Handcuff Clarity = 1

New Orleans: Starter - Deuce McAllister; Backups - Ki-Jana Carter, Aaron Stecker

Notes: As a number one overall pick, Carter was thought of as a better bet to backup McAllister. But Stecker was there as well and had showed some good skills as a backup with Tampa Bay.

Handcuff Clarity = 2

New York Giants: Starter - Tiki Barber; Backup - Ron Dayne

Notes: New HC Tom Coughlin was going to turn Dayne into a goal line back by placing donuts in the end zone.

Handcuff Clarity = 3

Saint Louis: Starter - Marshall Faulk; Backup - Steven Jackson

Notes: Marshall was at one time a stud RB. But Jackson was drafted to be the future Ram stud RB after Trung Candidate and Lamar Gordon didn't work out.

Handcuff Clarity = 3

San Francisco: Starter - Kevan Barlow; Backup - Jamal Robertson

Notes: Barlow was pretty hyped up going into the season as a breakout candidate. Robertson was thought to be the primary backup, although Terry Jackson and Maurice Hicks were lurking.

Handcuff Clarity = 2

Seattle: Starter - Shaun Alexander; Backup - Maurice Morris

Notes: Quite a clear cut situation

Handcuff Clarity = 3

Tampa Bay: Starter - Charlie Garner; Backups - Michael Pittman, Mike Alstott

Notes: Pittman was suspended for the first 3 games of the season for going demolition derby on his wife. Alstott always is good for some TD vultures. There was some thought that Garner would keep the starting job even after Pittman got back. Oops.

Handcuff Clarity = 2

Washington: Starter - Clinton Portis; Backups - Ladell Betts, Sultan McCullough

Notes: Betts was still trying to fulfill his potential as an NFL RB while McCullough was an undrafted "could have been" out of USC.

Handcuff Clarity = 2

IV. Statistical Breakdown of Handcuff Results

If you examine the data set here, you will see three different sets of data, which I will discuss one by one.

Draft-Starters (First tab)

This data set shows both the draft results by team in all 15 leagues, including the 3 MORBs, as well as the weekly results (through week 16) of every team and who the MRB was in each game.

Looking at the top section, you see the team, their RB1 and any handcuffs listed. Each league, numbered from 1 to 15, will show the draft positions of the RB1 and handcuff RB, as well as the next 3 RBs drafted in that league. These are the MORBs.

For example, in League 1 a team drafted Lee Suggs at 4.12 and William Green at 8.12. The 3 MORBs were Onterrio Smith, Willis McGahee and Tatum Bell.

You'll also notice the 3 teams with no handcuff drafted in any of the 15 leagues highlighted in red.

One interesting note is that the 15 leagues are, generally, numbered in order from earliest draft start date to latest draft start date. You'll notice that leagues 1 and 2 have the most handcuffs, with 14 and 13 respectively. The last couple of leagues have only 4 or 5. This makes sense because the closer to the regular season a draft occurs; the more information about possible injury concerns or handcuff prospects is available. This should lead to fewer handcuff needs as the situations sort themselves out.

In the bottom section is the list of each teams MRB throughout the first 16 weeks of the season. An entry highlighted by light blue is a week in which a drafted handcuff was the MRB. These light blue entries also list the number of fantasy points that handcuff scored as the MRB. An entry highlighted by yellow is a week in which an undrafted backup RB is the MRB. The six teams highlighted in green never had a MRB other than the pre-season starter.

For example, Carolina had Stephen Davis start the first game, and then DeShaun Foster started 3 games before getting hurt. Davis came back for a week before Brad Hoover, Joey Harris and Nick Goings finished off the season. Foster was the only handcuff drafted.

The next two rows show the number of times each week that both a handcuff and non-handcuff RB were the MRB. As we would expect, both numbers grow as the season went on due to injury or ineffectiveness by the starters. The highest week of handcuff MRBs was week 12 with ten handcuffs listed as the MRB.

After that is some league wide data comparisons concerning the use of handcuffs. There were 29 teams that had handcuffs drafted in my study. Each team played 15 games (not including week 17). Therefore there were 435 games total with a MRB. A drafted handcuff was the MRB in 98 of those games, or 22.5% of them. An undrafted handcuff was the MRB in 32 of those games, or 7.4%. So total handcuff MRB games were 130 out of a possible 435 or just under 30% of the games played. So in conclusion, last year in these 15 leagues, a handcuff RB was the main RB 30% of the time. This number is a little bit higher than I would have thought.

Finally there is a list of the 6 teams that never had a handcuff RB perform as the MRB in a game. So 6 teams out of 29 examined, or 20.7% of the teams in the league, did not use a backup as the Main RB at all during the season.

Starter Comparisons (Second tab)

The next data set is a list of all the times a handcuff RB was the MRB in every league, with their points and the names and points of the 3 MORBs listed as well. The average points are found for these 3 MORB, as well as the difference between the MRB and the average MORB. The Best Point total from a MORB and its difference from the MRB are also calculated. This is to show the possibilities had the owner drafted the highest scoring MORB that week.

For example, in week 1 William Green was a MRB and scored 9.2 points. He was drafted in leagues 1, 2, 3 and 9. In league 1, the 3 MORBs were Onterrio Smith, Willis McGahee and Tatum Bell. Their average number of points in week 1 was 8.4. So the William Green owner scored .8 more points that week than if he had taken the 3 MORBs. However, had he taken the best scoring MORB that week, Smith with 19.9 points, he would have gained the 10.7 more points that week that Smith scored.

The Total Averages listed at the bottom of the page were surprising to me. The average number of points a handcuff scored last year as the MRB was 13.3! This is much higher than the average MORB point total of 4.1 points, and also higher than the Best Point total of 8.5 points. So the only conclusion I can reach, based on these 15 leagues in 2004, is that when the handcuff RB was the main RB, he outscored any of the Missed Opportunity RBs you could have drafted. Of course, this does not account for the "Carrying Cost" of having those handcuff RBs on your roster when they are not the MRB.

Team Totals (Third tab)

This data set will help us examine the result of drafting a handcuff RB whom is never a MRB at anytime during the year. The data set lists all the drafted handcuffs and shows how many games, if any, they were the MRB during the season. If they were never the MRB, I look at their yearly point total compared to the yearly point totals of the 3 MORBs.

For example, Musa Smith was drafted in league 8 and never was a MRB. He had a total of 7.9 points for the year. The owner could have drafted Jerome Bettis, DeShaun Foster or Onterrio Smith instead, who averaged 113.2 points on the year. So in this league, the Musa Smith owner lost out on 105.3 points or an average of 6.2 points per game. That is quite a lot of points to be giving away each week. Obviously, this study does not examine whether that owner actually kept Musa Smith on his roster all season. But for our purposes, this shows us a significant loss of point production in this case.

As a side note, although I did not examine every owner's roster moves and record, I did check on the owner of LaBrandon Toefield in league 12. This owner picked up Toefield, who was never a MRB, to back up Fred Taylor over Michael Pittman, Onterrio Smith and Richie Anderson. He lost out on an average of 4.8 points per game. Checking with the league results, he kept Toefield on his roster the entire season and finished last in his division with a 3-11 record. There are many other factors that produced that dismal record, but certainly passing on those 3 RBs and keeping Toefield on the roster the entire season did not help.

At the bottom of the dataset are totals for these handcuff RBs whom never were a MRB. Their average yearly point total was 22.7 points, compared to the MORB point total of 57.3 points. This is a difference of 34.6 points during the season or about 2 points a game. So owners that drafted a handcuff RB whom was never the MRB lost about 2 points a game. While I don't consider 2 points a game to be significant, it likely did cost some owners a game or two.

However, this loss of 2 points a game for drafting a handcuff RB whom is never a MRB pales in comparison to the 9.2 points on average that were gained in the games when a handcuff RB was the MRB. To fully compare these two situations, we would also need to create a dataset that shows the point difference between the handcuff RBs the weeks that they are NOT the MRB and their 3 MORBs. Although this would likely lessen the 9.2-point average difference, it would not lessen the affect enough to approach a 2-point loss a game for holding a non-MRB handcuff.

V. Strategy Breakdown of Handcuff Results

So now let's examine the top 10 and bottom 10 handcuff RB point producers to see if we can learn anything for future years.

Top 10 Handcuff MRB Point Average:

1) L Johnson KC 23.0 points in 4 games

2) N Davenport GB 20.7 points in 1 game

3) A Thomas CHI 14.6 points in 2 games

4) D Blaylock KC 13.3 points in 3 games

5) M Pittman TB 12.5 points in 12 games

6) M Moore MIN 12.3 points in 4 games

7) D Foster CAR 11.3 points in 3 games

8) W McGahee BUF 10.6 points in 11 games

9) J Bettis PIT 10.0 points in 6 games

10T) O Smith MIN 9.7 points in 7 games

10T) S Jackson STL 9.7 points in 1 game

The first thing to notice about this list is that two teams, KC and MIN, appear twice. It would appear that these two teams are very good at plugging backup RBs into their offensive systems due to great game plans and/or offensive lines. These might be good handcuff targets for this year, with Larry Johnson behind Priest Holmes and Mewelde Moore behind Michael Bennett as the two best bets.

Also notice the wide range of number of games as MRB in this list. As great as Johnson was averaging 23 points in his 4 games as the starter, weren't you better off having Pittman averaging 12.5 points in 12 games? Or McGahee averaging 10.6 points in 11 games? You probably were because Pittman and McGahee gave you many more chances to start them as opposed to one game wonders like Davenport and Jackson.

However, this may show that Davenport and Jackson can deliver and are good bets in the coming year. Jackson has been named the Rams starting RB, but Davenport is still behind Ahman Green for the Packers. He may be a better bet than Tony Fisher as the handcuff to Green based on his work last year.

Looking at the actual running backs themselves, it's difficult to find too much of a pattern. All except Blaylock and maybe Moore were pretty highly regarding coming into the NFL. Most are younger, but Bettis is almost to retirement age and Pittman has been around for a while. The most telling aspect of these backs may be that only one of them, Mewelde Moore, is thought of as a situational or "3rd down" RB. The rest were regarded as every down RBs.

Finally, let's look at the team's situation going into last season. Looking back to section III above, the average Handcuff Clarity for these 10 RBs was 2.6. This tells us that most of the best performing handcuffs come from situations were the handcuff was very clear. The two Minnesota RBs are the exception to this rule.

Bottom 10 Non-MRB Yearly Point Averages

1) L Toefield JAX (5.2) points per game

2T) J Jackson CLE (4.1) points per game

2T) C Cobbs NE (4.1) points per game

4T) R Dayne NYG (4.0) points per game

4T) M Morris SEA (4.0) points per game

6) T Hollings HOU (3.9) points per game

7) C Perry CIN (3.7) points per game

8) M Smith BAL (2.8) points per game

9) D Anderson ARI (2.5) points per game

10) J Robertson SF (2.4) points per game

The first thing to notice about this list is that generally the teams represented had poor offensive seasons. This shows that even if the handcuff you drafted never starts a game, at least he should put up some points in garbage time if he is on a good team with a powerful offense. This could save you if you need to use him as an emergency starter some week during the season.

The second thing to notice is that these are all younger RBs whom had never really shown much in the NFL. While the list of Top 10 RBs above includes some younger players as well, most of them were highly acclaimed and drafted as such. These players, with the exception of Dayne, were not as touted coming into the NFL. Tony Hollings only started 4 games at RB in his entire college career!

Finally notice that these RBs are generally behind some of the more solid veteran RBs in the league, including Fred Taylor, Jamal Lewis (both with injury concerns, however), Corey Dillon, Tiki Barber, Shaun Alexander and Emmitt Smith. This shows me that merely finding older or injury-prone RB1s is not necessarily the best way to find a valuable handcuff RB.

VI. Overall Conclusions from the Study

"What does it all mean, Basil?"

If all this analysis of statistical data has left you as confused as Austin Powers, let me try to draw some conclusions using my article against handcuffing from 2 years ago. In that article, I listed 5 reasons why you should not handcuff your star RBs:

1) Not sure which RB is actually the #1 backup

I believe that this issue is still warranted. You must be sure of the actual backup to the RB1 before you draft your handcuff. The Ravens are a good example, as Musa Smith lost you 2.8 points a game over the course of the season, while Chester Taylor gained you 7.9 points per game in his 5 games as the MRB. So in those 5 games where you were trying to backup your starter, Jamal Lewis, you effectively lost 10.7 points per game if you drafted Smith instead of Taylor.

In some rare cases, such as MIN, you could get lucky and have both Onterrio Smith and Mewelde Moore perform as the handcuff. But those cases were few and far between.

2) Not sure that the backup can perform as well as the starter

I believe that this issue also still has some merit, although it is not clear-cut. When we looked at the top handcuff performers, we generally saw RBs who were pretty highly touted coming into the NFL and/or played on good offensive teams. Conversely, the poorest handcuff performers had not shown as much at the NFL level and were on ineffective offensive teams.

So even though some successful handcuff RBs did buck these trends, it still might be a good rule of thumb to only look for handcuffs who have shown some NFL skills and are on better than average offensive teams. This season, I might look at guys like Larry Johnson. Najeh Davenport, Anthony Thomas, Mewelde Moore and Derrick Blaylock.

3) Limited Roster Space Wasted

As shown on the Team Totals tab, this is a concern if the handcuff you draft never starts. In those cases, you likely passed on several RBs who could have contributed a couple points a game more for your team. However, the excellent success of the handcuffs that do start should override that negative of a wasted roster space.

To show this more clearly, we found out that on average, the handcuff RBs that start a game score 13.3 points. We also learned that handcuffs start 22.5% of the games. The other 77.5% of the games, your handcuff is sitting on the bench losing you about 2 points per game. So we can calculate the expected handcuff value by multiplying 22.5% by 13.3 points and adding it to the product of 77.5% and (2.0) points. The positive result is 2.99 points a game and negative result is (1.23) points a game. The total of these results, 1.77 points per game, can be thought of as the expected result of drafting a handcuff.

4) Limited Draft Possibilities

Nothing discussed here touches on the topic of limited trade possibilities, although I still believe that there is value in drafting another prominent handcuff and possibly trading him to the owner of the RB1.

5) Bulls-eye on your chest during the draft/auction

Likewise, there was not any discussion of this effect in my study, but again I still believe that this is a downfall of picking a well know candidate for a handcuff. I have already seen this in several drafts this year where the Priest Holmes owner has everyone trying to jump in and grab Larry Johnson before he can take Johnson as Holmes' handcuff.

VII. Final Thoughts

A couple of final thoughts:

1) Keep in mind that this study was done for a small dataset, 15 leagues in one season. It may be true that 2004 had many more successful handcuffs than a normal season. Or that these 15 leagues drafted many more handcuffs than a normal league.

2) The most important fact about drafting handcuff RBs that I have not yet discussed is that you must get them for a decent value. As mentioned above, I've seen Larry Johnson go way too early in several drafts this year. You are dramatically increasing the risk level of this practice if you take handcuff players too early because that increases the possibility of one or more of the 3 MORBs being great players. They may be full time RB starters as a matter of fact. You also miss out on great players at other positions.

3) I'd also like to mention that you've probably noticed that most of the MRB starts by handcuffs are later in the season. If you do draft a handcuff, you may want to use the strength of schedule indicator to try to grab some other players with favorable early schedules. Having Larry Johnson or Jerome Bettis run wild for you at the end of the year is likely no good if you started 1-6.

4) Finally, this study made me warm up to the idea of handcuffing star RBs. If done the correct way, with risk minimized as much as possible, you have the effective result of adding at least 1.77 points a week to your team. Obviously, you could add even more points than that.

So if the situation arises, don't be afraid to draft a handcuff RB as long as you look at all the success factors of that situation and minimize the risk as much as possible. It can help make you a Fantasy Football Champion!