Near the beginning of the NFL regular season I posted on the site my introduction to the Workhorse Metric, which measures the extent to which any college runner dominates the representative non-quarterback rushing production on his team. This present post is to apprise you of my journey down, through, and out of the analytical rabbit hole since then.
In the past few months, I have examined the last five draft classes (2010-14) and calculated the Workhorse Score (WS) for each running back drafted in those years as well as many undrafted free agent (UDFA) RBs to enter the NFL in that timeframe, including all of those with at least one top-36 positional campaign. I’ve also gathered a variety of other production statistics for each player as well as all the biophysical information I can find: age, weight, height, speed and agility measurements, etc. Basically, I’ve built the foundation of the eventual database that I hope will one day turn Jerry Jones’ RB dreams from dry to wet.
With all of this data in hand—or, rather, in computer—I’ve been busy looking for actionable trends, and I believe that I’ve found some. While I intend to continue researching the RB position, with the information I have accumulated so far I am now confident that I can make some definitive statements, such as this one: The Workhorse Metric could be an important predictor of RB potential.
Ever since we at RotoViz started looking at RB production in terms of market share, this question has presented itself: “Is market share of rushing production more predictive of NFL RB production than raw statistics are?”
When it comes to late-round (LR) and UDFA RBs who have had NFL success, the answer is unequivocally “Yes.”1
The Predictiveness of High Workhorse Scores in Late Round and UDFA RBs
Some highly drafted players with high WSs have done well—Le’Veon Bell, take a bow—but lots of early-round RBs without great WSs have also had NFL success: Eddie Lacy is a prime example. In general, as a group first- and second-round (R1 and R2) RBs don’t need strong WSs to have strong careers. I tend to like early-round RBs with high WSs (all else being equal), but I don’t think that a high WS is anything close to compulsory.
With late round and UDFA RBs, however, I think a high WS is almost necessary. Rare is the late round and UDFA Top-36 RB who does not have a high WS. In general, this cohort sees a noticeable correlation between WS and NFL success.
Let’s put some numbers to this: In 2014, out of all the R1 and R2 RBs to enter the NFL in the last five years, only two had their first top-36 campaigns—Jeremy Hill and Shane Vereen. Together, they have an average WS of 72.1. Meanwhile, five late round and UDFA RBs to enter the NFL in the last five years had their first top-36 campaigns in 2014: C.J. Anderson, Isaiah Crowell, Matt Asiata, Branden Oliver, and Denard Robinson. Collectively, they average an 83.80 WS. You probably don’t need me to tell you that those numbers are not atypical and that the difference between those numbers is significant . . . but I’ll tell you anyway.
Or, rather, I’ll let the numbers tell you. Here’s a table with the WSs for the last five draft classes and the standardized significance (Z-Scores) of those WSs:
|2010-2014 RBs||Mean WS||Median WS||WS Z-Score|
As one might expect, we can see a very clear trend: The higher the draft position, the higher the WS. Note that the WS for UDFAs is imprecise, because 1) the definition of “UDFA” is somewhat ambiguous—are all guys who go undrafted automatically UDFAs, or only the guys who at least make it to a NFL tryout or camp, or only the guys who actually make it onto an active NFL roster?—and 2) calculating the subpar WSs for all the pseudo-nameless UDFAs who have barely been in the NFL in the first place would’ve required countless hours I simply didn’t have. Rest assured, though, that when I have taken the time to calculate WSs for random UDFA RBs the trend we see here holds true: In general, the more draft capital the NFL invests in a RB, the higher his WS.
And yet against this trend we can see another trend—one much more fascinating. Here’s a table with the WSs for all the RBs to enter the NFL in the last five years and have at least one top-36 season.
|2010-2014 RBs||Mean WS||Median WS||WS Z-Score|
For RBs with NFL success, the higher the WS, the lower the draft position. We can see a clear trend of inverse correlation. And the trend is even more intriguing when we examine the tiered breakdown.
For R1 and R2 RBs, the WS means little if anything. In general, the successful R1 and R2 RB has a WS similar to that of the average drafted RB and actually lower than that of the average R1 and R2 RB. At best, for RBs selected with an early pick, a high WS means little.
For third- and fourth-round (R3 and R4) RBs, however, the story changes. Successful R3 and R4 RBs have WSs significantly higher than the average R3 and R4 RB and actually equal to that of the average R1 and R2 RB. And if you think about the production we’ve seen from guys like DeMarco Murray, Stevan Ridley, Lamar Miller, and Knile Davis, this statement feels true. These are R3 and R4 RB s who, years after entering the NFL, with the clear light of time are seen to have been equal as draft prospects to R1 & R2 RBs such as Mark Ingram, Trent Richardson, Doug Martin, and Christine Michael. In general, the R3 and R4 RB who eventually has NFL success looks a lot like the average R1 & R2 RB when they both enter the league.
For top-36 late round RBs the trend is even more pronounced. While the average late round RB has a 66.73 WS, the top-36 late round RB has an 80.22 WS—higher than the WS of even the average R1 and R2 RB. And if you think about it this makes some sense. Wouldn’t you say that Alfred Morris in his three-year career has been as productive as any R1 and R2 RBs of the last five years?
And, finally, despite coming from an original cohort with the lowest average WS to be found, the top-36 UDFA RBs have the highest WS of any subgroup, and if you look at the WS Z-Score for the top-36 UDFAs you can see that the degree to which it surpasses the average WS for all drafted RBs is in fact significant.
And who are these UDFA RBs who become contributors? Think of guys like C.J. Anderson, Joique Bell, and LeGarrette Blount . . . and, if I had included the 2009 rookies in this sample, Arian Foster. You know, big guys who can carry the ball many times per game and whom no defender really wants to tackle. In other words, workhorses.
So we have two intersecting data points that result from the trends above: late round and UDFA RBs as a total group have the lowest WSs of all RBs—and the late round and UDFA RBs with NFL success have the highest WSs of all RBs. That doesn’t seem coincidental.
In general, the higher a RB is drafted, the likelier he is to have NFL success. Here’s a table showing as much:
|2010-2014 RBs||Total||Top-36||Top-36 Percent|
In other words, throughout the whole population of RBs an overarching correlation exists: The higher a RB is drafted, the higher the WS, and the higher his historical odds are at having NFL success.
But is any causation to be found in any of these correlations? At least within the late round and UDFA RB cohort, yes.
Out of the 13 such RBs to enter the NFL in the last five years and have top-36 campaigns, all of them except for two had WSs better than the average WS for drafted RBs across the timeframe. That’s pretty good. And while the percentage isn’t quite as good for the UDFAs, if you look only at LR RBs with WSs no lower than 85—which is the unofficial number that starts to catch my attention—you’ll see that, of the eight such RBs to enter the NFL in the last half decade three have had at least one top-36 season. That might not sound like much, but that’s a 37.5 percent success rate out of a larger group that generally sees only a 13 percent success rate. That’s like mining for copper and finding gold… as long as you recognize that what you’re holding in your hand is gold.
The Workhorse Score over Raw Statistics?—Why Not?
Can’t someone just look at raw stats instead of workhorse scores and get the same result?—or at least an equivalent result? I don’t think so.
As one would expect, the more draft capital invested into a RB, the more pts/g he scored in his best college season.
|2010-2014 RBs||Mean Pts/G||Median Pts/G||Pts/G Z-Score|
The expected production-correlates-with-draft-position trend holds. But if you dig into the numbers, you see that the raw collegiate production of Top-36 non-R1 and R2 RBs isn’t as significant as their Workhorse Scores.
|2010-2014 RBs||Mean Pts/G||Median Pts/G||Pts/G Z-Score|
As is the case with the WS, by raw per-game stats the R1 and R2 RBs who eventually have NFL success are actually less productive than the R1 and R2 RBs who don’t. For these players, opportunity and non-production metrics seem to matter much more than collegiate production.
And for the Top-36 R3 and R4 and late round or UDFA RBs, their raw per-game stats are far less distinguishing than are their WSs.
|2010-2014 RBs||WS Z-Score||Pts/G Z-Score|
I don’t think that raw stats are useless—in fact, I think that the best practice would be to use raw stats and the Workhorse Metric jointly—but lots of RBs can be found who have decent raw stats who end up doing little in the NFL.
Or, to put it differently, the Workhorse Metric differentiates in a cleaner manner those likely to have NFL success from those who aren’t. By raw stats, all of the guys who have top-36 seasons look like versions of each other. They all average roughly a similar number of points/game, and they all look like guys selected in the top half of the draft, regardless of when they were actually selected. By WSs, however, the guys with top-36 seasons have greater distinction. While the top-36 mid-round RBs look just as good as the average early-round RBs, the late round and UDFA RBs really stand out, with WSs that are better than any other subgroup’s.
Raw stats are fine, but they don’t highlight the late round and UDFA RBs likely to become NFL contributors the way that the Workhorse Metric does.
The Value of Top-36 Late Round and UDFA RBs
Nothing is more important to building a championship team—whether it’s a NFL squad or a fantasy roster—than finding contributors late in the draft or on waivers. Everyone wants to have the next out-of-nowhere stud.
The problem is that, especially in dynasty leagues, these players are inordinately hard to find . . . at least at certain positions. For instance, in the 2014 season only three quarterbacks to enter the NFL as late round or UDFA players had a top-24 positional campaign. That’s a success rate of 12.5 percent. Additionally, in 2014 only four wide receivers to enter the league as late round or UDFA players had a top-36 positional campaign. That’s a success rate of 11.1 percent. In other words, rostering late round or UDFA QBs and WRs is a great way to lose your dynasty league every year. It’s not likely that you will be able to find the next Tom Brady and Antonio Brown.
But you do have a pretty decent chance of finding the next C.J. Anderson, especially with the Workhorse Metric. In 2014, a whopping 15 late round or UDFA RBs had top-36 seasonal performances, good for a success rate of 41.7 percent. That’s insanely high. Over 40 percent of this season’s top-36 runners were guys valued at almost nothing when they entered the NFL. Zero RB, indeed.
If finding late round and UDFA contributors is crucial to winning a championship, and if Zero RB is an ideal strategy through which one can leverage these contributors, and if the Workhorse Metric is the best tool we have for finding late round or UDFA players who are likely to become NFL (and thus fantasy) contributors, then, as I said at the beginning of the article, the Workhorse Metric very well could be an important RB statistic.
A lot of assumptions are in the foregoing sentence, and I’ll leave it to you to determine which of those are legitimate and which aren’t, as I’ve probably said enough already. And as for the Workhorse Metric, the numbers speak for themselves.
Matthew Freedman is a writer for RotoViz and is (not) the inspiration for the character in The League who shares his name. He serves as RotoViz’s (un)official ombudsman in the series The Dissenting Costanzan, and he also co-hosts the RotoViz Radio Football Podcast and writes The Backfield Report and The Wideout Report. He is the creator of the Workhorse Metric and the No. 1 fan of John Brown, the Desert Lilliputian.
- I consider any RB drafted no higher than the fifth round to be a late-round RB. (back)