2019
2018
2017
2016
The Model aka Tiers are based strictly on numbers. A players numerical profile dictates which Tier they end up in. Scouting reports are for funsies. The model was designed to detect players that will rush for multiple thousand yard seasons.
____________________________________________________________
Sorry for the RB Report being so late this year, I do apologize to anyone who was looking forward to it. In any event, enough time has finally elapsed where I’ve been able to analyze what went right, what went wrong, and everything in-between. It’s also a talented and diverse RB class that has been fun to evaluate, which helps. I’ve written quite a lot in a short amount of time, so if you see any mistakes feel free to let me know. And as always, I’m happy to answer any questions asked in good faith.
News and Notes
- There are 3 RB’s identified in this year’s report. Taylor, Akers, and Dillon qualify for Tier 1 (Power/Speed). No other RB hit for any other group although Swift was close, which is detailed later.
- I’m changing Tier 4 “flash in the pan” to Tier 3, and Tier 3 “replaceable backs and dart throws” to Tier 4. I think this better represents the value and certainty of each group.
- Enough time has elapsed where I was finally able to crunch some numbers on the hit rates of my model. If I did all my maths correctly, when we examine rb’s selected in the top 3 rounds from 2015-2017 and look for at least 1 1k yard rushing season, my model hit 83% of the time as opposed to 50% of the time for the field. If ignoring the 1st round rb’s and only focusing on the 2nd+3rd round rb’s drafted during the same time period, the difference is magnified to 75% hit rate for my model vs 33% for the field. More on this below.
- Problem! My model is likely identifying players at too great a rate. One reason could be that the rb’s entering the league now are much better than from 10-15 years ago. Another reason could be that my model is identifying some players that will bust. For now, I’m operating under the assumption that it’s a bit of both. More on this later.
- A trend I’ve been noticing is that the more data I collect, the more I’m able to simplify. I’m strongly considering combining Tier 1 (Speed/Power) and Tier 2 (Fake Bellcows) together at some point in the future. As time passes, I’m seeing the range of outcomes for each group (Tier 1 and Tier 2) expand into each other. Their criteria are exactly the same but for one minor difference, which if done away with means everything will be the identical. While I’ll hold off on making the change this year, I may start operating under the assumption that they are much more similar than I previously thought, and a change may be coming in future years.
- The NFL switching to a 17 game season presents a bit of a problem for me as I judge my model’s accuracy based on 1k yard rushing seasons over the course of 16 games, not 17. I’ve thought about this and feel like the best way to move forward would be to keep the 1k yard rushing season but also include a qualifier of needing to reach a required YPG average (62.5 YPG or better) which would equate to a 1k yard rushing season in a 16 game schedule. It may not be perfect (what if a player rushes for 200 yards in their 17th game or something, propelling them over the cusp) but I feel like those types of situations will likely be few and far between. If anyone has any better ideas, feel free to let me know.
____________________________________________________________
2015-2017 Overview
Full Tier 1 + Tier 2 History since inception aka predicting players [# of 1k yard rushing seasons]
2020 – Jonathan Taylor
2020 – Cam Akers
2020 – AJ Dillon
2018 - Saquon Barkley [2]
2018 - Rashaad Penny [0]
2018 - Derrius Guice [0]
2018 – Nick Chubb [1]
2017 - Leonard Fournette [2]
2017 - Joe Mixon [2]
2017 - D'Onta Foreman [0]
2016 - Ezekiel Elliott [3]
2016 - Derrick Henry [2]
2015 - David Johnson [1]
Given my research has shown we should allow a full 3 years for a player to break out, let’s examine the first 3 years of my model to see how it has fared vs the field. Since there has not been an ample amount of time for analysis of multiple hit seasons, for now we will define a hit as 1 1k yard rushing season. Also, the criteria for players like CMC were revised last year so we’ll leave him out of my models hits for arguments sake. For hits from my model, I’ll use only the main lists (Tier 1 and Tier 2) as those were the original Tiers which is the foundation of the rb report. They're the most predictive and their criteria is practically identical. The other tiers were created later and it’s yet to be seen how predictive they truly are.
Complete NFL Draft (first 3 rounds)(hit = at least one 1k yard rushing season)
2015 - 8 taken top 3 rounds, 3 hits (Gurley[3], Gordon[1] , DJ[1])
2016 - 4 taken top 3 rounds, 2 hits (Zeke[3], Henry[2])
2017 – 8 taken top 3 rounds, 5 hits (Fournette[2], CMC[2], Cook[1], Mixon[2], Hunt[1])
Prospects selected within the top 3 NFL Draft Rounds that hit for at least one 1k yard rushing season;
My Model – 5 of 6 hit, for 83% hit rate
All Players – 10 of 20 hit, for 50% hit rate
Prospects selected within the top 3 NFL Draft Rounds that hit for multiple 1k yard rushing seasons;
My Model – 4 of 6 (67%)
All Players – 6 of 20 (30%)
Within the group of players that hit, percentage of players who hit for multiple 1k yard rushing seasons;
My Model – 4 of 5 (80%)
All Players – 6 of 10 (60%)
So we have an increased hit rate vs the field in all areas. When the players my model does identify hit, they hit harder aka they are more likely to put up multiple 1k yard rushing seasons than your typical hit. Keep in mind this is having to leave out Gurley whom I believe would have qualified had he done his athletic testing, and CMC who I am leaving out because I retroactively made a small common-sense change in one criteria to include him and DeAngelo Williams.
Now, we just so happened to have a crop of 1st round RBs from 2015-2017 that all hit for 1k yard seasons. You might be saying, your model's hit rate is great but 1st round Rb’s are all good and so you have a high hit rate... so what? Well firstly, 1st round Rb’s are absolutely not going to hit at this rate for forever. In 2012, you only had a hit rate of 1 of 3, the lone hit being Doug Martin. In 2010 the hit rate was 2 of 3. In 2009 again the hit rate was 2 of 3. So the hit rate for 1st round RB’s to get at least 1 1k rushing yard season is not naturally 100%. Let’s take the analysis one step further and remove 1st round RB’s and see what we have left.
Players selected in rounds 2-3 of NFL draft that hit for 1k yard rushing season;
My Model – 3 of 4 for 75% hit rate
All Players - 5 of 15 for 33% hit rate
That is a significant difference and I believe this is where my model really shows its value. 1st round RB’s are more likely than not to be productive, but sifting through the 2nd and 3rd round rb’s without guidance can be a minefield.
____________________________________________________________
The Busts
As time has passed, one thing is quite clear. There Will Be Bust. When I first started this project I had no data for busts, presumably for two reasons. The data I was using was pretty reliable going back a few years before 2015 but before that it got pretty spotty. So the players who washed out of the league quickly were not included in that data set, as they were no longer in the league and therefore not in the playerprofiler database when it was created. If we analyze the period of 2005-2014 (so before my model was created), there were 13 players retroactively identified by my model, or roughly 1.3 per year. Because they were hits, they were still in the league, and therefore in the data set, and therefor able to get a hit from my model. However, since my models inception, there have been 13 players identified (predicted) in a mere 6 years, or 2.2 per year. 1.3 to 2.2 is a pretty big difference. Now, I think there are 2 factors at play here. One is that the RB talent entering the NFL these past few years has been significantly better than during the period of 2005-2014. However, the 2nd factor is that clearly there were players who would have been identified by the model but busted, and because they busted were therefor washed out of the league before they were able to have their data entered into the database and in turn then be picked up by my analysis. If we were to split the difference between these two factors (better talent entering the league now and missed busts in the past) we would get an average of 1.75 players who in theory should be being identified by the model per year that won’t bust, we’ll call them true hits per year. That means it is likely that roughly 2 (which would amount to 1.83 true hits per year) to 3 (leading to an average of 1.67 true hits per year) of the players identified by my model between the years of 2015-2020 will bust. But which ones? Let’s find out…
Looking back, there are players here and there that may have qualified for my model but I just don’t have accurate data for. Kevin Jones is a guy back from 2004 who might have qualified for my model and hit is rookie year but then his career was derailed by injuries. Let’s look at the current bust candidates. D’Onta Foreman is a guy that looks extremely likely to be a bust. He has not broken out for a full 3 years and his career has been derailed by injuries, among other things. Penny is the 2nd highest bust risk imo, who is looking extremely unlikely to break out in his 3rd year as he may start the year on the PUP list. You could say the same for Guice, however he is healthy going into his 3rd year and has looked great (or at least much better than Foreman and Penny) when on the field. So for now let’s examine Foreman and Penny and see what we can learn.
Both Foreman and Penny;
- 1 big final collegiate season (Kevin Jones also had this)
- Terrible at pass pro
- Issues with being overweight
- 1 cut and go runners, not overly dynamic, you could argue stiff hips
- Of the players identified as hits by my model, the fantasy community as a whole regarded Foreman and Penny the lowest in their respective classes
- Suffered injuries in NFL
Now I overlooked a lot of these issues because there is a lot to like about each player as well. But in hindsight they share a lot of the same problems. I think the pass pro and weight issues stem from a lack of work ethic, and I think that might be the most worrisome. I don’t believe any one issue is a huge deal in and of itself (you could argue Derrick Henry played a bit stiff and straight-linish, Fournette had weight problems, Melvin Gordon was not good at pass blocking, etc) but several or all issues combined may present a confluence of mitigating factors that a player just can’t overcome. I think injuries in the NFL, particularly significant injuries, can also result in a bust, however I’m not sure you can forecast that sort of thing. It was said by the Seahawks that Penny was drafted, in part, because of his supposed durability. In any event, I think it’s good to analyze what went wrong and see if we can identify any future players as potential busts. Interestingly enough, where a player was drafted (assuming top 3 round draft capital) doesn’t necessarily seem to affect bust risk as long as they fit my models criteria.
The 3 players this year are Taylor, Akers and Dillon. Dillon is the lowest rated of the group so let’s examine his bust risk factors;
- He did not have 1 final big collegiate season (none of the 3 did). All 3 were productive, with the 1st and 3rd being most productive.
- He is not terrible at pass pro (none of the 3 are).
- His weight is high but he’s very fit, muscular and well built… he carries his weight well.
- He is ideally a 1 cut and go runner, however he does have loose hips to me.
- He is the lowest rated of the hits
Trying to find an overall bust rate (and since only 5 years have elapsed I’ll again just use a single 1k yard rushing season for right now), currently the model has an 83% hit rate when all prospects are given a full 3 years to hit. The way it seems to me is that Foreman is trending to be a bust, and one of Penny/Guice is trending to be a bust (my guess is Penny for the reasons outlined above). That would be 2 busts for 10 players, or 80% hit rate and 20% bust rate. Given that none of the 2020 prospects display many bust-risk factors, nor Guice, plus the fact that Gurley wasn’t able to test athletically, I think the real hit rate may end up being somewhere around 85% with a 15% bust rate, which more closely mirrors the hit/bust rate for the players that have been given the proper amount of time as detailed above. Again, that is only for a single 1k yard rushing season, and we are really looking for 2 or more, ideally 3+. But for that we simply need more time, no other way around it.
____________________________________________________________
Player Scouting Reports
AJ Dillon
I really like his low pad level for a man of his size. For example, I thought Rashaad Penny struggled with things like pad level when watching his film, Penny was too upright a lot of the time. Dillon has good contact balance, paired with better than expected burst through the line. Given that he’s 247 lbs he’s much quicker than I thought he would be. Obviously he’s not the most agile player but he’s shifty enough, he’s got a decent jump cut and foot quickness is ok. Dillon has good top speed and is a load to bring down. Fluid athlete with flexible hips. The one big criticism I have of Dillon is that once he gets past the LOS he’s not the most creative runner. I had the same criticism of Chubb though and he just racked up nearly 1500 rushing yards this past year so it may not be that big of a deal. His weight is very high but he’s well built. That said, I think he could benefit from losing 10 lbs or so to add just a bit more quickness and flexibility. He’s not a plus in the passing game but I don’t think he’s not a minus either. He’s adequate at pass blocking and can catch most balls thrown at him. Most of his receptions will probably be the Fournette type though, screens, dumpoffs. Dillon is not a good route runner imo.
Comps
High – Ricky Williams
Low – Eddie Lacy
Jonathan Taylor
Very similar to Dillon but an upgrade in basically every area. Noticeably lighter and quicker. One noteworthy difference is that in the games I watched, Taylor was not utilized to pass block much. Instead he ran out into the flat most of the time. I can understand why a player like CEH would go ahead of Taylor is a team was looking for a sure-thing in the passing game. I don’t think Taylor is incapable though, he’ll just have to earn it similar to Fournette. His college target share (which to me is a predictive stat when transitioning to the NFL) is actually well above average.
Comps
High – Ladanian Tomlinson. Comping Taylor to LT is probably a bit of a reach, but given JT's incredible profile it's hard to think of anyone else to comp him to.
Low – Fournette type
DeAndre Swift
I really like Swift but one thing that stood out to me was a seemingly apparent lack of burst and power. I had completely forgotten what his combine measurements were and was somewhat shocked that he ran a 4.48. When I watch Swift I see somewhat of a savant, in that he’ll make up for any shortcomings with a natural feel for the position. I really like his contact balance and ability to run through arm tackles. I love his ability to create and overall feel for the game. He was asked to pass pro a lot and was good in pass pro and good as a receiver. Overall there is a lot to like here, and I think a Kamara comp is actually warranted. Pairing him with Kerryon who theoretically would handle more of the inside work and Swift the outside work seems to make sense.
While no other rb’s qualified for current criteria, Swift was very close to the “Flash in the Pan” group which is near certainty for 1-2 1k yard rushing seasons. Given that the sample size is so small for that group (and therefore I’m less able to refine the criteria along with their parameters), I’m not opposed to making a minor adjustment to the one criteria he fell short of. However, that is probably something I’ll do in the future after seeing how Swift pans out.
Comps
High – Alvin Kamara
Low – Devonta Freeman
J.K. Dobbins
I was a bit startled that he only weighed in at 209 lbs. Dobbins was maybe the most hot and cold player I’ve watched this year. I like his overall power and athleticism. I feel like he was questionable in the passing game. At times he could line up and stone an oncoming rusher, other times (notably against Wisconsin) he was pushed around a LOT by defenders. In the receiving game, he could make great plays and other time bad plays. For the most part I liked what I saw, but there were enough question marks to make me wonder if he’s going to be ready to unseat Ingram this year. My bet would be no.
If Dobbins would have put up these numbers at the combine or pro day, he would have qualified for the Tier 1 Power/Speed list. It's very unfortunate that he didn't get a chance to test so that we could know for sure.40- 4.32
20 yard dash - 2.48
10 yard - 1.44
Vert - 36.5
Broad - 129
https://www.cleveland.com/osu/2017/08/f ... _buck.html
Comps
CEH
Maybe someone who pays attention to college football can fill me in. Typically, when I watch film I start from the beginning and end at the end. However, this year I went the lazy route and just watched the games in order of how they popped up on youtube, which tended to be from last to first. The first few games I watched, he looked horrible. I was stunned and in disbelief. But the further back I went, the better he looked. I was watching 2 completely different players. I eventually got curious enough to google if he had an injury and sure enough, in December I think it was before his final few games he had a somewhat significant hamstring injury. So I’m inclined to believe that this is why he looked so different. Regardless, in the most recent games I watched he straight up looked like a turd. However the games further back, he looked pretty good. Good initial burst, ability to break arm tackles, and good shiftiness. A plus as a receiver but was not asked to pass block much and when he did it was spotty. There are a few concerns I have with him. When he’s completely healthy, yeah he’ll probably look good. But in the NFL you have to play through injuries a lot and I question how well he’ll play through them. Second is that he did not look like the most natural inside runner to me, which again makes me question if he’ll ever be the unquestioned bellcow. I think for the most part my concerns don’t really matter because his main value will come from catching passes, and I think he’ll be very good at that. But I think if you believe he’s gonna be CMC or anything close to it you may be misguided.
Comps
I think the Westbrook comp is very valid
Ke’Shawn Vaughn
Solid all-around player but looks more like a high quality backup that may get chances to start rather than a high quality starter. I’m not seeing any difference-making traits. I like him, I think he’s fine, but he would be clearly below the other 5 rbs in this class.
Cam Akers
I like his prototypical size, his burst, speed and athleticism. I like the fact that he was arguably the best rusher, passer and blocker on his entire team. The common criticism of Akers that I see is that he is “raw”, and while that may be the case I don’t know how anyone could judge anything he did behind that atrocity of an offensive line. I haven’t seen an offensive line that bad since I scouted David Johnson back in 2015. What really excites me about Akers is his upside in the receiving game. He was asked to pass block A LOT, and he did really well in that area. He extends his arms and uses his hands to catch the ball. He shredded the RB drills at the combine. I love his fit with the Rams. Goff isn’t great but it could be a lot worse. The O-line is said to be subpar, but the same was said about the Seattle O-line as an argument against Penny and yet Penny’s career YPC is 5.3 and Carsons YPC during the same time period is 4.5. I love the fit with an innovative head coach that will likely maximize his abilities and I love the fact that he has zero competition there in the long-term and close to zero imo (we’ll have to see given the shortened covid offseason though) in the short term given Akers proficiency in the receiving game. I have no idea why more people aren’t higher on him.
Comps
Lynch is one of a kind but from an overall prospect profile coming out of college, Marshawn Lynch. I realize I also comp'ed Guice to Lynch 2 years ago, but that was more running style rather than overall profile. I don't think Akers will be a Lynch-type of runner, but I couldn't come up with a better comp.
____________________________________________________________
Full Player Tier Listing
____________________________________________________________
Tier 1 – Long-Term Studs [Power/Speed]
2020 – Jonathan Taylor
2020 – Cam Akers
2020 – AJ Dillon
2018 - Saquon Barkley
2018 - Rashaad Penny
2018 - Derrius Guice
2017 - Leonard Fournette
2017 - D'Onta Foreman
2016 - Ezekiel Elliott
2016 - Derrick Henry
2015 - David Johnson
2011 - DeMarco Murray
2010 - Ryan Mathews
2008 - Darren McFadden
2008 - Chris Johnson
2008 - Matt Forte
2008 - Jamaal Charles
2007 - Adrian Peterson
2007 - Marshawn Lynch
2006 - Maurice Jones-Drew
2004 - Steven Jackson
2001 - LaDainian Tomlinson
Tier 1 – Long-Term Studs [Wiggle/Agility]
2017 – Christian McCaffrey
2013 – Le’Veon Bell
2009 – LeSean McCoy
2008 – Ray Rice
players on this list (at least the ones who were given ample time and opportunity so far) have produced a minimum of 2 1,000 yard rushing seasons. Most have 3+.
Foreman trending to be the first official bust of this list.
____________________________________________________________
Arian Foster, Frank Gore, Alfred Morris, Mark Ingram, Todd Gurley [no athletic testing for Gurley] are players who have attained 3+ 1k yard rushing seasons that don’t qualify for any of the above lists. A trend I’ve noticed is that outliers (good or bad) seem to miss the combine. Foster, Gore, Gurley, Foreman all missed the NFL Combine and athletic testing had to be taken from their respective pro days. Gurley was still injured so he did not have a pro day either.
____________________________________________________________
Tier 2 – Fake Bellcows
2018 - Nick Chubb
2017 - Joe Mixon
2012 - Lamar Miller
2008 - Jonathan Stewart
The players on this list are often good football players but just never really seem to live up to expectations. Seems to be a near certainty they will rush for 1-2 thousand yard seasons but none have 3 or more. As noted earlier, I’m contemplating combining this tier with Tier 1 Power/Speed by making the parameters of their criteria the same.
____________________________________________________________
Tier 3 – Flash In The Pan
2014 - Carlos Hyde
2013 - Eddie Lacy
2012 - Trent Richardson
2012 - Doug Martin
2006 - DeAngelo Williams
These are running backs that start off red hot but then cool quickly. Great potential for at least 1 1k yard season but maximum of 2. No additions this year.
This year Swift is very close to this Tier and I may look to make a change to it in the future.
____________________________________________________________
Tier 4 – Replaceable Backs and UDFA Dart Throws
Year__Pick____Name
2019 – 5.02 – Ryquell Armstead
2018 - 3.07 - Royce Freeman
2017 - 3.22 - Kareem Hunt
2016 - 5.34 - Alex Collins
2016 - 4.36 - Kenneth Dixon
2016 - 5.11 - Jordan Howard
2015 - 5.13 - Jay Ajayi
2011 - 1.28 - Mark Ingram
2009 - 1.12 - Knowshon Moreno
2008 - 6.10 - Jalen Parmele
____Undrafted
2019 – Alex Barnes
2015 - Thomas Rawls
2015 - Zach Zenner
2014 - Isaiah Crowell
2014 - Terrance Cobb
2008 - BenJarvis Green-Ellis
I separated the list into drafted and undrafted players because as you can see the drafted players have a much better hit rate and also a much higher ceiling overall. Drafted players on this list are likely to get at least 1k rushing season under the belt but none have more than 2.
Since this lists inception a few years ago (I believe I created this in 2017), none of Dixon, Freeman or Armstead have hit yet, which gives me pause to the veracity of this Tier.
____________________________________________________________
Tier 5 – Pass Catchers
Year__Pick____Name
2015 - 3.13 - Duke Johnson
2014 - 2.22 - Bishop Sankey
2013 - 2.05 - Giovani Bernard
2011 - 2.24 - Shane Vereen
2008 - UDFA - Danny Woodhead
2005 - 4.29 - Darren Sproles
Test Cases
Year__Pick____Name
2018 - 4.34 - Chase Edmonds
2017 - 5.39 - Aaron Jones
This tier is composed of players that likely won’t be 3-down RB’s (none have had a 1,000 yard rushing season) but may be valuable pass catchers (most have had multiple years of 40 or even 50+ receptions) and carry good value in PPR leagues. They have good longevity, as you can see the list is composed of players stretching from over a decade ago that are still active in the league.
The test cases are players who missed the requirements by a slim margin. I may take off the test cases next year.
____________________________________________________________