Gut Check No.351: Week 6

Charcandrick West and SPARQ, Green vs. Sherman, Travis Benjamin-Josh McCown, Gio Bernard, and the scouting lag.

Welcome to my in-season column. If you're seeking strategic tips, match-up advice, game observations, IDP info, and dynasty advice with a little bit of current "football" events opinion (if anything involving the madness of the NFL commissioner's office actually qualifies) added into the mix then you've come to the right place. Think of the in-season Gut Check column as a department store with a variety of areas: 

  • Re-Draft
  • IDP
  • Dynasty
  • Futures (Rookies and College Prospects)
  • Opinion
  • Wild Side

How much attention each department earns depends on what I observe. If you missed the first two weeks, you can pay a visit down memory lane. Here are weeks 123, and 4.  

Last week, I wrote about players capable of filling bye-week gaps and many of them have the potential to become permanent starters. The names included, Gary Barnidge, Allen Hurns, Willie Snead IV, Marquess Wilson, Duke and David Johnson, Stefon Diggs, Ladarius Green, Jamison Crowder and Dwayne Harris. If any of these players are still on you waiver wire or you're in the market to negotiate a deal that either involves one of these players as an added component to match what you're giving up, read here

Re-Draft: SPARQ Score and Chiefs RB Charcandrick West

Should you blow your waiver stack on the second-year back from Abilene Christian? It's the most common question I've seen on Monday after Jamaal Charles suffered the second ACL tear of his NFL career. The answer from the athlete centric crowd is yes.

They love West's SPARQ metrics. The Chiefs new starter had the sixth-best score among 2014 rookie runners, besting Tre Mason, Dri Archer and Charles Sims. If you're not familiar with SPARQ, Wikipedia is the best place to learn. If you don't care, skip the next 11 paragraphs and look for the bold heading: What about West?

The Seattle Seahawks value high peformers in the SPARQ rating. As the NFL became more interested in the score, Nike removed the breakdown of the rating from the public eye. You'll find a somewhat reverse-engineered version that both Field Gulls and Rotoworld like to feature. 

I say somewhat because it's not perfectly reverse-engineered and that's a bit of a problem for data reliability. The best public version of the SPARQ rating is likely overfitting the data. How do I know this? I had two trained statisticians use the same process that Zach Whitman describes at Field Gulls and both independently concluded that, unless Whitman has specific data that he's not sharing, it's impossible for him to be getting .95+R^2 without overfitting the data.

I wrestled with the idea of writing about this for a few months because I think Whitman has done a good job of bringing SPARQ to the public eye. He's an engaging guy and I think his reverse-engineered rating gives the public a basic idea of the kinds of things that are on the forefront of the NFL's process to determine the kind of athletes that the teams value. But for the sake of transparency, people I trust in this arena have questions about the validity of the accuracy of the reverse-engineered score as a reported. 

The first trained statistician is my brother, who holds a masters in the subject from Columbia. He's currently earning his PhD at Harvard (feel free to dislike him for that fact, although I find Penn or Dartmouth far more snobby when it comes to the name-dropping factor.). The second trained statistician works in the league and he's intimately familiar with the SPARQ score.

Since it wasn't in the second person's best interest to do any leg work, I had my brother attempt to reverse-engineer the SPARQ rating the way that Zach Whitman has detailed at Field Gulls. I gave my brother all the combine data that's available, sent him to the Field Gulls post, and let him get to work. 

My brother's first brush with skepticism occurred after reading the article: 

rSPARQis a formula based on five inputs: weight, 40-yard dash, shuttle, powerball toss, and vertical jump. Whitman explains that combine data does not provide powerball toss, so he essentially substitutes bench press for his value based on some conversion. To make this conversion he relies on high school data. I'm not convinced that Whitman's conversion is appropriate. First, because he is just using bench press as the sole predictor of powerball toss, he would actullay "explain more variance" (i.e. R^2 values closer to 1) in rSPARQ using the raw bench press. 

Second, I'm not convinced that the relationship between bench press and powerball toss in high school would be the same in college athletes entering the draft as it would be in high school since the powerball toss relies, in part, on innate coordination. Athletes may bulk up in college considerably, but not become more coordinated in a domain that the powerball exercise also tests. 

Is there any available data that has both the bench press (and ideally other predictors) and powerball toss? If so, I would like to redo Whitman's analysis in this area. 

After I couldn't find any powerball toss data, my brother reported his initial findings: 

With the five variables Whitman indicated (using benchpress instead of the powerball toss), it looks like the best that I could do following Whitman's approach is produce a model with a predictive R^2 = 0.714. A perfect fit would be R^2=1 and Whitman is reporting impressive R^2's in the 0.98-0.99 range, which raises red flags for me because there is no way that bench press is that predictive of powerball toss to warrant R^2s in this range even if Whitman perfectly specified the model). 

Improving fit requires look at what possible variables we are omitting. We know one omitted variable is powerball toss, but my guess is that there is something else. Perhaps SPARQ has a different formula for offensive players different than defensive players or lineman positions have a different scheme than skill positions. 

It turned out that my brother didn't see the complete combine data I provided him (those pesky Excel spreadsheet tabs can be so elusive), so he returned to work with the years of combine data I provided him. 

I'm up to the R^2=0.9 range by throwing in the kitchen sink (all the tested combine variables, plus adjusting for position) into the variable selection algorithm. Of course, this makes the formula rather lengthy and there is some missing data that I need to deal with to report anything to you with confidence. Still, 0.9 is, I must say, pretty darn good. Things like Whitman's R^2=0.98 is likely unattainable, and honestly makes me skeptical. 

At this point, class was in session and my little brother disappeared into Ivy for the fall. I took this information and ran it past my NFL guy. I knew he'd be uneasy about doing any legwork that might violate his protocol, but he'd be ok commenting on something that someone else had done. 

Looked at the SPARQ stuff. No way you get .95 + R^2 without overfitting the data; I went through the same process, hit .8ish range. 

With two people I trust looking at public SPARQ data that Whitman touts, the conclusion is that the data gives you a basic idea of how the private rating works, but it's not a perfect fit. I plan to follow up with Whitman this year to see if he can provide additional information that might not be included in his article that can help the two statisticians either replicate his results or further confirm that the data is overfitted.

This is not some major controversy and it doesn't not speak poorly of Whitman. If you want a basic picture of this cutting-edge view, Whitman's work provides value and it's also entertaining and engaging. If you're seeking precision, there's reason to be skeptical. It's good that he put the work out there for people to see. 

What About West?

Now that we can at least get a crude, if not somewhat accurate, picture of West's athletic profile with the public version of SPARQ, let's keep in mind that great athletes are not great football players. Of the top 10 SPARQ runners from 2014's draft, 6 are still in the NFL, and 3 have shown starter potential:

  1. Jerick McKinnon
  2. Stephen Houston
  3. Torrance Hunt
  4. Lache Seastrunk
  5. Bishop Sankey
  6. Charcandrick West
  7. Tre Mason
  8. Dri Archer
  9. Fitzgerald Toussaint
  10. Charles Sims

The three potential starting-caliber talents on film are McKinnon, Mason and Sims. Sankey has always been a committee-level talent to me. Seastrunk isn't willing to work at the game like a professional despite his raw on-field talent. Archer may earn an expanded role somewhere in the NFL within the next 2-3 years.  

I haven't seen a ton of pro film on West. What struck me about his performance in relief of Charles on Sunday was his decisive, downhill mentality. He hits the crease or intended area without hesitation.

This has been Knile Davis' problem. Davis, another player that athlete centric crowd loved, had clear conceptual flaws that he has still not corrected and it not only cost him a real opportunity to create a running back rotation with Charles, but it cost him the backup role to West, who spent all of 2014 on the Chiefs' practice squad. 

West has burst and agility, but he's not at that effortless level that we've seen from Charles. West's efforts appeared labored when he was changing direction at high speed and he wasn't earning much after contact. West, like Charles, is a smaller back (205 pounds). Unlike Charles, it's clear that West is still thinking on the field more than he's reacting. 

Charles has been great because he has mastered the ability to integrate the playbook, his feel for the run scheme, and what he sees on the field with his footwork. He has excellent vision. West isn't there yet and he appears slower--even if his approximate SPARQ rating is among the best of the 2014 rookie backs. 

Compounding matters is the Chiefs offense. The line has earned high-end "Mid-Tier" ratings from Footballguy Matt Bitonti through Week 4, but they entered the season as the best of the "Low-Tier" units. Bitonti noted that the changes to the lineup balanced the line and gave them a mid-tier value. 

Although a unit's grade is independent of a running back, a great runner maximizes the unit's efforts and a subpar or inexperienced back can ruin a good thing. I expect West to get more comfortable in the staring role and that added feel for his teammates should tap into more of his speed, quickness, and change of direction. How much more comfortable is the bigger question.

The fact that the Chiefs are working out Ben Tate and Pierre Thomas tells me that Kansas City kept West and Davis as younger, cheaper options. But when a team begins looking at players who will demand higher veteran minimums, it means they plan to get their money's worth. West should get 1-2 weeks to audition for the permanent role and because any signing of Tate and Thomas occurs after the opening week, their contracts aren't guaranteed for the season, which means Kansas City won't feel it overspent on a veteran runner if West plays well and they opt to keep the more experienced guy on the bench. 

Based on what I've seen, West can run through an arm tackle and he can bend away from defenders with some quickness, but I haven't seen him turn potential losses into gains. Nor have I seen him show that combination of agility and balance to do come close to Charles' skills. 

He earned the backup role because he makes fewer mental mistakes reading the line of scrimmage than Davis, but it doesn't make him a player I'd blow my waiver wire position or salary to obtain. Unlike Thomas Rawls, a player I think will be reduced to a bit part within the next week or two, West lacks the quarterback talent to benefit as a runner. Rawls at least as a quarterback that will challenge the back end of a defense with his arm while also gashing the front end with his legs. Alex Smith doesn't loosen up defenses and it means that even if Kansas City's line has performed better than Seattle's, it also faces more challenging fronts because opponents want the Chiefs to throw. 

If you're desperate for a RB, can't trade for a more promising back, and you can get West at 35-40 percent of your waiver budget then it's worth a shot. I don't like recommending this high of a number because of the potential for a veteran to come into Kansas City and take the role within 2-4 weeks and I don't think West has matured into this physical skills yet. But if you need a back that will get touches, go for it. 

West does not get a ringing endorsement for those of you trying to find an upgrade to existing starters. If you're seeking a season-changing fantasy option, West is nothing more than a gap-filler until he proves that his increased workload will make him significantly more productive. 

I'd rather sit tight on Andre Ellington, LeGarrette Blount, David Johnson, stick with Thomas Rawls if you already have him, grab a cheap share of Ryan Mathews, or even wait another week or two on Christine Michael. These six backs have the downside of losing the starting job or never cracking the starting lineup, but they all possess the skills to become low-end RB1s for the rest of the fantasy season. I'm not convinced West is more than a low-end RB2, at best. 

Re-Draft: Week 5 Notes and Year-long Implications

The A.J. Green-Richard Sherman Duel and Fantasy implications: This was one of the more enjoyable one-on-one events of the week. Of the seven plays that I counted where Green and Sherman faced off in single coverage, Green won four and Sherman three. Green dominated the early going with sudden footwork and hand usage to beat Sherman outside before breaking off his stems sharp enough to leave the cornerback a step behind.

Green even used a technique to work under Sherman's arm on a third down route from the slot that turned around the corner and helped the receiver break free for a first down. It was a similar technique I profiled Steve Smith using on Leon Hall two weeks ago. 

As the game progressed, Sherman found a good leverage point to ride Green up the sideline on multiple routes or undercut breaks and earn position to defend targets. One of the best moments was a route where Green tried an outside release that Sherman handled and the receiver turned into Sherman and squared the corner as if he was setting up a run block before releasing further down field. It was a well-designed move by Green, but Sherman didn't buy it and when Green released upfield, Sherman was step-for-step until he broke underneath and cut off the target. 

Despite a near-even matchup, much of Green's production came against Sherman. Seattle's Cary Williams was targeted with greater success, including Marvin Jones Jr drawing two pass interference penalties against the veteran and Green spanking the corner for a 72-yard touchdown nullified by an Andrew Whitworth holding foul.

As I mentioned on last Sunday's Audible pre-game, the Seahawks struggle against ultra-quick wide receivers. Green, Jones, Emmanuel Sanders, Randall Cobb, Odell Beckham, and Julian Edelman all had WR1-caliber fantasy outings against this crew during the past 12-18 months. John Brown, Antonio Brown, Steve Smith, and Travis Benjamin are all decent candidates to surprise the fantasy community as quality matchups against the vaunted Seahawks' secondary.

Giovani Bernard, a Better Fit in 2015: Jeremy Hill thrived in an I-formation set last year. He did the same at LSU. Give Hill a lead fullback and he'll stay focused on working off that blocker, which generally keeps him moving downhill and using his size and strength to earn yards. Now that the Bengals are at full strength with its skill positions, the offense is opting for more one-back sets and Hill is struggling, bouncing too many runs outside and getting met in the backfield .  

Single-back and shotgun sets are a better environment for the quick, shifty, and more disciplined Bernard. While lacking Hill's power, I have always maintained that Bernard's skill set and physical traits place him on a spectrum between Brian Westbrook and Ray Rice

Bernard will rarely dominate from the I-formation, but he can do competent work from there. He thrives when his offense has weapons to spread the field and he can crease emptier boxes up the middle.

What makes Bernard more viable this year is the Bengals' tendency to flood the short and intermediate zone to one side with three players: a receiver, a tight end, and a back. Cincinnati repeatedly forced Kam Chancellor into difficult decisions with alignments where Tyler Eifert and Marvin Jones Jr were tight to one side of the formation and Bernard worked from the backfield into the same zone. 

Eifert scored twice in the red zone from this look, once forcing Chancellor into a bad decision and earning a post-touchdown rebuke from safety Earl Thomas for missing the assignment. The second time, Eifert ran by Chancellor for a score in a similar area, but the coverage responsibilities were different. 

The Bengals varied the route combinations to test Chancellor at every turn. Sometimes Bernard broke inside and Eifert outside with Jones running past. Other times, Bernard broke outside, Jones to the inside, and Eifert up the seam. Bernard's skill in this aspect of the game and his short-area quickness as a ball carrier makes him the more attractive option with this offensive game plan. 

Josh McCown-Travis Benjamin Can't Close the Deal: The quarterback-receiver relationship comes down to a simple idea of two people with a potential opportunity to establish a mutual connection. In this case, it has some similarities to dating. To take this analogy further, when I watched Browns quarterback Josh McCown try to establish an on-field connection with Travis Benjamin Sunday, there were moments where it appeared more like this...


McCown may be the more polished quarterback in all the aspects of the game that Johnny Manziel is still trying to develop, but as I've written earlier this year, Manziel improvises better and he has a better downfield game. McCown is flat-out awkward in these areas. 

Overall, McCown seems to view Benjamin's talent the way an awkward, unconfident guy views someone he finds attractive--a mixture of unhealthy fear and desire that creates uncomfortable moments. 

Benjamin got open on a deep post on a route combo where Benjamin clearly got open before McCown even decided to check-down, but the quarterback ignored the opportunity. The quarterback then undershot Benjamin on a 2nd and 5 throw to the opposite hash that the receiver still nearly caught and then led Benjamin too far into traffic in the middle of the field. Later, McCown threw the ball blind on a comeback that got undercut for an interception with 10:45 left in the game. 

The two are connecting better than the people in this video, but Benjamin is routinely beating press coverage and getting deep and McCown isn't closing the deal. If Johnny Manziel earns time down the stretch, trade for Benjamin. 

Marcus Mariota and Scouting Lag: The Titans quarterback's stats against the Bills were 21-32 for 187 yards and an interception. Rex Ryan's defense may have a tough reputation, but Mariota was the first quarterback this year who didn't throw at least two touchdowns against Buffalo.

To be fair to Mariota, the Titans only threw the ball about 40 percent of the time compared to previous opponents, but 32 attempts is still a full day for an NFL quarterback. It's his first bad statistical outing.

With a schedule that includes Miami, Atlanta, Houston, New Orleans, Carolina, Jacksonville (twice), and Oakland during the next 8 weeks, the Titans face some reasonably soft pass defenses, but at half of them are known to get pressure on the quarterback. Buffalo had success--often fast enough without blitzing--that it limited the Titans' offense. 

The Bills also told the media after the game that they knew the Titans offense well enough that they had success anticipating many plays. This comes from advanced scouting that every NFL team does. The notable thing about this process is that teams tend to get--or gain confidence in the reliability of--scouting after the first 4-5 weeks of the season. 

It's this time of the year where opponents begin to implement game plans based on offensive performances from this year. This is the point of the season where I'm looking forward to seeing how Mariota responds to teams that will address some of his favorite plays. Will he adjust and remain productive, or will it cause the rookie to stumble? 

If Mariota mostly thrives through the Carolina game, there should be reason for optimism about his future. Either way, the Texans, Patriots, and Jets aren't a trio of teams I'd want my fantasy starter drawing during the playoffs unless he's a stud.