Finding The All-SPARQ: Impact Defensive Ends In The 2017 NFL Draft

Back in 2015, we introduced Cowboys Nation to the concept of SPARQ on these pages. SPARQ turned out to be a concept that resonated powerfully with many fans. A metric that was previously virtually unknown outside of the most hardcore advanced-stat-aficionados (and a bunch of Seahawks fans) now has Cowboys social media all a-buzz and a-twitter.

Over the next days and weeks, we'll look at many different position groups in this year's draft class via SPARQ, so today I'll provide an extended (re-)introduction to SPARQ before we move on to this year's defensive end class.

SPARQ

Many elite athletes in the various college programs find that once they enter the NFL, their previously elite skill set is - at best - par for the course on an NFL team. As a matter of principle, NFL players are bigger, faster, stronger, and more talented than college players.

Which is why NFL teams are obsessed with athleticism over almost anything else, and which is why we as fans pore of 40-yard dash times and other Combine measurables so much. You can teach most players to recognize when a defense is in man or zone, but you cannot teach a player to outrun a faster opponent.

A little over a decade ago, Nike developed a metric called SPARQ. The idea behind SPARQ was to have a composite metric that would allow you to quickly assess the athleticism of a player with a single number. Think of it as an SAT score for Football Players. This "SAT" score, or SPARQ rating, does not trump the evaluation of game tape, a person's character and competitiveness, interviews with coaches, and medicals. It is just another tool for coaches to use, but it does encapsulate one simple truth about the NFL:

Given the same level of talent, the bigger/faster/stronger players almost always win.

And that's where SPARQ comes in. The SPARQ metric is calculated using eight inputs. There is no height or arm length component involved, but SPARQ blends an athlete's weight, explosive power, speed and agility into one metric.

(1) Player Weight: this "normalizes" the score, giving credit to a heavier player who displays similar movement skills to a smaller, quicker player.

(2) Explosive power bench press, broad jump, vertical jump

(3) Speed and agility: forty-yard dash, ten-yard split, short shuttle and 3-cone drill.

Unfortunately, Nike never published the exact formula for the SPARQ metric. But an enterprising blogger, Zach Whitman, reverse-engineered an approximation of the formula, and while he doesn't divulge the formula either, at least he publishes the results of his calculations at 3sigmaathlete.com.

Here's Whitman explaining how SPARQ can be used.

What’s the use of SPARQ? What we see often in pre-draft analysis is an over-emphasis on the forty-yard dash, for which there are two main reasons: (1) speed is important, and (2) we’re familiar with the common forty benchmarks. A 4.4s 40 is fast and sounds good, and there’s an inherent understanding of what it means.

The problem is that the forty-yard time isn’t fully indicative of a player’s overall athleticism. Most people don’t know off-hand what a good broad jump is for a wide receiver, and even fewer are aware of what they should expect from a defensive end. SPARQ is a way to standardize these different parameters and gain a more circumspect view of a player’s natural ability. [...]

SPARQ isn’t perfect. Player test results have error and, even if they were perfect, don’t fully represent the ability of an athlete. The goal here isn’t to build an airplane. SPARQ is just a method by which we can better understand players, and it’s important to not let perfect be the enemy of good.

So what, if any, correlation does SPARQ metric have with actual NFL production? Here's a chart courtesy of Zach Whitman at 3sigmaathlete.com explaining that exact correlation.

The chart uses Approximate Value (read up on that metric here) as a measure of NFL production and the SPARQ score as a measure of athleticism. SPARQ here is expressed as a player’s ranking relative to his peers at his position (a 0 z-score is average, a 2.0 is two standard deviations above the peer average). Whitman explains the rest:

What we see is that there’s a clear trend toward more athletic players producing a higher AV3. If there was no relationship between athleticism and production, this line would be flat, parallel to the x-axis (i.e., zero slope). This relationship is statistically significant with a p-value of approximately zero.

2017 Defensive Ends Class

Whitman publishes all the pSPARQ numbers on his website, and the following table summarizes his SPARQ data and adds the Production Ratio for this year's defensive ends (click on the blue column headers to sort):

SPARQ & Production Ratio, 2017
POS	CBS Rank	Player	School	Proj. Round	Ht	Wt	pSPARQ	z-score	NFL Perc	Prod. Ratio
DE	2	Solomon Thomas	Stanford	1	6-3	273	1397	1.5	93.4	1.37
DE	8	Derek Barnett	Tennessee	1	6-3	259	110.8	-0.7	25.4	2.10
DE	24	Taco Charlton	Michigan	1	6-6	277	121.6	0.2	56.0	2.14
DE	32	Charles Harris	Missouri	1-2	6-3	253	118.2	-0.1	45.9	1.94
DE	34	Takkarist McKinley	UCLA	1-2	6-2	250	113.8	-0.4	33.1	1.67
DE	37	Tyus Bowser	Houston	1-2	6-3	247	137.8	1.4	91.3	1.47
DE	40	Carl Lawson	Auburn	1-2	6-2	261	124.2	0.3	63.3	1.33
OLB	42	T.J. Watt	Wisconsin	2	6-4	252	139.7	1.5	93.4	1.06
DE	56	Tarell Basham	Ohio	2	6-4	269	121.9	0.2	56.9	1.58
DE	64	Jordan Willis	Kansas State	2	6-4	255	140.5	1.6	94.2	2.06
DE	74	Derek Rivers	Youngstown State	2-3	6-4	248	130.9	0.8	80.1	2.13
DE	103	Trey Hendrickson	Florida Atlantic	3	6-4	266	134.7	1.1	87.1	2.17
DE	121	Deatrich Wise Jr.	Arkansas	3-4	6-5	274	122..8	0.2	59.3	1.06
DE	124	Daeshon Hall	Texas A&M	3-4	6-5	266	129.3	0.7	76.7	1.50
DE	158	Tanoh Kpassagnon	Villanova	4-5	6-7	289	118.0	0.5	69.1	2.11
DE	161	Dawuane Smoot	Illinois	4-5	6-3	264	110.8	-0.7	25.4	1.79
DE	188	Hunter Dimick	Utah	5-6	6-3	274	122.6	0.2	58.9	2.30
DE	204	Josh Carraway	TCU	6	6-3	242	109.9	-0.7	23.4	1.52
DE	241	Bryan Cox	Florida	7	6-3	265	92.4	-2.0	2.1	0.68
DE	258	Garrett Sickels	Penn State	7-FA	6-3	261	107.5	-0.9	18.2	1.04
DE	270	Darius English	South Carolina	7-FA	6-5	237	11.9.0	0.0	48.1	1.35

Whitman's data is now updated to include pro day results. A few other notes on the data:

• pSPARQ is the single metric designed to summarize a player's athleticism.

• z-score calculates a player’s ranking relative to his peers at his position. A z-score of 0 means a player is average, while a 2.0 means he’s two standard deviations above the peer average.

• NFL perc. is the z-score translated into percentiles. A 50.0 percentile would represent a player who rates as a league-average NFL athlete at the position.

• Production Ratio shows the number of sacks and tackles for loss per game over a player's last two college seasons. A number above 1.5 is often indicative of premier production for a pass rusher. Production ratios marked in yellow indicate a player is from a small school, and that his high production ratio is at least in part the result of playing against inferior competition.
  

Going by the pSPARQ score, the top defensive ends in this year's draft class are Jordan Willis, Soloman Thomas, and T.J. Watt (though he's likely to show up as an outside linebacker on the Cowboys' board, and not as a DE)

For comparison, 2015 Cowboys draft picks Randy Gregory and Ryan Russell had pSPARQ scores of 132.8 and 122.3 respectively, 2016 pick Charles Taper had a pSPARQ of 133.7, all of which should give you an idea of what type of athleticism the Cowboys are looking for in their pass rushers. As a further point of reference, some of the better pass rushers to enter the league in recent years like J.J. Watt, DeMarcus Ware, Jadeveon Clowney, Justin Houston, or Cameron Jordan all scored above 140.

So now we know who the superior athletes in this defensive ends class are. But by itself, that won't help us all that much. After all, the history of the NFL draft is littered with superior athletes who never made it in the NFL.

Back in January this year, we looked at the college production of the defensive ends in the 2017 draft class. To do that, we used a metric called the 'Production Ratio' that adds up sacks and tackles-for-loss and divides the sum by the number of college games played. The resulting ratio is one tool among many - albeit a pretty good one - that measures the playmaking potential of front five players coming out of college.

If we combine the two metrics, SPARQ and the Production Ratio, we should be able to find the most productive AND the most athletic DEs in this draft. The graph below plots the Production Ratio against the SPARQ score for the 21 DEs from the table above.

The two red lines divide the graph into above average and below average performers. Players with a Production Ratio of 1.5 or more (the top two quadrants, "A" and "C") delivered an above average production in their last two college seasons. Players with a SPARQ score of more than 120 (the two quadrants on the right, "A" and "B") are above average athletes relative to their NFL peers.

The A quadrant (top right) is where you shoud find the players most likely to succeed at the NFL level. They have a strong track record of production and have the pre-requisite athleticism that should allow them to compete at the NFL level. Six defensive ends from this year's draft class populate this quadrant, which makes this a solid DE draft class.

The B quadrant (bottom right) shows superior athletes whose college production has been below average. And while this doesn't automatically invalidate them as potential prospects, it does raise questions. Teams need to understand why these guys didn't have the kind of production other players, often with inferior athleticism, did. Was it the scheme they played in, the players they played next to, the opponents they played against, the role they were asked to play, or are they simply not very good football players?

The numbers here won't answer those questions, but those are questions teams will have to answer satisfactorily via film study, player interviews, coaching interviews, or other means.

In this quadrant, T.J. Watt probably projects more as an off-the-line linebacker; Daeshon Hall has good athleticism, but weak production for a guy who should hove had tons of space playing opposite Myles Garrett in college; Solomon Thomas projects more as a DT in the NFL, and his numbers will look much more impressive when compared to DTs rather than DEs.

The C quadrant (top left) features players with a strong record of production at the college level, but who have questions regarding their athletic ability. Again, being in this quadrant is not necessarily a bad thing - Demarcus Lawrence for example was a C Quadrant player (113.8 SPARQ, 2.28 Production ratio). However, if you don't have the athleticism to compete at the next level, you're going to struggle mightily - regardless of your college production. It's just an extra question teams will have to answer.

Two edge rushers most frequently associated with the Cowboys (Charles Harris, Takkarist McKinley) both find themselves in this quadrant, even if they are pushing towards the A-quadrant.

[Note: the data and graph have been updated with pro day results since this article was originally published]

It is worth noting that the Combine data the pSPARQ values are based on is just a single snapshot of a player's athletic performance. If any or all of these guys simply had bad days at the Combine, they could easily improve on their numbers at their Pro Days or at private team workouts and subsequently move into the 'A' quadrant.

The benefit of the Combine is that it allows teams to collect data from prospects performing in the same environment and under the same circumstances. But players can have an off day, some players prepare better for the event, some players are still recovering from injuries; in short, the athletic markers that go into the SPARQ calculation can differ depending on where and when they are measured.

last year for example, Joey Bosa reportedly improved his 40 time from 4.86 at Combine to 4.78 at the OSU pro day, improved from 24 to 28 reps in the bench press, and added an inch to his 10-foot Combine broad jump. That performance would easily give him a higher SPARQ score.

Also important, especially for tall players like Tanoh Kpassagnon (6-7), Taco Charlton (6-6), or Daeshon Hall (6-5), is that SPARQ does not take into account any size or length measurables other than weight. Size and length can be important for football players, yet they are missing in SPARQ, and if a key component of a particular player's game is his height or arm length, SPARQ will not reflect that.

The D quadrant (bottom left) is a tough one to be in. Below average production and below average athleticism don't promise a great future in the NFL, but once more, you need to understand each individual case before closing the book on a prospect.

With the limited data we have at our disposal, we can only provide the snapshot we see in the chart above. But with the understanding that the data points for each player may not be quite as rigid as the chart suggests, our snapshot nevertheless provides a good starting point from which to discuss these players.

The mandatory caveat: There are a multitude of factors that determine how well a prospect will do in the NFL. College production and athletic markers are just some of them, but at the very least, they provide some interesting input into the evaluation process.

Given these numbers, and given what you know about these prospects, in which rounds would you be looking for a defensive end, and which one would it be?

As an addendum, here are some historical numbers for defensive ends as a comparison as well as the charts for the 2015 and 2016 draft classes a little farther down. It's important to note that while great pass rushers seem to cluster in the A-quadrant, there are great players in the B and C quadrant as well, even if they are not quite as plentiful.