clock menu more-arrow no yes

Filed under:

Alabama Football Recruiting: Another look at SPARQ

New, 23 comments

Recruiting isn’t an exact science, but that’s never stopped me before

NCAA Football: Cotton Bowl-Michigan State vs Alabama
Derrick Henry is a poster child for SPARQ metrics
Jerome Miron-USA TODAY Sports

Sometimes, I get derided around these parts for an over-obsession on SPARQ. If you’re new to the site since last February, you may have been confused if you saw Roger Myers taking pot shots at me in the comment section of a random article, particularly if cornerback Nigel Knott was involved in the discussion (more on that later).

So, what is this SPARQ?

It goes back to the early 2000s when Pete Carroll was at USC. I won’t go into details— mostly because I don’t know them— but he worked with Nike and some other prominent names in coaching, recruiting, NFL scouting, and analytics to help create a five-variable formula based on a series of common athletic tests (like the ever-popular 40-yard dash) the come up with a single number that they though best represented a player’s “explosiveness”. In other words, it isn’t just talking about who can run the fastest 40-yard dash, but factors in the 20-yard shuttle, vertical jump, kneeling powerball toss, and the player’s weight as well.

Nike integrated a combine for high school recruits into their “The Opening” summer camps, and we started getting public and standardized results for the tests around 2011.

One of the best and easiest examples is former Tide running back Derrick Henry. When talking about measureables, most people go to the 40-yard dash first. Derrick ran a 4.52 forty, which is pretty good but nothing special. But if I told you he ran that fast while at a hefty 243 pounds, those of you that have followed football for any length of time would agree that’s pretty impressive, even if “impressive” wouldn’t really show up on a spreadsheet. Big and fast is a lot scarier than little and fast.

Well, Henry had a 41-inch vertical jump, which is better than nearly every prospect at the NFL combine each year, and also had really good marks on the shuttle and powerball. All said, he ended up with 141.87. The number itself is arbitrary, unless you know the context around what is a good SPARQ and what isn’t (rule of thumb, though, higher is better).

Around 2013, a guy I followed from the Seattle Seahawks side of sports blogging, Zach Whitman, pulled some crazy stats wizardry and managed to back calculate the actual formula and then used and modified it for NFL draft prospects. It was his work that got me interested and also gave me the idea of normalizing SPARQ scores by position group. That way you can get the sense of how any individual player, say a linebacker, compares athletically to the rest of the linebackers in D-1 college football.

So, get your math caps on, and I’ll do some layman’s explaining. Feel free to email or Tweet at me if you want deeper explanations.

A common term in stats is a standard deviation. It’s basically a number that represents where a subset of a population falls within a bell curve. The idea is that 68% of a population falls between -1 and 1 standard deviation, with 0 being the middle mark of the curve.

When assigning the number of standard deviations of a specific data point (in this case, a player), the term Z-score is often used, and the one I usually go with. So, if I say that a player has a 0.5 Z-score, then it means he is half of a standard deviation above average. And if we assume that half of all players are below average (by definition they are), then it means that he is more athletic than about 70% of everyone at his position in college football.

If you’re visually inclined, here’s a good couple of graphs showing how Z-scores work with percentages.

So back to Derrick Henry. His 141.87 SPARQ score is a really high number. You know that. But could you have told me just how high it was if you didn’t know what was a good score? What if I told you his Z-score was 2.36? Going back to our percentiles, which is easier to understand, Henry is more athletic than 99% of all running backs.

When comparing across positions, you have to remember what the number actually means. If I say that Emil Ekiyor has a 2.36 Z-score, compared to a 2.05 Z-score from Cameron Latu, it doesn’t mean that Ekiyor is necessarily more athletic or explosive (he’s not, Latu has 120 SPARQ, compared to 102 from Ekiyor). But it does mean that Ekiyor is further above the average lineman than Latu is above the average linebacker. Make sense?

2018 Recruiting Data

Last Name First Name Position State Town Stars National Rank Position Rank Height Weight 40-Yard Dash 20-Yard Shuttle Vertical Jump Power Throw SPARQ Z-Score
Last Name First Name Position State Town Stars National Rank Position Rank Height Weight 40-Yard Dash 20-Yard Shuttle Vertical Jump Power Throw SPARQ Z-Score
Ekiyor Emil OC IN Indianapolis 4 113 3 6030 339 5.18 4.78 26 50 101.97 2.36
Latu Cameron WDE UT Salt Lake City 4 131 7 6050 236 4.82 4.38 36.7 45.5 120.69 2.05

Now, as I mentioned at the start, my focus on SPARQ over the years has become a bit of a running joke around here, and many like to point at highly athletic players that didn’t pan out as anecdotal evidence of why you should just “stick the eye test” and “hustle beats athleticism” and other common platitudes.

Well, yes and no. Both ways of looking at things are accurate to a degree. But neither are the whole picture without looking at the other. A player that’s extremely athletic but has little football skill or attitude will flame out. A less athletic player can be very successful if he can maximize the right skills and master positioning and reacting.

But man, it sure helps when you can just flat outrun the other guy.

In an effort to nullify some of the naysayers, I turned once again to the numbers. Because numbers don’t lie*.

*but context sure does

I asked Erik, Josh, and CB969 to each give a rating to every single Alabama recruit with a SPARQ number since 2013. It was subjective, but we were each rating on a scale of 1-10 on how successful that player’s career had been (be it at Alabama, another school, or the pros). I took the average out of the four of us to hopefully eliminate some of the subjectivity involved, and then plotted career success vs. both SPARQ and Z-score.

The data wasn’t as conclusive as I had hoped. While there was a definite positive trend in both cases, it was still very scattered. In both cases, we hovered between a 0.1 and 0.15 r-squared value (definition and explanation here), which isn’t great, but also isn’t terrible for subjective data with less than 100 points to work with.

In SPARQ particularly, you can see a very significant positive trend, but three outliers on the bottom right side of the graph with really high SPARQ scores wound up bringing the trendline further down. For some trivia, I’ll give bonus points and a #refund to whoever can correctly guess the three.

The Z-score plot actually didn’t fit a regression model as well as the SPARQ, even though I had hoped the normalizing aspect of using standard deviations would help to smooth it somewhat. That said, there was still a general positive trend, indicating that more athletic players are more likely to be successful, even if the chances of variation from that trend are very high.

Ultimately, SPARQ and athleticism are just one factor in the entire evaluation of a player. It isn’t everything. We have to watch for techniques, skills, playstyles, scheme fits, on-field attitude, and, of course, athleticism.

In the coming weeks, I’ll be breaking down what every single new freshman will be bringing to the Tide’s 2019 football squad, and I will be referencing SPARQ freely and liberally, so bookmark this link if you need.

In the meantime, enjoy our coverage of National Signing Day.