NFL Draft 2014 SPARQ Profiling, Part 10: Database

If you missed earlier entries in the series, check out the SPARQ storystream.

I'd like to thank Tony Wiltshire, who contributed all data for this analysis. Without him, it would've been impossible to assemble and run SPARQ for every eligible player.

Z-Score

SPARQ articles have combined for around 25,000 words over the last five weeks, so there's really not much ground to cover there. If you want to know about how any specific Seattle position look statistically, that's already available; however, we hadn't yet looked at how positions compare to each other. Because SPARQ will generally be lower for an offensive lineman than a running back, you can't really compare Russell Okung to Christine Michael.

Here enters the z-score. As discussed in previous articles, the z-score simply represents the number of standard deviations from the mean. This calculation requires a mean and standard deviation, so it's important to figure out what exactly we define as an average NFL athlete.

Using the excellent 2013 Wiltshire database, I calculated all 2013 SPARQs. Retrospective data allows us to place players in groups: drafted, undrafted players who make it to a camp, and players who fall out of the league entirely. I'll probably go into depth in this study later in the summer, when news is slower, but the general result is pretty interesting: drafted players are about a standard deviation above undrafted players, who are about a standard deviation above players who fall out of the league entirely. It's not just Seattle who values SPARQ -- athletic players tend to be drafted.

I decided that the "average NFL athlete" is represented by everyone who's in an NFL training camp. The "rSPARQ z-score" and "pSPARQ z-score" statistics are thus based on this "average training camp" athleticism. I ultimately decided against defining the average based on a sample of drafted players; I feel that the training camp numbers are more stable year-to-year, and there aren't a large number of players drafted at each position each spring, so it's harder to obtain reliable statistics.

The result of this is that we can relate a player's SPARQ to their global athleticism percentile, and then put all athletes in the same doc and sort by this new normalized parameter. It's just a way to get a feel for SPARQ without needing to remember the range for a specific position.

If you're interested in turning SPARQ into percentiles, look at this handy PDF courtesy of Johnny Football University. Basically, a 1.0 z-score is an upper 16% NFL athlete, 2.0 is upper 2.5%, and 3.0 is upper 0.15%.

This implies that if there are roughly 2000 players in the NFL, there are around 50 players with at least a 2.0 z-score.

Database(s)

There are two SPARQ databases now available on google docs: one has all players currently on the Seattle roster and most high-profile players acquired during the PCJS era. The other contains all athleticism values for the 1746 players in the Wiltshire 2014 database. His doc actually lists more than 1746, but I've only included those who have the requisite tests to calculate SPARQ.

2014 NFL DRAFT SPARQ DATABASE: GOOGLE DOC

EXISTING SEATTLE ROSTER SPARQ DATABASE: GOOGLE DOC

Wrapping Up

Every number that can be run is now run. Note that there are a lot of players available in this release who weren't covered in the initial positional pieces. The most athletic player in the class is Marshall OT Garrett Scott, who wasn't covered previously, and there are plenty of others who fit this criteria as well.

When looking for a Seattle athlete in the tables, film is obviously a huge component. SPARQ is just a tool, but I think it's a good way of making the sample you're judging a lot smaller. A Seahawk will almost always be above-average in SPARQ, preferably tending toward a 1.0 z-score. There are certainly other physical factors to consider besides the SPARQ calculation, like 32" arm length for cornerbacks, a requirement which we discussed a few weeks ago and Dan Quinn confirmed today.

I'd like to re-emphasize that I don't believe SPARQ is the be-all, end-all. If it was, Jameson Konz would be a star. However, the correlation of draft picks to SPARQ is too high to dismiss. There's value here, it's just a matter of combining objective analytics with subjective scouting ability, personality evaluation, and common sense. No piece of the puzzle is everything, and no piece of the puzzle is worthless.

I plan to live-update the SPARQ table as the draft goes along, and will be transferring players over to the Seattle roster doc as they're drafted. If you're interested in live SPARQ analysis, I'll be tweeting throughout the draft about new Seahawks and how they fit into the roster.

Two days left.