Introducing SPORK

tl;dr Here are the coefficients to calculate SPORK, a close approximation of pSPARQ (r2 .959), which itself is an approximation of SPARQ.

It’s turtles all the way down.

(Intercept) 181.924227
HT           -4.592111
WT            0.361056
X40         -15.159259
X10         -45.532307
Shuttle     -18.731278
X3Cone       -5.985102
Bench         0.252199
Vert          1.327290
Broad         9.557272

Here is a link to a Google doc applying the formula.

Sports analytics nerds are a funny lot. We obsess over things like mean squared errors and trying find year over year correlations for the beard growth on Julian Edleman. Our curiosity drives us to do strange things in the pursuit of knowledge.

It has always been this way. Hackers with the same drive reverse engineered arcade machines and built emulators so they could play Dig Dug on their home computer. Then they posted the source code for free for all to see and, more importantly, to contribute to.

Game recognizes game in the nerd community. When you bust your butt to figure something out and then share your results with the world, you earn street cred and admiration.

Which was why when Zach Whitman back-calculated Nike’s proprietary SPARQ metric (rumored to be used by the Seattle Seahawks for player evaluation) and published them on the Seahawks fan blog Fieldgulls, I was suitably impressed and excited.

Here was another in a long line of curious nerds that figured out the answer to something interesting. Moreover, the answer he discovered was being kept hidden from us. Indeed it was the fact the Nike took their SPARQ calculator off the the internets that prompted Zach to begin working on modeling it. He didn’t like not knowing. Information, as they say, wants to be free.

Then a funny thing happened. After freeing the SPARQ formula from the clutches of a multi-national corporation bent on secrecy, our intrepid hero…kept it for himself! Didn’t see that one coming.

Still, Zach posted an google spreadsheet with all his results for nerds to add to their personal databases. Which was something. I guess. Then this year he launched his own blog and began writing articles for Rotoworld. Awesome. But shockingly, in both of these venues Zach made the decision to post his data… in pictures. Pictures! PNGs, if we’re being precise.

Why does this matter? Well, data presented this way must be hand entered into another nerd’s database. That’s just plain rude. Makes Hulk want to smash.

On his blog he says he’s trying to “figure out a way” to present the data in a tabular format. Really? So, we’re to believe that a nerd that can back-calculate SPARQ can’t learn enough html to post a table of data? I call shenanigans.

No, this is blatantly transparent douchbaggery. In what can only be considered some vain attempt to build his personal brand, Zach is impeding the dissemination and reuse of information. This is the exact opposite of the hacker ethos.

Compare Zach’s actions to Marcus Armstrong’s. Marcus wanted to replicate Madden scores for real NFL players. So, he built a site that does just that. It’s awesome. Even better, Mockdraftable gives away its data, which is listed in tables for easy copy and paste. Nerdgasm.

Or compare Zach’s actions to those of your humble blogger. When I hand collected injury data for MLB players and cross referenced it all with retroIDs for easy analysis, I released it to the world. Baseball Prospectus incorporated it into PECOTA. David Phillips at Georgetown used it for his paper in the Journal of Quantitative Analysis in Sports. Beyond the Boxscore looked at wrist injuries and power. In short, knowledge was created.

This is how things can be when nerds don’t act like d-bags.

So, I took 10 minutes today and ran a regression on pSPARQ. I provided the coefficients at the top of this post. It’s not a perfect approximation, but it’s very close. It’s certainly good enough to tell you if a player is an elite athlete or not, which is the entire goal. Add it to your spreadsheets, run some analysis using Marcus’ combine data. Compare 125 SPORK score athlete’s AV to 100 SPORK score athlete’s AV and see what shakes out.

Create some new knowledge, and share it with the rest of us nerds. Maybe one of us will buy you a beer.

 
78
Kudos
 
78
Kudos

Now read this

Projecting Running Backs using Age Adjusted Correlation Curves and Historical Comps

tl;dr Fantasy players who draft based on efficiency will be consistent league losers. A player’s per game production in year Y correlates to Y+1 production to varying degrees depending on the player’s age. During a player’s prime years,... Continue →