Admission: I wouldn’t describe myself as an individual who gets terribly excited about draft prognosticating.
However, I do like comparative analytics. I like data and analysis and statistics and all that good stuff. So much so I spend my days doing it for good money. One thing I’ve learned in my career bridging the IT and Business gap is that you can make numbers say whatever you want. They’re putty, and if you’re a crafted artisan you can shape them into all kinds of great things.
I could care less whether the Jaguars draft Geno Smith or Dion Jordan or Bob Barker with the second pick of the 2013 NFL draft. I’ll watch. I’ll drink beer while I watch. I’ll text my friends after the Jaguars pick and we’ll exchange lively but ultimately uninformed banter regarding the quality of said pick.
I do, however, care about the data and sundry methodologies available to compare players to each other.
Confession: I read Big Cat Country all the time. And I have something like a man crush on Alfie. I like his wit. I respect his reliance, where possible, on numbers and such. He likes beer, and I homebrew. I’ve only recently actually signed up for an account, and what prompted me to do so was sifting through the myriad exchanges (recently, mostly) on Geno vs. Barkley. So I thought I’d take a look at the numbers and see whether, just maybe, I could get some putty and start to shape something.
I’ll organize this somewhat stream-of-consciousness so you can follow the journey my mind took while crafting this small, but potentially telling, bit of analysis. I welcome feedback on the metrics and approach. I’m happy to modify or extend it. I’m new to football metrics. Be kind friends.
Thought #1: How do I normalize outcomes of two distinct-but-related collections of data? (checks algorithm running on a box at work, starts digging up data via ESPN stats). Or, how do I compare two quarterbacks playing against different levels of competition, in different offenses, in different environments. I immediately start freaking out about confounding variables. I calm down. I have to start somewhere.
I start with attempts. In 2012 Geno had 518 of them and Barkley had 387 of them. Hmm…looks like Geno threw it more than Barkley. That’s good to know, we can use that later.
I go to completion percentage. In 2012 Geno completed 71% of his passes and Barkley completed 63%. Geno threw it more than Barkley and completed, comparatively, a higher number of said attempts.
So I can somewhat safely deduce from these two pieces of data that Geno threw more in his offense than Barkely did in his. I’d have to grab possessions data to really round this out but let’s just work off of this assumption for now. After all we’d want to know if they both had the same opportunity to throw (this is choice bias for some of you geeks). It’s ok, not a big deal.
What we think we know so far: 1) Geno threw it more, and 2) Geno completed more throws.
Thought #2: What if Geno is completing shorter passes because his offense is geared around guys running out three yards and turning around? Ok, I’ll need to know if they were throwing deep balls (arguably more difficult to complete) or short stuff (arguably easier to complete).
I go get YPA. In 2012 Geno had 8.12 YPA and Barkley had 8.46. Advantage Barkley, but only just barely.
But wait. Wait. Wait. Wait. We’re missing something. We don’t know if these guys are playing against the same level of competition. Everyone in Jacksonville tells me the SEC is better than other conferences so I figure I better try to smooth out the competition difference.
I go get Attempts and YPA against smoothed out talent. All I have to rely on is comparables against AP ranked talent. Not a pristine method but the best I have available.
I go get Attempts against AP ranked competition. Geno throws it 183 times with a 72% completion rate against ranked opponents and Barkley tosses it 133 times for a 56% completion rate. Holy wow. Geno threw it more than Barkely against AP ranked opponents while completing passes at a substantially higher clip than Barkley.
I go get YPA against AP ranked competition. Geno snags a 9.07 YPA against AP ranked opponents and Barkley logs a 7.81. Wait a damn second.
You mean to tell me that against AP ranked opponents Geno actually increased his YPA and completion percentage against his blended (AP ranked vs. non-AP ranked) season stats while Barkley dropped against his blended season stats in both areas? That stands out to me.
I also checked touchdowns and Geno threw twice as many touchdown passes against AP ranked opponents than Barkley.
So, for those of you who care, if you put me in a position to draft Geno or Barkley I’d tell you I don’t draft people because I’m not a GM. I’d tell you that the numbers might just be telling us a little something about which of these two QBs is the more accurate passer.
When answering the question of who’s more accurate: Geno or Barkley? You may just have your answer…
Thoughts on the method? Should we look at other outcomes?
*Source: Pulled splits via ESPN stats. I used the interweb. The Google, specifically.