One of the enjoyable parts of this gig is the ability to get nearly instant feedback on my work. Readers will leave comments, hit me up on Twitter or Facebook, or send me an email. Even when I wade through the 95 percent of feedback advising me to either retire or die (or both), there’s a few thoughtful readers out there who provide some interesting and/or helpful suggestions.
After publishing my article on Consistency Rankings (CR), many of you wrote in to ask a simple question: how to best use the concept of consistency rankings in conjunction with scoring?
Of course, I’ve taken a stab at answering that question in The Best Damn Draft Theory — but that was on a positional level. How can we apply it to individual players? Turns out, that’s not a simple question.
My first inclination was to try and answer it using a clustering technique, such as k-means clustering. However, I ran into two problems: first, I’m not sure that there’s enough differentiation between players’ performance to yield good results, and secondly, this is the formula for defining k-means clusters:
So after a couple of hours of tinkering (I might still figure out the whole clustering thing, but a deadline is a deadline), I decided to take a different approach: I took every player from last year whose CR was greater than zero, looked at their per-game output, and wrote a back-of-the-napkin formula to rank them within their own positional group.
The results seemed promising, so I decided to publish them for you here. But as payment for my hard work, I’m going to ask you to give me a name for this new consistency/performance ranking. Leave your suggestion on our Facebook page, Twitter feed or send me an email firstname.lastname@example.org.
On to the player rankings …
It’s a bit eye-opening to see guys like Blake Bortles and Colin Kaepernick ranked above guys like Philip Rivers and Ben Roethlisberger, but keep in mind what this ranking is trying to put across: guys who not only performed well, but who also did so consistently.