Prior to free agency, I used a chaos theory-influenced draft simulation to demonstrate how accurately predicting the NFL Draft is an impossible task. Admittedly, it was not chaotic enough, nor was it robust. Thanks so much to reader feedback from multiple forums, I modified the simulation by… well… by adding more chaos. The result, I think, is a better illustration of the chaos inherent within the NFL Draft.
The NFL draft is a unique event, drawing comparisons often to war. Decisions are made in the War Room and team executives launch at each other counter measures and counter intelligence to ensure no one other than themselves understands the plan of attack. Andy Reid, he of draft day successes and failures, once said this of the NFL Draft, “Best friends lie to best friends. It’s like the Civil War.” So, as you can imagine, when 32 enemies enter the field of battle simultaneously, successes and victories are difficult to measure, let alone predict. Good thing prediction is fun.
My initial simulation model was based on a statistical coin flip: to draft based on team need or best player available (BPA). That still holds true, however, drafting by team need is more dynamic in this new model, as is drafting by best player available for Kansas City Chief’s first pick (remember, Tabor defines a chaotic system as one whose outcomes are very sensitive to initial conditions). The new model is also more robust. Instead of five simulation runs, there are 1,000. Doing this allowed me to better illustrate the high amount of variation from one simulation to the next. It also allowed me to do some cool summarizing of results.
In a way, these new simulations are not incomparable to parallel universes – each simulation could be possible and likely to occur if not in our universe, then in another. Or, we can view them through the “Copenhagen Interpretation”, that is, each simulation is an event that occurs simultaneously with the others. It’s just those with higher probabilities occur more often. So it’s possible then, that one of these thousand simulations could occur in our universe. Or, one of the infinite possibilities not included here will occur. Or… maybe I watch a little too much Fringe and Big Bang Theory.
At any rate, the primary decision for teams in these simulations remains to draft by need or best player available. This is determined by comparing two random numbers. If one is greater than another, then teams draft BPA, else by need. The decision to draft BPA versus need across all simulations averages 50%, with some individual simulations ranging from 40% to 60% in favor of either BPA or need.
For Kansas City’s first round draft choice, a best player available selection is a random decision among the top four BPA prospects (I chose four arbitrarily). It seems oddly fitting that Andy Reid will launch the draft’s first grenade in his first year not in Philadelphia, the shock wave of which will impact the Philadelphia Eagles, even three picks later. After Kansas City’s first pick, teams that draft BPA do so based on updated prospect rankings from CBSSports.com, accounting for those prospects already drafted by need (I wanted to randomize this but could not figure out a way to reconcile the best player available selections against those drafted by need without repeating prospects already drafted).
Teams that draft by need, on the other hand, do so based on another random decision: which need? Need is determined by randomizing need-priority within a two-to-three unit range (based on updated and modified need rankings from Walterfootball.com). For example, if the Eagles pick by need in the first round, they will do so randomly among their top two positions of need.
Speaking of our Philadelphia Eagles (finally), how did they fare? According to 1,000 simulation runs, and based on the simulation rules defined above, the Eagles have nine primary options with their first round pick and have close to a 60% chance of drafting an offensive player in the first round. Assuming Andy Reid and Chip Kelly are not best friends, Geno Smith is the favorite, though not overwhelmingly, drafted 184 times in one thousand:
The second round is even more chaotic. With more choices by teams before them, the variation in the Eagles’ picks grows. Texas defensive end Alex Okafor is drafted most, followed by Syracuse quarterback Ryan Nassib:
Even though I think this model better illustrates the chaos of the NFL draft, there are obvious limitations to these simulations. The glaring one is the lack of trades. Unfortunately I didn’t have the time to incorporate them into this model, but perhaps next year. I would also have liked to run this for 10,000 iterations, pushing Excel beyond its limit, but running this simulation with a macro took thirty minutes. Increasing tenfold would take three hundred minutes, patience for which I do not have. Also on my wish list for the future: a BPA list for each team and randomizing BPA picks. However, doing this would be outside the scope of this article, which is to illustrate how chaotic and complex predicting the draft really is, no matter which universe you live in, nor which friend you believe.
Below are some additional highlights of the simulations. For more detail, download the Excel file and use the interactive pivot tables to see what you’d like, including full draft results by individual simulation and individual team, and complete average draft position (ADP) rankings. Also, I welcome feedback, so leave comments below.
Here is how the Kansas City, Jacksonville Jaguars, and Oakland Raiders draft picks look in the first round:
Here are the average draft positions of the first 32 picks, along with min and max ranges:
And here are possibilities for Indianapolis’ Mr. Irrelevant. Go Khalid Wooten:
Tags: 2013 NFL Draft, Alex Okafor, Chaos Theory, Geno Smith, Jacksonville Jaguars, Kansas City Chiefs, NFL, NFL Draft, NFL Mock Draft, Oakland Raiders, Philadelphia Eagles, Ryan Nassib