Guest columnist Zachary O. Binney fact-checks a story in a national publication and finds that everyone makes mistakes.
26 Apr 2012
Guest column by Nathan Forster
Welcome to the Mocktopus, our first attempt at a mock draft created through statistical analysis. The Mocktopus first estimates the probability that a given team "on the clock" will select a player from one of eleven position groups--quarterbacks, offensive tackles, running backs, wide receivers, tight ends, interior offensive linemen, interior defensive linemen, edge rushers (college defensive ends), linebackers who are not edge rushers, cornerbacks, and safeties. A separate model then estimates the probability that the team on the clock will select the first, second, third, or fourth "most valuable" player available at the position. Both of these estimates are conducted using a technique called logistic regression. Using software that I specifically designed to execute the models, the Mocktopus simulates the NFL Draft 10,000 times to create a mock draft.
The principal components in the Mocktopus are other mock drafts. Although mock drafts are frequently criticized for missing on their picks more often than they hit, they are nevertheless significantly more accurate than a chimpanzee pulling random names out of a hat. My research suggests that those who have had the most accurate mock drafts in the past can also be expected to produce above average mock drafts in the future, although their level of accuracy can be expected to drop somewhat. Thus, as part of its calculations, the Mocktopus assigns weights to mock drafts from the most successful draftniks based on historical accuracy, but it also assumes that the accuracy of their mock drafts will regress towards the mean in the future. The most historically accurate mock draft according to the model is Mike Mayock of NFL Network’s mock draft, and you will see his strong influence on the model.
The second component is a metric that measures value by combining the "Top 100" overall prospects lists from the Draft Countdown and NFL Draft Scout websites. Mock drafts tend to focus too much on team needs and not enough on prospect value, so including a metric for "value" helps control for a bias for pairing teams with low value players in order to address the teams' most critical needs. Moreover, when a team drafts a player at a position that few expect, it often does so at a position where good value is available (for instance, St. Louis' selection of Robert Quinn and Detroit's selection of Nick Fairley in the 2011 Draft). Interestingly, value was not a statistically significant factor as to quarterbacks and offensive tackles, which are positions highly-driven by team needs, and therefore the Mocktopus does not use value to predict whether a team will draft a quarterback or offensive tackle (although it does use value to help it predict which quarterback or offensive tackle will be drafted).
Finally, the Mocktopus also incorporates team performance variables that have been statistically significant predictors of player choice in past drafts. As to quarterbacks, the Mocktopus' predictor variables include the age of the team's projected starter (teams with old quarterbacks tend to draft new ones). As to wide receivers, the predictor variables include the collective yards per catch of the team's projected starting wide receivers (teams tend to draft wide receivers when they lack down-field threats on the roster). There are other team-based factors that barely missed the cut for inclusion in the Mocktopus because they approached, but were not, statistically significant (such as team sack rates for edge rushers) so there is definitely the potential to add additional team-based factors in the future.
Although this sounds silly to type, this project has also raised an oddly philosophical question about mock drafting: what is the purpose of a mock draft? I would imagine that most fans, when they read a mock draft, are most interested in the player that the mock draft projects for their favorite team.
However, traditional mock drafts are clearly not optimized for this purpose. An interesting discovery that I made while developing the Mocktopus is that creating a mock draft for the purpose of providing projections for thirty two separate fan bases almost invariably involves projecting the same player to multiple teams. This is a consequence of the fact that the probability that a team will select a particular player, given the team’s 100+ options, is typically well under 50 percent.
For instance, imagine that we think that the Cleveland Browns have a 35 percent chance of drafting Ryan Tannehill, a 25 percent chance of drafting Morris Claiborne, a 25 percent chance of drafting Trent Richardson, and a 15 percent chance of drafting Justin Blackmon. Now imagine that there is a 0 percent chance that any team between the Browns and the Dolphins would take Tannehill, but that the Dolphins had a 90 percent chance of selecting Tannehill if available. If our goal was to project the Browns’ and Dolphins’ most likely pick we would give them both Tannehill.
Because I was not so bold to throw internal logic to the wind, I designed the Mocktopus to eliminate the doubles and replace them with highly-rated players with uncertain destinations.
In making these decisions, the Mocktopus optimizes its mock draft to score as highly as possible under the scoring system popularized by the website the Huddle Report: two points for a player correctly matched to the team that drafted him and one point for correctly slotting a player in the first round.
I can only speculate on how accurate the mock drafts produced by the Mocktopus will be. Although there have been 223 first-round picks made in the time period we analyzed to create the Mocktopus (2005–2010), those sample sizes are a lot smaller when you drill down on a position by position basis (only eleven interior linemen drafted). The Mocktopus is also prone to human and computer error. The Mocktopus relies on a large amount of data that changes fast and often, and moreover, I’m a new hand at computer programming and have been snuffing out bugs in the program up to the last minute. So, the Mocktopus is far from a finished project—hence the beta moniker. Think of it as a fun experiment that we’re putting out a little early rather than keeping it under a rock.
No analysis necessary here. The Mocktopus would give Luck to the Colts anyway, but his selection is so certain that I locked him in.
Kalil appeared here in almost everybody’s mock draft—until yesterday, when drafntiks finally started to become convinced that the Vikings might take Claiborne. The Mocktopus still likes Kalil here.
There has been some late movement towards Justin Blackmon, but the Mocktopus prefers Richardson for this spot for a couple of reasons. First, the Mocktopus considers Richardson a better value than Blackmon. Second, mock drafters have traditionally had a much harder time projecting wide receivers than running backs, so the Mocktopus weighs the votes for Richardson a little heavier than the votes for Blackmon. A Blackmon selection, however, is entirely possible and would really shake up the top six, which until recently, looked almost set in stone.
The Mocktopus likes Claiborne here, but would strongly favor the Buccaneers to jump on Richardson if he falls.
The choice here seems to be between Blackmon and Claiborne. The Mocktopus thinks there is only about a six percent chance that Trent Richardson drops and the Rams take him. Also, Claiborne instantly becomes the favorite if the Browns take Blackmon.
Mock drafts have been mocking an edge rusher to the Jaguars for years but they haven’t taken one high since the Derrick Harvey experiment.
Aside from Luck and Griffin, Tannehill is the Mocktopus’s biggest lock of the draft by far. Why? First, the mock drafts are nearly unanimous. Second, there is a huge gap between Tannehill and the next most highly rated player of his position group, Kirk Cousins.
This is a four-man race between Fletcher Cox, Quinton Coples, Justin Blackmon, and David DeCastro (in that order). Fox, Coples, and Blackmon, however, each appear in a few more simulations for other teams, so DeCastro, a highly rated player who is hard to find a home for, goes here.
The Mocktopus thinks that Luke Kuechly, who it values as the seventh best player in this draft, would be a great value pick over the flashier pick, Michael Floyd and the dark horse, Mark Barron.
A few days ago Barron looked like a lock to go to the Cowboys at number 14, but some late movement in the mock drafts and an increase in his perceived value moves him up to the 11th pick. He edges Stanford guard David DeCastro by the slimmest of margins.
The conventional wisdom for this pick is that the Seahawks could use a pass-rushing threat on the other side of Chris Clemons, and that Pete Carroll is such a great motivator that he can get the best out of Coples.
This pick likely comes down to whether the Cardinals prefer Reiff or Cordy Glenn. Reiff gets a decent edge because he has an advantage in both the mock drafts and the value charts. A dark horse could be Mark Barron, who would be considered great value if he slipped.
The mock drafts still favor Barron for the Cowboys, but his value is so high he is unlikely to slip to the Cowboys. The Mocktopus gives the Cowboys an almost 75 percent chance to draft Barron if he is available, but he was only available in just under 25 percent of simulations.
There are rumors that the Eagles covet Cox so much that they could trade up to take him. Although the Mocktopus does not yet have the capacity to project trades, it nevertheless thinks that Cox has a strong chance to end up here because if others pass on him, the Eagles are a great candidate to snatch him up.
If Courtney Upshaw tops 18.0 sacks in his first two years, he and Jason Pierre-Paul can meet up at some fancy New York restaurant and drink champagne and go all 1972 Miami Dolphins at my expense.
The Bengals need a cornerback, and Stephon Gilmore’s late rise suggests that Kirkpatrick will be the best option here.
Since Larry English was a bust, the Chargers have to try again.
The late-rising Chandler Jones is likely to go in the first-round, and the Bears could use pass rush help, so it’s a decent enough fit.
There will probably be lots of good options here. The Mocktopus is pretty evenly split between Hill and the Titans breaking the fall of a number of players that are not generally expected to be here, like Stephon Gilmore or Melvin Ingram.
This is where chaos theory really takes hold and the Mocktopus starts spitting out low probability projections. Poe is the leader to go to the Bengals here and he does so in only 800 simulations out of 10,000.
Wright has been here since there was talk of the Browns taking Robert Griffin III at fourth overall. Wide receivers are hard to project, but the Browns are as likely to take a wideout as anyone.
Did somebody let Matt Millen into the building? Nah. The Lions are just hard to mock and Floyd is hard to place because he keeps coming out on the losing end of narrow battles for spots in the upper-to-mid first round. Glenn may actually be the most likely option for the Lions here, but he shows up on the next team in more of our simulations.
The Mocktopus wants to give the Steelers Dont’a Hightower or one of the tackles.
The Mocktopus likes Worthy to go to the Broncos because, for some reason, the mock drafts have formed a consensus around Worthy and the Broncos despite the logjam of interior defensive linemen in this draft.
The Mocktopus likes Fleener to the Texans more than any other pick at the end of the first round. Why? Mock drafts historically rack up points for where they project tight ends—and some good ones have Fleener going to Houston—while Rueben Randle, the other favorite of the mock drafters, would not be good value here.
This pick could work because Perry can bring the heat off the edge and has the size that the Patriots covet in their outside linebackers.
There is a solid mock draft consensus consolidating around McClellin to Green Bay. Despite the cluster of similarly valued edge rushers at the end of round one, and all of the craziness that can happen between picks 3 and 27, the Mocktopus puts the chances of McClellin going here at a healthy 37.3 percent.
The collection of 3-4 teams at the end of round one gives Hightower a great chance of hearing his name called on Thursday night.
This is a close race between Zeitler and Amini Silatolu.
This would be a great move for the defensive back-starved Patriots, but unfortunately, Gilmore is unlikely to last this long. The Mocktopus just thinks that the Patriots are so unpredictable that it’s better off placing someone who is almost certain to go in the first round somewhere. Given the Patriots’ drafting history, that’s probably a smart move.
This is a pretty big slide for a player who has been hyped as having the athleticism to play left tackle. The Mocktopus likes the Giants to break his fall at a pretty high clip—he appeared here in 23.2 percent of simulations.
59 comments, Last at 03 May 2012, 11:02am by chemical burn