Decision MediaWorks

You Are Not A Savvy College Football Fan Until You Know This

Quick Note: After posting we found a data set that included the 2012-13 season and draft stats. When we updated with these numbers, FSU made a leap to a .55 Pearson’s r. Essentially, FSU was an outlier until last season. 

During the initial round of college football games last weekend, all three of Florida’s major programs won their season openers. As I was in the midst of devouring a salsa-covered chip, a friend announced, in sage-like fashion, that the Gators will “clean up” when the next NFL draft comes around.

Around here we question assumptions with data, and that got me to thinking… I began to wonder…

1) Does a team’s success have an effect on the draft prospects of its players?

2) Are some teams better at converting talented players (draft picks) into wins?

So, I brewed some coffee and dove into the data. I entered the number of draft picks for the Gators, Seminoles and Hurricanes each year in the last ten years and the number of wins in the previous season. I then calculated the Pearson’s r over those years for each team. The Pearson’s r is a convenient little statistical tool that tells us the strength of a relationship between two variables (in this case the number of players drafted and the number of wins). The Pearson’s r may be either positive or negative but will always fall between 0 and 1. In short, the smaller the number, the weaker relationship and the larger the number, the stronger the relationship.

What I found was nothing short of bizarre.

Over the last ten years, the Gators had a moderately strong relationship between wins and draft recruits. The Florida Pearson’s r is .52. The Miami Hurricanes aren’t much different, with a Pearson’s r of .57. This means that when these two teams perform better on the field, their players benefit by having better luck in the NFL draft. But the shocker is Florida State. Their Pearson’s r is .06. This is extremely weak and more or less non-existent. You can see the Pearson’s r comparisons in the graph below.

This lack of a relationship is baffling and means that if the Seminoles perform well, their seniors and juniors are no more likely to be drafted than if the team collapses into a pile of sweat and loses.

For a moment, I suspected this could be explained by the Seminoles’ reputation, so I decided to perform the same calculation on two other historically well performing football teams, the Alabama Crimson Tide and the Notre Dame Fighting Irish. This didn’t clear up anything. The Tide had a very strong relationship, with a Pearson’s r of .63. The Fighting Irish had a Pearson’s r of .41. So, the players benefit more from better seasons.

You can check all the data in the interactive table below. If you learned something interesting here today, please share it with your friends.

What Tailgate Town is Most Likely To Get You In A Fender Bender?

It is a pilgrimage for many in Florida. Fans heading back to their college towns for the opening game. The energy, the tailgating and of course, the traffic.

One of the most definitive rivalries in Florida is the one between UF and FSU. For those of you sporting “House Divided” décor we wanted to settle a score. What college town has the worst drivers?

We decided to look at the number of traffic accidents in Alachua and Leon counties and compare them to the state of Florida as a whole to find out if what I’m experiencing is something more than happenstance. The figures in the graph below are the ratio of accidents to population.

We found that in Alachua County, you are 25% more likely to be involved in a crash compared to the rest of Florida. But you are even more likely to be heading to the body shop after a visit to ‘Nole Country. In Leon county you are almost 35% more likely to be in an accident compared with the rest of Florida.

So, according to the data, both schools are excellent at producing terrible drivers, but FSU takes the top ranking.

This post was contributed by Alabama native @mrdanieldean…Roll Tide!

The Moneyball of You

This month’s issue of ESPN The Magazine features a flattering yet realistic description of the challenges faced by the Jacksonville Jaguars analytics team and its leader, Tony Khan.

The comparisons of my beloved Jags and the Oakland Athletics of Moneyball fame are many. Small market teams struggling to compete seek a competitive edge using an “unproven” method that contradicts the “gut” of old school types.

Yet, from a data science perspective, football is far different from baseball. The ESPN article says this:

“At the heart of baseball is a one-on-one battle — pitcher vs. batter — that allows for easy collection of clean, accurate, predictive data. In football, though, there are 22 moving parts on each play, along with an infinite number of variables, including score, field position and down and distance.”

In football it is difficult to collect data on all the possible variables that could be influencing the outcomes we are attempting to forecast. From a data perspective, football is messy, inconsistent and blurry but not unpredictable. Football is like life.

Football, like life, is about to change.

The football field of 2018 will look a lot the same…except for the sensors. Every player will have helmets that monitor impacts and brainwave activity. Shoulder pads will record body temperature and heart rate. Every player’s exact position and speed on the field will be recorded by GPS, and Google Glass-like recordings will be made on every play by every player on the field.

When the data becomes as dynamic as the game, a strange thing happens.

The Tony Khan’s of the world will no longer reside in the windowless bowels of NFL stadiums but will run game day war rooms alongside coordinators…the real-time data pouring in and being processed instantly. Khan will use this data to score potential plays and personnel packages based on their probability of success and deliver a menu of “best chance” plays to Jaguar coaches.

Think that sounds like science fiction? To see how ubiquitous sensors have become in our daily lives, I want you to do an experiment for me. Take your phone out of your pocket and try to count all its sensors:

Camera, spectrometer (light sensor), GPS, accelerometer, microphone, gyroscope, clock, WiFi.

All these sensors turn the seemingly random and chaotic activities of our lives into neat, structured data.

Authors Viktor Mayer-Schonberger and Kenneth Cukier call this trend of measuring our live “datatization.” Datatization means that you will have access to incredibly detailed information and analytics about you. Want to know what evening habits lead to your best nights of sleep? Want to know at what temperatures you are most likely to feel happy? Want to predict your likelihood of divorce? Its already possible and this is only the start.

The next generation of wearable computers will increase the amount of datatization a thousand-fold. Today’s sensors measure our outside world, and tomorrow’s will datatize our inner world.

Google Glass can see what you see. Biometric bracelets will report heart rate and body temperature. Our physiological and mental condition will be datatized and mashed up with data about our driving, web browsing, eating and shopping habits to reveal hidden insights into our lives.

It’s the Moneyball of you and you will be the Tony Khan of your own life.

*Awesome Data Cat cartoon from ESPN.com (http://a.espncdn.com/photo/2013/0819/mag_jags-illo01jr_400.jpg)

10-20-Life. Does it matter?

After we posted last week on the relationship between concealed carry permits and violent crime, Greg Newburn of Families Against Mandatory Minimums tweeted to ask about similar data on the policy of “10-20-Life.” You can learn more about this mandatory sentencing policy here.

We did some research and here is what we found:

- Easily accessible public data on Florida firearm crime is not available going back more than a decade. This is unfortunate because 10-20-Life became law in 1998. The bottom line is that we can’t follow the trend all the way back, but we can go back ten years.

- While violent crime in Florida has fallen dramatically over the last decade, gun crime actually spiked in the mid-2000s before falling again in the last few years. In turns out, the trend of using guns for violent crime is a bit disjointed from overall violent crime rates and from the number of concealed carry permits (i.e. more guns possessed by Floridians). Hence, there is no easy connection between overall violence and gun crime or the number of guns and gun crime.

- We can’t make any clear judgments about 10-20-Life without having data from before the policy was enacted. But, you can see the available data for yourself and form your own hypothesis. Ask yourself what trend the overall gun crime rate (purple line on the chart) mirrors. For example, our researcher @mrDanielDean thought gun crime seemed to rise and fall with the economy.

The chart below is interactive- just hover over points to see the data. As always, if you like what you see, please share it!