Home /  News & Events / Extension Blog / Moneyball, fantasy football and predictive analytics

Moneyball, fantasy football and predictive analytics

By David Washburn

You may not know it, but if you play fantasy football these days you’re likely benefiting from the burgeoning field of predictive analytics. And it’s also touching just about every other aspect of your life.
It doesn’t get any more old-school than the JDFL – the fantasy football league run by legendary San Diego Union-Tribune sports reporter Bill Center that over the years has included a who’s who of local sports media personalities.
Founded in 1983, the JDFL is one of the oldest leagues in the country, and Center rules it with an iron fist. If you’re late to the JDFL draft and miss the first round, then you lose your first-round pick. Same goes if you’re in the bathroom when it’s your turn to pick in subsequent rounds.
Each week during the season you have to pick up your phone and call your lineup in to Center. If you forget to make the call, too bad. At the end of play each week, he tabulates the results by hand and blasts them out in an email, which is the only time he uses a computer.
“We still use the same old formula,” said Center, who retired from the UT in 2014 and now works for the San Diego Padres baseball team. “We’ve refused to join the 21st Century, or whatever century we’re in.”
The JDFL, to use a statistical term, is an outlier.
Fantasy football leagues have exploded in popularity over the past couple decades and, along with all the other fantasy sports, are now conducted almost entirely online. Players have instant access to their teams and can make lineup changes, look up stats and get real-time scoring by clicking a mouse or touching a screen.
Even the annual draft – the high holy day of fantasy football – no longer needs to be done with everyone in the same room. Sites like ESPN or CBS Sports have snazzy interfaces that do all the work that Center does and much more.
And if you’re late to the draft or have to step away when it’s your turn to pick, the programs have you covered. They will pick the “best player available” by instantly analyzing a slew of ratings and statistics on every player in the NFL.
The technical term for all of this is predictive analytics, which has not only transformed fantasy sports, but also real sports and many other aspects of our lives. Using machine learning and data mining, statisticians can make predictions on everything from what we’ll buy online, to the shows we’ll watch on TV and how we’ll vote.
In the sports world, it’s best known as “Moneyball,” a term coined by the author Michael Lewis in his book about Oakland Athletics General Manager Billy Beane, the first sports executive to embrace predictive analytics for player evaluation.
The Athletics made it to the playoffs in four straight years during the early 2000s, despite having one of the smallest payrolls in Major League Baseball. The success was largely attributed to Beane’s embrace of sabermetrics, a method invented by baseball historian and statistician Bill James that challenged the conventional wisdom of player evaluation.
Historically, player evaluation was the domain of the team’s scouting departments, which consisted mainly of crusty old guys who’d spent the majority of their lives playing, coaching and watching baseball. These guys relied almost exclusively on their experience and intuition to pick players. When they did use statistics, their favored measure was batting average, which represents the number of hits a batter gets per x number of at-bats.
Beane took player evaluation away from the scouts and gave it to the stat nerds, who, by using sabermetrics, determined that statistics like on-base percentage or slugging percentage (which is the percentage of a batter’s hits that go for extra bases) were better predictors of a player’s impact on a game than batting average.
So, for several years the Athletics were able to get players who were undervalued by other teams at bargain prices, and also win on trades involving players that other teams overvalued.
But as is almost always the case with innovation, the pack eventually caught up with Beane and it was the Boston Red Sox, not the Athletics, who were the first to cash in on sabermetrics. In 2004, the Red Sox constructed a team using sabermetrics and ended the 86-year-old “Curse of the Bambino” with a World Series victory.
Fast-forward to today and every team in pro sports employs statisticians who use predictive analytics in player evaluation and development. As a result, newspapers and sports websites are filled with new stats like WAR, which stands for “wins above replacement,” and OPS, which melds on-base percentage and slugging percentage into one stat.
A byproduct of this new era is that sports teams are demanding more out of their scouts and looking beyond athletic departments and into math departments to make hires. One person very much aware of this new reality is Ash Pahwa, who teaches a class called “Sports Predictive Analytics” for UC San Diego Extension.
The class delves into the history of predictive analytics, including the Moneyball story, and covers the building blocks of the discipline, which include calculus, linear algebra and regression analysis. The students are a mix of mathematicians interested in a career in sports along with some employees of team scouting departments who need to learn the math.
“All the professional sports teams are looking for data scientists -- I had one student who was with the Angels [Los Angeles Angels baseball team] and one who works for the Lakers [Los Angeles Lakers basketball team],” said Pahwa, who is a tech entrepreneur whose most recent company is Irvine-based A+ Web Services, which provides internet marketing and web analytics services.
“If you are a good mathematician and you take my course there is a good chance you could get a job with a sports team somewhere.”
And if a sports team doesn’t hire you, don’t fret Pahwa said, there are plenty of other opportunities. “Can you tell me any company that doesn’t use a calculator? You can’t,” he said. “The same thing is happening with predictive analytics, we won’t be able to live without it.”
Pahwa takes his students back to the 1700s when Carl Gauss, who is considered the greatest mathematician to ever live, developed the theory that eventually led to the field of regression analysis, which is the basis of predictive analytics. But it wasn’t until the 1970s, when computer chips became fast enough to crunch large data sets in a reasonable amount of time, that the discipline went from theory to practice.
“In the early days, it would take five days to complete the regression analysis between two points,” Pahwa said. “Now we can do it in a blink of an eye.”
Pahwa is a self-described sports nut, but doesn’t play fantasy football because he eschews gambling. If he did play, he probably wouldn’t like the JDFL – he’d be happier playing with Leonard LaPadula.
LaPadula is the founder of Advanced Sports Logic, a New Hampshire-based company that markets an online product that uses predictive analytics and advanced probability for all aspects of fantasy football – from drafting to setting lineups and picking up players during the season.
A computer chip designer who designed the first GPS chip, LaPadula first started playing fantasy football in 2003. He wasn’t much of a football fan and it showed in his team’s performance during his first few years playing – he regularly finished near the bottom of the league.
But things started to turn around in 2009 when he used his computer chops and love of math to come up with a program that created probability distributions for all the players in the league. That year he used the program to make all of his drafting and lineup decisions.
The results spoke for themselves – LaPadula won six straight games to begin the season and made it to the playoffs. By 2011, he was marketing his product to the wider fantasy community and attracting a devoted following among the more serious players.
“We found that people using our program were winning three times as much as the average player,” said LaPadula, who put the program on hiatus this year while he develops a new interface.
However, even LaPadula and Pahwa will tell you that not even the most advanced predictive analytics can guarantee victory. Consider that in 2009, the first year LaPadula made the playoffs using his program, he entered the first round of the playoffs with a 73 percent chance of winning.
He lost.
“At the end of the day the data can only take you so far, you still need intuition,” Pahwa said. “The classic example is David and Goliath. If you had the data on Goliath, you’d have bet on Goliath to win.”

If you're interested in seeing if predictive analytics is your new passion, check out our Data Science programs and courses, or contact the program manager at unex-techdata@ucsd.edu.

Blog post currently doesn't have any comments.

About Extension

UC San Diego Extension is recognized nationally and internationally for linking the public to expert professionals and the knowledge resources of the University of California.