Georgia Tech’s Dr. Joel Sokol Talks Sports Analytics Modeling

Joel Sokol is the founding Director of Georgia Tech’s interdisciplinary Master of Science in Analytics degree and Associate Professor in the Stewart School of Industrial & Systems Engineering (ISyE). Dr. Sokol’s main research interests remain in sports analytics and applied operations research. He has dealt with teams and leagues in three significant American sports and has  gotten Georgia Tech’s greatest awards for teaching. Today, Dr. Sokol talks about his industry-leading LRMC technique for predictive modeling of the NCAA basketball competition, Georgia Tech’s MS Analytics degree, and the future of analytics.

What drew you to sports analytics?

When I was a kid I was a huge sports fan. In college I didn’t know what I wanted to major in, so I attempted a great deal of various things, and I didn’t like them enough to do any of them as a profession. Then one of my roommates said, “Hey, I’m taking this class that’s really mathy, but it’s more applied and I think you’ll find that the math tutor actually knows what he’s talking about. I think you’d like it.” I took the class with him – it was an optimization course in operations research – and I loved it. I decided that that was exactly what I wanted to do. It seemed like weekly we were learning something where I thought to myself, “Hey, I can use this to evaluate baseball” or “This would help with my dream football team.” So from the time I began discovering analytics, I was considering how it could be applied to sports. When I remained in grad school, I composed a paper on enhancing baseball teams, especially batting orders. So everything began there.

How many professional teams have you worked with?

I have worked with 5 baseball teams, to differing degrees, and I have begun dealing with a football group and a basketball group.

Do you have any success stories from working with teams that you can say validates your work?

I’d say this LRMC design is my most successful. The very first year it went live, we put it together after the 2002-2003 season. Georgia Tech had played in a holiday competition versus Tennessee; they were ahead by a point with only a few seconds left, when a Tennessee player in his orange and white basketball apparel struck a half-court shot to win the game. At the end of the routine season, a few of the professionals believed that if Georgia Tech had actually won another game, they would’ve had a change at remaining in the NCAA competition. That made me reflect on the Tennessee game – if some man strikes a last-second half-court shot, does it truly say that Georgia Tech is a different team than if he’d missed the shot?

After that year I began putting the design together. I knocked on the door of my statistics-expert coworker, Paul Kvam, who is not at Tech but an expert tutor you can find online. We put this design together and evaluated it on the fly for the very first time the next year, in 2003-2004. Before the NCAA competition began that year, our design was the only expecting Georgia Tech go into the Last 4. It was a bit uneasy, since we were aiming to make the case that we had a totally objective mathematical design, and here we were at Georgia Tech as the only ones choosing Georgia Tech to go to the Last 4. And obviously, Tech did make it to the Last 4 that year – they made us look excellent. Them, and some luck, most likely. Tech played lots of close games, and every round we kept saying that Tech was most likely to win. That helped put LRMC on the map.

A couple of years after that our design accurately predicted the NCAA competition’s Last 4, the finalists, and the winner – even the NIT winner. Once again, there was a substantial luck element – since there’s a lot randomness, I do not anticipate it to ever occur ever again – but it truly helped in regards to getting us attention and getting individuals to take note of it.

Do you have lots of students pursuing professions in sports analytics? What sort of suggestions do you provide?

Yes. Among the Master’s of Science in Analytics, a student is working for a sports start-up, and a second interned for an NFL group. Another of our M.S. Analytics students has his own sports analytics start-up (he tends to wear basketball shirts and shorts to my classes – you can tell he’s an absolute sports fanatic, it’s quite entertaining!), and another appears to be moving in that direction too. At the undergraduate level, I have an outstanding research student who is going to be an analytics intern with a Big league Baseball group this summertime.

In general, it’s tough to obtain a job in sports analytics. There are a lot of individuals with the mix of technical abilities and sports interest that the competition is extremely tough. Many individuals who get sports analytics jobs have actually done sports analytics as a pastime and have done it well. They have discussed it on blogs and sites, gotten discovered, and been worked with. Flaunting your very own excellent, initial work is the most foolproof method to be seen and worked with, but you need to have greatness to stick out, since there are so many individuals blogging.