Philip Maymin, “An open-sourced optical tracking and advanced eSports analytics platform for League of Legends”, presented at the 12th MIT Sloan Sports Analytics Conference, Boston, MA, 2018. [PDF]

eSports games should have lots of advanced data at their disposal, but the reality is that they don’t.  In fact, most of the statistics available in games are primitive summary statistics. Maymin and his team sought to change that by developing a player tracking system that is live, remote, and non-intrusive, which allows for the collection of an extremely rich data set and more sophisticated metrics than currently possible on an eSports platform. Most importantly, this work has been put out to open source.

[Winner of the 2018 MIT Sloan Sports Analytics Conference Research Paper competition.]

This is the start of what I hope will be a mini-series within my Paper Discussions series — interviews with winners of the MIT SSAC Research Paper competition about their paper and their research. First up is the 2018 winner, Philip Maymin. Prof. Maymin has a unique and frankly intimidating resumé: bachelor’s, Masters, and PhD degrees in Computer Science, Applied Mathematics, and Finance from Harvard University and University of Chicago, a licensed attorney (in California), a journalist, a computer programmer, a portfolio manager of prestigious hedge funds (including his own), and an Associate Professor of Finance at the University of Bridgeport after a successful stint at NYU. He’s also an editor of two journals, one of which is the Journal of Sports Analytics, the Chief Analytics Officer for Vantage Sports, and a consultant for multiple NBA teams. Whew!

Prof. Maymin’s research broke new ground in sports analytics, especially in an emerging area such as eSports. His talk at the conference was a tour de force, and if you got everything that he was saying, congratulations! Far from being intimidating, he is actually quite personable, and he graciously accepted my invitation to an interview. I asked him some questions about his work and the fallout from winning such a prestigious award. The transcript is below, and my questions are in bold.

What motivated you to pursue this research and write this paper?

The time has come! Esports will soon dominate traditional sports. But the analytics for eSports has lagged a bit.

Does an eSports analytics community exist like those in traditional sports?

Far more informally.

You wrote in the first sentence of your paper that “…eSports should have vast quantities of data, but sometimes they do not.” I admit to not knowing a lot about eSports, but that seems surprising to me. Why has that been the case?

The game publishers control the data and they do not always want to reveal all of it.

Which eSport game did you choose to focus on and why?

League of Legends, because it is one of if not the most popular multiplayer game in the world.

So many times we hear of analytics being a data collection problem when it’s not a data analysis problem. How did you attack both problems? And did you have to get far enough on one area before you could address the other?

We were able to work on both tracks simultaneously, and improve the analytics as we got more data.

In your paper you formulated a notion of “worthless deaths” and “smart kills”. Could you explain those terms and how they might improve our understanding of player performance?

A worthless death is one that does not give you any net advantage. A “worth” death is one that helped your team overall; for example, maybe you killed two of your opponents before dying yourself. A smart kill is one that gives your team a net advantage. A dumb kill for example is just chasing someone around and eventually killing them, but you wasted so much time that it didn’t end up helping your team.

One thing that I find very interesting is that with eSports it’s possible to collect data on thousands, or even hundreds of thousands, of games. Do you foresee more advances in modeling of expected outcomes or contributions now that you have data on player actions and interactions?

Yes, that was one of our main results.

I was most excited to learn that you plan on releasing your software to open source. Was that your plan from the beginning? What has the response been from the eSports community?

People are very excited that we are doing this, and we hope it helps propel eSports analytics to new heights.

What has the public response been like after winning the Research Paper contest?

A lot of interest from professional eSports teams as well as traditional sports teams, and a lot of inquiries from players and eSports enthusiasts. People want to improve but it is hard to track and measure, so they’re excited there now seems to be a way to do it.

What are your future plans with this project?

Probably extend to other eSports as well.

[END]

You can watch Prof. Maymin’s presentation below.