Tim ‘Magic’ Sevenhuysen is the Founder of Oracle’s Elixir and one of the world’s foremost League of Legends statisticians. Many have called for those involved in esports to use stats in a similar manner to the way they’re utilised in traditional sports such as baseball, basketball and increasingly football too.
Perhaps esports could have its own Oakland Athletics and Billy Beane, or maybe the established teams need to step up first and make better use of available data to make superior decisions on rosters and which tactics to employ against different opposition. Magic gave us his thoughts and was also kind enough to make some Spring Split predictions to boot.
Esports Insider: Why did you launch the site, who is the data for and why did you decide to make 2016’s match data files downloadable for public use?
Tim: I created Oracle’s Elixir for two main reasons. One, I saw a niche in the LoL esports ecosystem that wasn’t being too thoroughly filled. Two, my interest in LoL esports in general was growing, and I figured collecting and writing about stats would be a fun way to further embrace my hobby.
At first I wasn’t too concerned about who the audience might be for my site. I was mostly focused on satisfying my own curiosity and my desire to create things. It became clear pretty quickly that writers and team analysts were excited about what I was doing, along with hardcore fans more generally, so I’ve tried to keep in touch with these groups and ensure that they’re finding value in both the numbers I report and what I write about.
I’ve been thinking about sharing some of my data publicly for a long time. First though I had to think about how much to share, and what the implications were. I eventually decided that since Oracle’s Elixir has mostly become a source of information for writers and analysts, then providing data for those audiences fits within my mission. I’m excited to see what people can produce with this data, and hopefully they’ll share their work publicly and generate lots of discussion and idea-sharing!
“In my opinion, every team should try to have an analyst on staff who knows how to interpret and make use of performance statistics”
Esports Insider: How deep can you go with the data analysis of players? What are the differences in terms of assessing individual players and teams, and which offers a greater challenge?
Tim: There’s a lot we can learn about players and teams from statistical analysis, but there are also some key limitations. At the most basic level, stats help us test our existing analysis against another source of information. For instance if we think a player is good in the laning phase, we can look at their CS difference at 10 minutes and see whether the numbers agree with us.
On a deeper level, we can measure a team’s control of various objectives, like towers or dragons or jungle camps. We can check a player’s ward output and check it against what we see whilst watching the games. We can check farm distribution and damage output. If any of those numbers tells us something unexpected, it forces us to rethink our opinions and maybe rewatch some games to try to reconcile our perceptions with the numbers.
“I’m sure teams look at stats while building their rosters, but most final roster choices are based on trying players out in scrims and getting a closer feel for a player that way”
Generally speaking it’s easier to use stats to assess teams than players. Team performance can be ultimately reduced down to wins and losses, and we can use stats to measure what they are doing to achieve those wins. Compared to team stats, it’s easier for player stats to be influenced by their teammates’ performances. It’s also possible for a player to perform well in a loss or perform poorly in a win, and the stats don’t necessarily make that easy to see.
Esports Insider: To your knowledge and in your opinion, are major orgs in esports making good use of this type of data available when preparing for tournaments and making roster changes in League and other titles? Should all teams have a data analyst on staff?
Tim: There are definitely teams using statistics, mostly for their game preparation while scouting upcoming opponents. I’m sure teams look at stats while building their rosters, but most final roster choices are based on trying players out in scrims and getting a closer feel for a player that way. I wouldn’t advise teams to sign a player based on stats alone, without giving them a proper tryout first.
In my opinion, every team should try to have an analyst on staff who knows how to interpret and make use of performance statistics. But it may be not necessary to have someone who can gather stats direct from the source and create big databases to work with. If an organisation can afford someone with those technical skills then great, but for organisations with smaller budgets I’m hoping that the data I’ve released goes part of the way to giving their analysts something to work with.
“For now, I see Cloud9 as slight favorites to take the spring championship”
Esports Insider: Are you League’s Billy Beane and could we see a Moneyball situation in 2017 NA LCS. Are certain elements of LoL gameplay analysis being overlooked or are some being given too much emphasis?
Tim: I wouldn’t claim that title for myself, but I do like to think that my work over the past two years has brought some greater stats awareness into the LoL esports scene.
I wouldn’t necessarily pinpoint specific statistics as being underused in LoL analysis right now, but I think different teams approach stats is very different ways, without much consensus on the best practices. With more people working on LoL stats over time, and more people having access to good data, I hope we’ll start to find some agreed-upon approaches to build on.
Esports Insider: Care to make a Spring Split winner prediction and who will be the MVP? We see you’ve put Cloud9 and TSM in your tier 1 in the power rankings…
Tim: A rematch of the TSM vs. Cloud9 finals is a pretty good possibility, though there are four or five other teams who could legitimately threaten to climb to the top.
For now, I see Cloud9 as slight favorites to take the spring championship. TSM’s Bjergsen is a very good bet to repeat as regular season MVP, especially now that he won’t be sharing spotlight with Doublelift.
Those interested in Oracle’s Elixir League of Legends data can download the available packages here