In October, the Toronto Defiant, an expansion franchise in the one-year-old Overwatch League, hired Dennis “Barroi” Matz. Matz, 25, is the founder of Winston’s Lab, a popular website for Overwatch stats and analytics that broke down player and team performance through an advanced video-tracking program. The program that powered the site uses image recognition via machine learning to scan match footage from Overwatch, a futuristic first-person shooter, and transmute the action into a log of in-game events. The log is then run through a script that spits out an array of metrics not found in official league statistics, including which in-game characters are used most frequently and how those characters' special powers contribute to an overall team victory, to name only a few. For teams looking for any competitive edge in a league that features millions in prize money, the hope of the Defiant is that Matz’s alchemy can turn those numbers into gold.
Matz, who studied computer science in college and lives in Aachen in his native Germany, was focused on artificial intelligence and planned to move to Japan to continue the pursuit. But in 2016, he was intrigued when Overwatch debuted and Blizzard Entertainment wanted to build an esports league around it. It was around that time when he noticed what he believed to be a crucial absence, something that has helped popularize other sports and rooted the debates that surround them in objectivity.
“When the first tournaments started to pop up and the whole scene began to grow, there was no place to get statistics," Matz said. "You were not able to look up who played which hero or even a team’s win rate on a certain map-type.”
Eventually, Matz got tired of waiting. He used Tim Sevenhuysen’s League of Legends site, OraclesElixir.com, as his model for Winston’s Lab but wanted to go even further. “I wanted to create something similar, but I wanted a platform that is used by top analysts and the public alike," he said.
Matz believes analytics of the sort he tracks have the power to alter the game’s prevalent strategies, just as they have in traditional sports, pointing to the spike in three-point attempts by NBA teams in the modern era.
“In basketball, the conclusion, which was drawn out of analytics, … changed the dynamics of the game,” he said. And such a change may have already occurred in the Overwatch League.
“When I found out how incredibly valuable the first kill of a fight is, teams began playing more cautiously and conceding fights earlier, resulting in a smaller loss of time, thus having a higher chance to win the map,” he said.
Julien “daemoN” Ducros, coach of the OWL’s expansion franchise in Paris, said analytics let coaches quickly diagnose issues and help the team improve.
“The stats are so important to me to quickly understand the problem that you just need it," Ducros said.
If a team is relying too heavily on one tactic, the data can present a solution for a counter, just as baseball teams will overload one side of second base against a batter who hits the ball to that half of the field more often than not.
“My job is to identify those weaknesses and exploit them, or if it’s about our own team, fix them," Matz said. "I also will help to come up with new team compositions and attack routes … essentially a playbook similar to what you have in football.”
Matz’s PTK model, which focuses on the different player roles within a team, could have a huge effect on lineups and how future OWL teams are built. “The overall trend … is that [damage per second] players — players whose job is to put out as much damage as possible — have a way-smaller impact on success than is generally perceived," he said. "Championship teams are normally those who have more reserved DPS players and tanks or supports that make the big, flashy plays.”
Those are the sort of insights the Defiant hope Matz can provide them in their inaugural season, and why they’re moving him to Los Angeles to study and train the team as it competes in a season that begins Feb. 14 and will be broken up across four stages of five weeks each. To that end, the success of the Defiant could go a long way toward validating the role of advanced analytics in esports, just as the rise of the low-budget Oakland Athletics helped popularize the stat-driven principles of “Moneyball” in Major League Baseball.
“The analytics arms race is coming,” said Sabina Hemmi, a pioneer of esports analytics who co-founded Dotabuff and Overbuff, analytics sites dedicated to Dota 2 and Overwatch. Hemmi said the success of her sites has led to multiple teams and organizations asking her to provide them with scouting tools to identify the next esports stars.
While teams continue to embrace analytics, Blizzard itself is already serving up a series of data points from each match, as well as highlighting for them the performances of players in the Overwatch League’s “Path to Pro” circuit, essentially the OWL’s minor league. Ben Trautman, the Overwatch League’s official statistics producer, isn’t concerned that all of the statistical data could lead to players exploiting video games, which at their core, are powered by advanced equations.
“We have a very strong relationship with our players and they’ll help verify the integrity of the game to avoid demonstrating exploits to the general public,” Trautman said. In his mind, the more data the better.
“Data ultimately is created by players, player performances are not created by data,” Trautman said. “Data helps in strategy and game planning, but as we’ve seen throughout the inaugural season of the Overwatch League, there’s a certain ‘X factor’ that some players bring to the table and do amazing and surprising things that exceed anyone’s expectations.”
Even with all the data in the world, analysts such as Matz are still left with a big challenge when relaying their findings, be it to players, coaches or fans: They must make it digestible.
“I can tell you all day that if a number X correlates perfectly to Y, it is a big indication that your average win rate increases. But will you understand what it means? More importantly, are you able to make use of that info? The answer is probably no," Matz said. "The tough part is to make all those abstract concepts tangible. It is all about finding correlations, interpreting them in a meaningful way and then dividing them into small pieces of advice everyone can remember.”
Should the Defiant take the Overwatch League by storm in 2019, the role analytics played will be hard to forget.
Noah Niederhoffer is a freelance writer who produces national radio shows at SiriusXM and develops, produces and launches podcasts in addition to covering sports and politics. He’s an Atlanta native and a graduate of the University of Maryland’s Philip Merrill College of Journalism.