The quest to identify the best League of Legends support player can spark intense debate among fans, yet establishing a definitive answer remains elusive. Currently, there is no systematic, data-driven framework for assessing player performance in esports. Instead, discussions are often shaped by a small number of analysts and fueled by public sentiment on platforms like Twitter and Reddit.
Evaluating player skill is critical for several reasons. For one, a player's value to their team directly impacts salary negotiations. Questions arise concerning individual skill, adaptability, and team success: does a player's excellence hinge on their current team setup, or do they have the potential to shine in any environment? Engaging discussions about player merits gain depth when backed by quantifiable data.
The appeal of baseball comes from its rich statistical foundation, as highlighted by Moneyball. Esports, too, collects similar statistics for its players, lending itself to comparable analyses.
Unlike traditional sports, which have only recently begun employing analyst teams, esports is poised to lead in research. While organizations like the NFL and NBA have only recently started to prioritize data analytics, esports inherently generates data through every match that occurs on a server. This standard data collection, supported by Riot Games through Bayes Esports Solutions, captures player positions, item purchases, and actions taken, essentially preserving a comprehensive game state with high temporal resolution.
Furthermore, it's possible to track mouse and keyboard inputs, which traditionally assist in cheating detection but could also measure player reaction times and focus.
The designated roles in games streamline research analysis; for example, an AWPer in CS:GO can be identified by their weapon, and a jungler in LoL can be marked accordingly in datasets. This allows researchers to quickly assess player capabilities rather than spending time categorizing roles after the fact.
However, as with any analysis, game patches pose challenges. While patches affect high-stakes decision-making and team dynamics, understanding how players adapt to these changes could provide valuable insights. At present, the statistics available are largely limited to what the games provide, and assessing performance through different patches remains an underexplored area.
Given the abundant data available in esports, there is a strong argument for it to take the lead in research around team composition and performance. Instead of looking to traditional sports for quantitative guidance, the reverse should hold true.
Dr. Darina Goldin serves as the director of data science at Bayes Esports. Her journey in competitive gaming began during her graduate studies with Team Fortress 2. While she no longer plays actively, her passion for esports continues. At Bayes, she has developed numerous predictive models for games like Counter-Strike, Dota 2, and League of Legends, and when she isn't analyzing data, she trains in Brazilian Jiu-Jitsu at the gym.
