- SAP Managed Tags:
Competitive esports is a field, where speed and precision decide between winning and losing. Analysts and players typically have the daunting task of sifting through hours of game play to get valuable insights that can change the course of a match. Team Liquid, one of the world's most successful esports organizations, is always seeking new methods to achieve that competitive edge. In the Next Level Esports Center on SAP Business Technology Platform (BTP), Team Liquid can analyze various aspects of all pro-games in an effort to prepare for their next League of Legends match-up and it already helped them to win a number of championships in the past years.
In the past it was not feasible to detect teamfights, since it’s a time-consuming task to analyze hundreds of matches every month. That's why the team required a solution to automate the process of team fight detection in their League of Legends matches.
This article discusses how Team Liquid uses SAP to automate and optimize their team fight detection process, saving time, minimizing costs, and having improved insights into their gameplay.
The Challenge: Manual Analysis in a Fast-Paced World
In traditional analysis, identifying and studying team fights — the pivotal moments in a match when two teams clash — is a labor-intensive process. Analysts were required to manually watch every fight, keeping track of vital information, such as which players took part, the duration of the fight, and how the fight impacted the game overall. Because one game can last between 25 and 45 minutes, this process could consume hours of analysts' time, with the outcome being prone to human error.
For a high-performance team like Team Liquid, this inefficiency was a major issue. They needed a more precise and automated system that would quickly detect team fights and provide valuable insights for coaching and strategy.
The Solution: Automated Team Fight Detection with SAP
The solution designed for Team Liquid is based on an automated team fight detection system that works very harmoniously with the existing tools on SAP BTP. The system works by analyzing in-game data for specific events, such as damage dealt, player proximity, and the timing of interactions. A range of machine learning models were trained to evaluate their performance with respect to the rule-based model. Using the resulting feature importances together with manual parameter optimization allowed the rule-based system to perform on par with the machine learning models. By defining a series of conditions, the system can automatically identify when a team fight begins and ends — no manual tracking necessary.
The key technology enabler behind this solution is SAP HANA, which provides the computational power needed to process and handle enormous volumes of data in real time. Using the Predictive Analysis Library (PAL) of HANA Cloud, the team was able to implement advanced clustering algorithms, which dynamically group players based on how close they are to one another and their participation in a team fight. The different clustering methods and their parameters were evaluated using PAL's slight silhouette score and Team Liquids’ expert game knowledge. This combination of methods ensures that all the team fights, no matter how complex or dynamic, are captured.
Additionally, the solution utilizes machine learning algorithms to improve the detection process over time, continually increasingits speed and accuracy. By detecting players movements and interactions, the system is able to track team fights across multiple matches, ensuring that analysts can quickly identify and examine critical moments without having to manually comb through dozens of hours of gameplay.
Key Benefits for Team Liquid
The impact of this automated team fight detection system has been game-changing for Team Liquid, both in terms of time savings and cost reductions. The team estimates that the new system saves 10,000 working hours annually — time that was previously spent manually identifying team fights, a task which is now completely automated. This time-saving benefit allows analysts to focus on strategically more relevant tasks, such as strategy development and player performance analysis, ultimately enhancing the team’s competitive edge.
From a financial perspective, the system is expected to save $250,000 annually by reducing the time spent for manual analysis. With saved time and cost savings, Team Liquid can shift resources to other resources, such as training players, scouting, or equipment, all of which support their continuous success in the extremely competitive arena of esports.
Real-World Impact: Actionable Insights for Coaches and Analysts
While the ability of the system to save time and expense is impressive, the real value lies in the actionable insights it provides to Team Liquid’s coaches and analysts. By allowing the team to automatically detect team fights, the team has better and more stable data available that can be used immediately to improve their decision-making process.
For instance, coaches can now quickly see which players were involved in specific team fights and how those fights impacted the game. They are now able to monitor player performance within such key situations and identify the areas that need to be re-evaluated and improved. With 16.000 analyzed pro-games in the database, this is helping them to recognize patterns in their own and opponents’ behavior to make respective adaptations to their strategy.
Conclusion
The team fight detection feature developed for Team Liquid using SAP HANA Cloud represents another milestone in the long-standing co-innovation partnership. By automating the time-consuming process of detecting team fights, Team Liquid has saved thousands of hours, reduced costs, and gained invaluable insights into their gameplay. This solution has empowered the team to focus on what matters most — improving their performance and strategy, both individually and as a team.
As esports continues to grow and evolve, leveraging machine learning and artificial intelligence with SAP products will be key to staying competitive. For Team Liquid, this new system is not just about saving time — it’s about using data to drive better decisions and ultimately achieve success in the world’s most competitive gaming tournaments.
Big thank you to @Thomas_Woelkhart for the design and the development of the feature together with our partner Team Liquid.
2 Comments
