Patches in esports games are often introduced to bring about changes and keep the gameplay fresh. These patches can have significant effects, such as eliminating building types or altering heroes. They may also involve subtle changes, like adjusting cooldowns or item prices. The impact of these patches on gameplay is difficult to determine immediately and often requires observation of how professional teams adapt. As a result, a new “meta” game develops until the next patch is released.
When it comes to creating betting odds for MOBAs (Multiplayer Online Battle Arenas), patches can pose a challenge. Side markets focusing on objectives like killing certain monsters need to consider the frequent changes in these monsters’ properties and priorities. Simply relying on existing models can leave room for clever punters to take advantage. However, adjusting models takes time as new data needs to be generated. It can take several weeks to gather enough data to ensure model accuracy for side markets, causing delays or increased risk.
Imagine if software could predict meta game changes by analyzing patch notes. This would revolutionize the industry! One possible method is using self-play, where a computer steers two teams in games like League of Legends or DotA2 until a new meta emerges. By comparing the shift in strategies and collecting statistics, new prediction models could be created. With this approach, relevant matches from top teams could be gathered in a matter of hours instead of weeks. While still a significant step, it is feasible with a dedicated team and access to high-quality data provided by companies like Bayes Esports.
In the future, it is possible that AI assessments of patch notes may replace traditional expert opinions. OpenAI has already demonstrated success in teaching machines to play Dota2 at a professional level. Modern AI can learn strategies, coordinate teams, and optimize for long-term goals. As a data scientist with a passion for sports and research, Dr. Darina Goldin leads the research team at Bayes Esports, a company employing a data-driven approach to esports. In her column, Dr. Goldin explores ideas for products that could benefit the esports community, though they may not be easy to accomplish.
