Optimizing numerical systems and mechanism design is crucial for enhancing player experience in Massively Multiplayer Online (MMO) games. Traditional optimization approaches rely on large-scale online experiments or parameter tuning over predefined statistical models, which are costly, time-consuming, and may disrupt player experience. Although simplified offline simulation systems are often adopted as alternatives, their limited fidelity prevents agents from accurately mimicking real player reasoning and reactions to interventions. To address these limitations, we propose a generative agent-based MMO simulation system empowered by Large Language Models (LLMs). By applying Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) on large-scale real player behavioral data, we adapt LLMs from general priors to game-specific domains, enabling realistic and interpretable player decision-making. In parallel, a data-driven environment model trained on real gameplay logs reconstructs dynamic in-game systems. Experiments demonstrate strong consistency with real-world player behaviors and plausible causal responses under interventions, providing a reliable, interpretable, and cost-efficient framework for data-driven numerical design optimization.

As the need for more sophisticated artificial intelligences (AIs) grows, the challenges that they must face along the way have to evolve accordingly. Real-time strategy (RTS) video games, unlike turn-based board games such as chess, can serve as a vast playground for pushing the limits of AI. In particular, StarCraft II (SC2), one of the world's most popular and skill-demanding RTS games, has already been the object of a few groundbreaking AI-related studies. In SC2 matches, each player has to build up and command an army of varied units to defeat their opponent using wit and grit. While AI-based systems can excel at many aspects of the game, improving their decision-making regarding when their units should be sent to or relocated during a battle is remarkably difficult.