GEEvo: Evolving Balanced Game Economies with Genetic Algorithms

By: Prof. Dr. Kai Eckert | Thu, 25 Jul 2024

Novel framework uses evolutionary algorithms to generate and balance complex graph-structured game economies

Florian Rupp and Kai Eckert present “GEEvo: Game Economy Generation and Balancing with Evolutionary Algorithms” at the 2024 IEEE Congress on Evolutionary Computation (CEC).

The Challenge of Game Economy Design

Game economies—complex systems of resources, crafting recipes, currencies, and progression paths—are fundamental to many modern games. However, designing balanced and engaging economies is notoriously difficult. Designers must ensure that no single path is overwhelmingly advantageous while maintaining interesting strategic choices for players. This balancing act traditionally requires extensive playtesting and iterative refinement.

Graph-Based Economy Representation

The research recognizes that game economies are naturally represented as graph structures, where nodes represent entities (resources, items, currencies) and edges represent relationships (crafting recipes, conversion rates, trade opportunities). This graph-based perspective enables the application of sophisticated algorithms for generation and optimization.

Evolutionary Algorithm Approach

GEEvo employs evolutionary algorithms to generate and balance game economies through an iterative process:

  1. Generation: Random mutations are applied to graph edge lists to create candidate economies
  2. Evaluation: Each economy is assessed for balance and desirability
  3. Selection: Better-performing economies are retained
  4. Evolution: Selected economies undergo further mutation and refinement

This bio-inspired approach allows the system to explore vast design spaces and discover balanced economy configurations that might not be obvious to human designers.

Balancing Through Simulation

A critical component of GEEvo is its use of simulation to evaluate economy balance. By simulating player interactions with the economy, the system can identify imbalances such as:

  • Resources that are too easy or difficult to obtain
  • Crafting paths that provide disproportionate advantages
  • Dead-end progression paths with poor rewards

Variable Complexity Generation

An important feature demonstrated in the research is GEEvo’s ability to generate economies of varying complexities suitable for different types of games. Simple mobile games might need streamlined economies with few components, while complex simulation games benefit from intricate economic systems with many interconnected elements.

Practical Applications

This work has immediate practical applications for game developers, particularly:

  • Rapid prototyping of economy designs
  • Automatic balancing of existing economies
  • Generation of procedural economies for roguelike or sandbox games
  • A/B testing of alternative economy configurations

Citation: Florian Rupp, Kai Eckert (2024): GEEvo: Game economy generation and balancing with evolutionary algorithms. In 2024 IEEE Congress on Evolutionary Computation (CEC).