MarketEntryGame

Description

MarketEntryGame is an environment for evaluating agents on strategic decision-making in a market entry game with capacity constraints. This environment wraps the MarketEntryGame implementation from TextArena, a framework for text-based game environments.

Capabilities

• Risk-benefit analysis with limited information
• Strategic communication and coordination
• Game-theoretic reasoning with capacity constraints
• Predicting opponent behavior under uncertainty

Compute Requirements

MarketEntryGame does not require a sandbox. It has minimal compute requirements.

License

MIT.

Tasks

There are two splits: train (150 tasks) and test (150 tasks). Each split contains 50 tasks across each of 3 variants:

• MarketEntryGame-v0
• MarketEntryGame-v0-train
• MarketEntryGame-v0-raw

Each task is seeded for reproducibility.

Reward Structure

This is a sparse reward environment. Rewards are mapped from TextArena's native range of {-1, 0, 1} to {0.0, 0.5, 1.0} via (raw + 1) / 2.

We do not use LLM graders for this environment; reward is determined programmatically.

Data

Game state is generated procedurally by the TextArena engine using seeded randomness. No external data files are required.

Tools

Agents are given three tools:

• send_message(message): Send a chat message during the communication phase.
• enter_market(): Enter the market for this round.
• stay_out(): Stay out of the market for this round (guaranteed 5 points).

Time Horizon

MarketEntryGame is a multi-turn environment.

Environment Difficulty

Medium. Agents must balance the guaranteed payoff from staying out against the higher but risky payoff from entering the market, while communicating with opponents to coordinate or mislead across 5 rounds.

Other Environment Requirements

This environment requires an OpenAI API key (passed via secrets) to power the LLM opponents.

Safety

Agents in MarketEntryGame interact only with an economic simulation and have no access to external systems, the internet, or sensitive data. The environment does not present safety risks.

Citations

@software{textarena2024,
  author    = {Guertler, Leon and Banting, Wilfried and Pignatelli, Eduardo},
  title     = {TextArena},
  year      = {2024},
  publisher = {GitHub},
  url       = {https://github.com/LeonGuertler/TextArena}
}