Home / Insights / Market simulation
Research topic
Market simulation puts agents into a market with fixed rules and records what they do. Economists have built markets this way for decades, from laboratory double auctions with human subjects to agent-based models of financial markets. LLM agents now bid, set prices and negotiate in the same settings.
The research here covers what those studies find, how simulations are designed and checked against theory and human results, and what the findings mean for anyone who deploys an agent that buys or sells.
Research report
2024 to 2026
A review of published studies that put LLM agents in auctions, pricing, negotiation, marketplaces and trading: the biases observed, the collusion and deception findings, and the evaluation gaps.
Methods note
How to test LLM agents as bidders: induced values, known efficient outcomes, measures read from bids alone and human reference agents built from published experiments.
Rillor tests LLM agents in simulated markets and auctions built on the designs of experimental economics. Rules, values, news and prompts are frozen per run, and agents receive induced values, so the efficient allocation is known before the first bid. Agents from many model families bid side by side, and every bid is logged. Each agent is measured for efficiency, price convergence, biases, collusion and distance from human results.
AI outputs can be wrong. Important decisions should include human review.
Rillor is not a registered investment adviser or commodity trading advisor and does not provide investment or trading advice. Research, forecasts, data and software described on this site are for research and engineering use. Nothing here is an offer or recommendation to buy or sell any security, commodity interest or digital asset. Past or simulated results do not indicate future results.
Send us the agent you want tested, the market or auction format and the behavior you need measured.