Search code, repositories, users, issues, pull requests...
MarketAgents is a microeconomic simulation framework designed to model agent-based market interactions. It includes features for simulating double auctions, tracking market dynamics, and modeling agent behavior. The framework is built to be extensible and adaptable for various economic experiments. This library is designed for implementing turn-based Double Auction simulations with literate economic agents capable of reading/writing news and communicating with each other. The library aims to study the role of news and information processing in the pricing dynamics of simulated virtual financial markets. Traditional economic models have struggled to acknowledge the role of information networks in price formation beyond simplistic implementations due to the inability to model human communication using natural language. The latest generation of LLMs now makes this possible.
This library seeks to merge and validate formal economic models of trade and learning with recent agent-based modeling capabilities. Large language models can implement, model, and formalize three fundamental aspects of realistic economic agents that have been known weaknesses of traditional economic models:
