marketsai / marketsai

A modular framework designed to simulate economies and markets using Reinforcement Learning.

marketsAI

marketsAI is a modular framework designed to simulate economies and markets. Each economy or market is an OpenAI Gym compatible environment.

Installation

Because the library is still on heavy development we are not yet on PyPi. You can however install the package using pip:

pip install https://github.com/marketsAI/marketsAI/archive/alpha.zip

Structure

The main code is in the marketsai/ folder. The relevant sub-folers are:

economies/: In this folder there are two types of scripts:

• Scripts starting in run_ are desgined to run environments. There you change the environment, specify the configuration of the training and create custom metrics.

• Scripts starting in analysis_ take trained models and evaluate them on the eval_model of the environemnts. Here we create graphs as well.

Economies

• Growth Model
• Stochastic Growth Model
• Krusell Smith
• Heterogenous entrepreneurs
• Townsend

marketsai's Issues

Solving the models

Hi! I am an econ student and find your repository very interesting. I am also implementing rl in economics questions. May I know whether you had try to use any algos (e.g. PPO?) to solve some of those Envs and get the exact solutions as the analytical ones?