This project is a MVP for an agent-based model of RAI's market behavior. The RAI system, the Uniswap pool and the agents are represented as classes. The agents each have their own strategies and make decisions on whether they should buy, sell, provide liquidity, mint etc based on said strategy in discrete timesteps.
The simulation tool does not exactly represent the way RAI system is deployed on mainnet, and is missing some aspects of the feedback mechanism described in the Reflexer whitepaper. As such, none of the outputs of the simulations should be treated as conclusive evidence of the way the system would behave in the real world, but are rather intended to help build intuition.
Currently the project is intended to be used for simulations with 1-hour timesteps.
The code is structured as follows: the agents module contains the classes defining the different kinds of agents who will interact with the different protocols in the protocols module (the RAI system itself, the Uniswap pool, later some auxiliary lending markets perhaps). The utils module is intended to contain miscellaneous utility functions as the project requires it.
Install Python 3.7 or higher.
Install Matplotlib:
pip install matplotlibInstall Numpy:
pip install numpyClone the repo.
Open the config.ini file and enter your desired simulation parameters following instructions there. Save your parameters.
Run the simulation:
python simulation.pyThe results are saved in the results folder.
Note: as of writing this, using LongETH agents makes the system diverge really quickly. Any feedback on why you think that might be the case is appreciated.
This project is open to any contribution. Please open an issue if you have anything in mind so that development can be easily coordinated. The priority in my opinion is to look for bugs and add more agents with different strategies.
Thanks to Ameen Soleimani for the idea of an agents based model.