Brownian-Market-Maker

Adapting Avellaneda & Stoikov market making using ccxt for KuCoin

Influced by this repo: https://github.com/fedecaccia/avellaneda-stoikov

Avellaneda-Stoikov Market Making Simulation

This project implements a simulation of high-frequency trading using the Avellaneda-Stoikov market-making model. The simulation utilizes Brownian motion to model stock price dynamics and implements market-making strategies to manage inventory and optimize profit.

Components

1. main.py

This is the main script that orchestrates the simulation. It initializes the market maker, simulates trading scenarios, and visualizes the results using Matplotlib.

2. brownian.py

Contains the implementation of the Brownian motion simulation, which is essential for modeling stock price movements over time.

3. brownian_path.py

Provides an example of generating a single realization of Brownian motion and plotting it using Matplotlib. This can be useful for understanding the stochastic process used in the simulation.

4. requirements.txt

Lists all the Python libraries and their versions required to run the project. Use pip install -r requirements.txt to install them.

Installation

1. Clone the repository:

   git clone https://github.com/your_username/avellaneda-stoikov-market-making.git
   cd avellaneda-stoikov-market-making

2. Install dependencies:

   pip install -r requirements.txt

Usage

Running the Simulation

To run the simulation with default settings (symbol: BTC/USDT, 100 simulations):

python main.py

Customizing the Simulation

You can customize the simulation by modifying parameters in main.py:

• Adjust n_sim in simulate() method to change the number of simulations.
• Modify parameters such as sigma, T, N, dt, gamma, k in the simulate() method to explore different scenarios.

Visualizing Brownian Motion

To visualize a single realization of Brownian motion:

python brownian_path.py

This will plot the path of Brownian motion over a specified time interval.

Results

The simulation results include:

• Final inventory held
• Last price observed
• Cash balance
• Final wealth (cash + inventory value)
• Maximum and minimum inventory held during the simulation
• Average and standard deviation of Profit and Loss (PnL) across simulations

Parameters

The parameters used in the simulations can be found and adjusted in main.py. These include:

• sigma: Volatility of the stock
• T: Total time period of simulation
• N: Number of steps in the simulation
• dt: Time step size
• gamma: Risk factor influencing bid/ask spread
• k: Market model parameter influencing intensity of order placement

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Federico Caccia https://github.com/fedecaccia/avellaneda-stoikov
• Marco Avellaneda & Sasha Stoikov for their pioneering work on market making models.
• Open-source libraries: CCXT, NumPy, Matplotlib, and SciPy and ChatGPT