MOB

Modeling Order Books.

Installation

Our requirements where produced using python3.8

Install torch (version 1.13):

pip3 install torch --extra-index-url https://download.pytorch.org/whl/cpu

Install other requirements:

pip install -r requirements.txt

Set up the MOB environment variable for reading/storing data:

os.environ["MOB"] = path/to/folder

Install shap from

python -m pip install git+https://github.com/dsgibbons/shap.git

Data

Data is available here. This contains:

• 1 month of Orders Books history in HOURLY
• 6 year-length dataset of exogenous features in data/datasets/Lyon
• Experimental results

Scripts

The script folder contains the following:

• example.py to plot several Order Books (aggregated curves).
• compare_methods.py will solve the Dual Problem for several implemented solutions. Solutions with parameters (e.g. : k for sigmoid approximation) will test several values.
• obn.py fits an Order Book Network and inspect the forecasted order books along time.
• order_book_analysis.py extract information from the order book dataset : price distribution, acceptance distribution.