README.md

Deep Hedging

Pricing Options (Derivatives) using Deep Learning models

The main either loads data (real datas), or generate datas (according to the Brownian motion). The input datas are stock prices, and an information set (which can be any information a human trader might use).\ The different models, defined in the deep_hedging_models folder, take the inputs, and simulate a N days trading days. Each day, the Agent infer a strategy and takes an action (see ).\

To run a model, you can simply run the main_Deep_Hedging.py, and specify the following arguments: , or enter main_Deep_Hedging.py -h to display the possible arguments Other options such as the volatility of the generated datas, or the strike price can be set at the begining of the main.\

The following package are mandatory to run the code:

• tensorflow
• keras-tcn
• QuantLib
• sklearn which can be installed through pip