The library uses Python 3 with the following modules:

• numpy
• scipy
• matplotlib
• pandas
• bspline (QLBS)
• tensorflow (DH)

Python library for QLBS learner

• Incoporates transaction costs into the Q-learner
• Alternative Q-learner reward function with sqaure root of the variance terms (standard deviation) + learning factor

Run QLBS_master.ipynb -> QLBS with transaction costs

Run historical_crypto_data_qlbs.ipynb -> running QLBS on historical Crypto options data from Deribit

Run QLBS_std.ipynb -> QLBS with std reward function and smoothing

Python library for Deep Hedging

• Incoporates transaction costs into the Deep learner

Run NN_options_master.ipynb.ipynb -> Simple NN and RNN with transaction costs

NN_model.py -> NN and RNN model architecture

Igor Halperin, (2017). “QLBS: Q-Learner in the black-scholes (-merton) worlds." (QLBS) Hans Buhler et al, (2019). “Deep Hedging.” (Deep Hedging)