A new and more effective method for dynamic hedging, developed during my research internship at Tsinghua University. The LSTM neural network outperformed 5 baseline models in several markets. Please see the project presentation slides.
The code is implemented in Python with Tensorflow, and the baselines were implemented in R.