--> Repo contains parts of my MSc dissertation project, for reusage. Code provides a framework written in R for automated computation of VaR based on different time series models (GARCH, CMM, historical simulation) and experimental neural networks (LSTM-MDN architecture, implemented in python using TensorFlow/Keras framework).
The repo mainly contains an automated pipeline downloading stock data, training various models for VaR calculation and running backtesting on each model. Code can be fully run via one-line execution (by running the execution.R script).
TBC