This repository contains the code for the thesis "The estimation of Value-at-Risk and Expected Shortfall based on deep generative models"
Author: Belonovskiy Peter Ilich, HSE DSBA
Supervisor: Naumenko Vladimir Vladimirovich, HSE Associate Professor
The dependency managament in project was implemented via poetry. To clone this repository and set up the environment, run the following commands:
git clone https://github.com/BELONOVSKII/var_es_dgm.git
cd var_es_dgm
poetry install
poetry shellNote, poetry should be pre-installed in your system.
Thesis uses daily stock prices data from yahoo finance. To parse the yahoo finance and download data run:
python var_es_dgm/data_parcing/parse_yfinance.py This downloads individual stocks's data and produces combined file data/complete_stocks.csv that would be further used in the experiments.
- Variance Covariance:
var_es_dgm/basic_models/parametric.py - Historical Simulation:
var_es_dgm/basic_models/hist_sim.py - TimeGrad:
var_es_dgm/TimeGrad/
- Univariate:
experiments/univariate - Multivariate:
experiments/multivariate
All figures from the thesis could be created by running notebooks in visualisations/.