A study of Artificial Intelligence for VIX and VIX futures
Details
- Authors: Nikolaj Brux
- Title: A study of Artificial Intelligence for VIX and VIX futures
- Supervisior: Prof. Dr. Jörg Osterrieder
- Degree: Bachelor of Science
- University: Zurich University of Applied Sciences
- Year: 2022
- Status: Working Paper
Summary
Exploring the feasibility of forecasting the VIX index by only using VIX future prices by using machine learning and deep learning techniques.
Abstract
Forecasting techniques using some forms of artificial intelligence have become very popular. Within a similar time period, the Cboe Volatility Index (VIX) became increasingly popular. Occasionally, questions are raised about the lead-lag structure of the VIX and its derivatives, pointing to inefficiencies or market manipulation. In this study, we will explore the predictability of machine learning models and neural networks by forecasting the Cboe Volatility Index (VIX) by only using VIX futures prices. With a simple ‘by hand’ approach, four models are constructed and another four ‘ensemble models' are created by combining equally weighted forecasts of the first four constructed models. Comparing the eight models against the benchmarks (naïve approaches) yields no meaningful results. The neural networks interpret the data as a random walk, indicating poorly constructed models and/or an unsuitable approach for this problem.
Important links
Data
- Cboe Volatility Index (VIX)
- VIX futures daily
Data source VIX: Yahoo Finance
Data source VIX futures: Bloomberg