Investment Strategy for VIX Futures based on a Bayesian Approach to Term Structure Modelling

Investment Strategy for VIX Futures based on a Bayesian Approach to Term Structure Modelling

Oostdam, Oskar Oostdam (TU Delft Electrical Engineering, Mathematics and Computer Science)

Bierkens, G.N.J.C. (mentor)
de Graaf, Kees (mentor)
Papapantoleon, A. (graduation committee)

Delft University of Technology

Applied Mathematics

2021-08-25

The VIX index, which is the expected volatility of the S&P 500 index in 30 days, is of interest to a lot of investors on the US financial market. Allowing the volatility of the financial market to be used as a trading tool gives rise to interesting investment opportunities, such as hedging and speculation. In this thesis we will be creating an investment strategy on VIX futures by modelling the term structure with a Bayesian approach. Using Markov Chain Monte Carlo (MCMC) methods, we will simulate a posterior distribution of our term structure, which results in credible intervals of our futures prices. We will compare the efficiency of Metropolis-Hastings algorithms against the No-U-Turn Sampler, which is a Hamiltonian Monte Carlo algorithm. Eventually we find that the No-U-Turn Sampler significantly outperforms the Metropolis-Hastings algorithms. The resulted credible intervals of our futures prices will be used to determine whether a contract is overvalued or undervalued. The strategy consists of taking a combination of long and short positions on VIX futures contracts which we consider to be mispriced. We will therefore take a long position on undervalued VIX futures, while taking a short position on overvalued VIX futures. We eventually find that this investment strategy is very risky due to the high volatile behaviour of the VIX index.

VIX Index
Bayesian statistics
Term structure modelling
Metropolis-Hastings algorithm
Hamiltonian Monte Carlo
No-U-Turn Sampler
Investment strategy
VIX Futures

http://resolver.tudelft.nl/uuid:c2909082-f9fe-49dd-b2b5-4ced0c14d75b

Student theses

master thesis

© 2021 Oskar Oostdam Oostdam