Mean-Reverting Logarithmic Modeling of VIX

Since March 26, 2004, when the CBOE Futures Exchange (CFE) began trading futures written on S&P500 volatility index (VIX), volatility has become a widely accepted asset class as trading, diversifying and hedging vehicle by traders, investors and portfolio managers over the past few years. On February 24, 2006, CBOE introduced options written on VIX index and since then VIX option series has now become the most actively traded index option series on CBOE. This thesis focuses on mathematical modeling of spot VIX with standalone approach. Unlike the consistent modeling approach in literature, which starts with specifying joint dynamics for SPX index and its instantaneous stochastic volatility then derives expression for spot VIX and price VIX derivatives based on this expression, standalone approach starts with directly specifying dynamics for spot VIX and prices VIX derivatives in this simpler framework. Although there is work in literature that studies the mean-reverting logarithmic model (MRLR), no work has been done in considering stochastic volatility in MRLR to capture the positive implied volatility skew of VIX option, nor have they compared the pure diffusion version of MRLR with its jump and/or stochastic volatility extensions. Furthermore, most of the literature only focus on static pricing formulas for VIX future and VIX option, no work has been done in investigating the dynamic feature of VIX future, calibration and hedging strategies of mean-reverting logarithmic models, as well as the convexity adjustment of VIX future from forward variance swap, which has a liquid variance swap market to back out the vol-of-vol information in mean-reverting logarithmic models. In this thesis, I present four versions of MRLR models. The first model is a pure diffusion model where spot VIX follows a mean-reverting logarithmic dynamics. Then I extend this basic MRLR model by adding jump or stochastic volatility into spot VIX dynamics to get MRLRJ and MRLRSV models. Finally, I combine jump and stochastic volatility ...