Volatility Management From The Perspective Of Option Market Markers:Modeling And Forecast Of The Implied Volatility Surface

Options occupy an important position in the financial markets. From the international experience, as basic exchange-traded products, options, futures and spots together con-struct a complete system for risk management. On February 9th 2015, the Shanghai Stock Exchange (SSE) launched its first ETF option, based on the SSE 50 Index ETF as the de-but product. Study on option pricing and related issues becomes more significant for the development and improvement of China’s financial markets.Black-Scholes model is widely criticized for its inconsistency, due to the fact that the volatilities implied from observed option prices are different across strikes and time to maturities. The obvious shortcomings of the Black-Scholes model have triggered a large number of literatures in which both academics and practitioners tried to account for its lim-itations. One strand resorted to static volatility models, such as stochastic volatility models, jump-diffusion models, local volatility models. As suggested by a lot of empirical studies, the implied volatility surface also changes dynamically over time. These features inspired us to build a dynamic model for implied volatility surface, which can not only capture cross-sectional information on the relationship of implied volatilities among contracts with different strike and maturities, but also describe how the implied volatility surface evolves over time. We call such a kind of model the dynamic implied volatility model.Implied volatility surface is the core of option pricing, risk management and trading strategies. It is also the connection between exchanged-traded derivatives and over-the-counter derivatives. From the perspective of market making for exchange-traded options, the pricing errors should be small and preferably within the bid-ask spread, as a market maker can rarely take views against the whole market. From the perspective of pricing and hedging exotic options, the model should fit the market implied volatility data accurately to make sure all derivatives are priced under a consistent system, in case the formation of arbitrage opportunities.This thesis focuses on the construction of the implied volatility surface. The main work of this thesis is:1. Summarized five methods for constructing the implied volatility surface. After com-paring the fitting accuracy, calibration speed, theoretical background and stability of model parameters, we analyzed the reasonability of each model for being the inter-polation model of single implied volatility surface, from the perspective of market makers.2. Improved the calibration process of the SABR model, SVI parameterizations and JW-SVI parameterizations, to increase the calculation speed and stability of model parameters. Tested the stability of model parameters by using intro day data of Gold ETF options, to prove the necessity to develop the dynamic implied volatility model.3. Went deep into the dynamic implied volatility model and gave an introduction to the existing three strands of this kind of model, including market-based dynamic im-plied volatility model, Vega-Gamma-Vanna-Volga model proposed by Peter Carr and Liuren Wu, factor-based dynamic implied volatility model. To conquer the disadvan-tages of existing models, a general factor-based model for the dynamics of implied volatility surface is proposed, without the linear restriction on volatility function for-mat. To propagate the nonlinear system, unscented Kalman filter is introduced, and a dynamic approach to provide optimal estimation and efficient forecast of the implied volatility surface is proposed.4. By using panel data of implied volatility surface of S&P500 index option, the fore-cast and estimating accuracies of the were tested. Comparisons were drawn among several related models, to justify the meaning of this approach from both theoretical and practical point of views.