Use of volatility prediction methods when trading selected options strategies

This thesis examines the use of volatility forecasting methods in the construction of long and short straddle option strategies. It is divided into a theoretical and a practical part. The theoretical part defines basic concepts, characterizes option strategies, volatility, and describes the GARCH and HAR models including their variations. In the practical part, six volatility forecasting models are tested on 30 stocks from the Dow Jones Industrial Average: GARCH, GJR-GARCH, HAR, HAR-X, AHAR, AHAR-X. The strategies are constructed from ATM options with a ten-day investment horizon. The selection of stocks to enter positions is controlled by the difference between implied and predicted realized volatility. The HAR-X and AHAR-X models achieved the best results. These were models with an exogenous dummy variable with information about the company`s earnings announcement. Adding this infromation led to a (slight) improvement in Sharpe ratios compared to simple HAR and AHAR. On the contrary, models from the GARCH family showed weaker results and higher drawdoown. In conlusion, it can be stated that volatility prediction can be used in the construction of options strategies, but practical implementation is limited by transaction costs, bid-ask spreads, and margin requirements.