Conditional volatility forecasting in a dynamic hedging model

This paper addresses several questions surrounding volatility forecasting and its use in the estimation of optimal hedging ratios. Specifically: Are there economic gains by nesting time-series econometric models (GARCH) and dynamic programming models (therefore forecasting volatility several periods out) in the estimation of hedging ratios whilst accounting for volatility in the futures bid-ask spread? Are the forecasted hedging ratios (and wealth generated) from the nested bid-ask model statistically and economically different than standard approaches? Are there times when a trader following a basic model that does not forecast outperforms a trader using the nested bid-ask model? On all counts the results are encouraging – a trader that accounts for the bid-ask spread and forecasts volatility several periods in the nested model will incur lower transactions costs and gain significantly when the market suddenly and abruptly turns.