This paper presents a manageable and effective way of nesting two popular, yet distinct approaches to obtain optimal hedging ratios - time-series econometrics (GARCH) and dynamic programming (DP). The nested DP-GARCH model is then compared to a DP-GARCH model that accounts for variability in the bid-ask spread often unobserved (and hence ignored) in most studies. Results from an empirical application using data from an importantly traded commodity " sugar " suggest that a DP-GARCH model that incorporates the bid-ask spread still outperforms more traditional models. Moreover, the hedging ratios are far less volatile, and statistically different, than those recommended by the traditional GARCH methods that ignore the spread.