Can abnormal returns be earned on bandwidth-bounded currencies? Evidence from a genetic algorithm

Most of the studies about the Foreign Exchange market (Forex) analyse the behaviour of currencies that are allowed to float freely (or almost freely), but some currencies are still bounded by bandwidths (either disclosed or undisclosed). In this paper, I try to find out whether two bandwidth-bounded currencies, the Hong Kong dollar (HKD) and the Singapore dollar (SGD), present opportunities for abnormal returns. I consider a set of trading rules, and I use a genetic algorithm to optimise both the subset of rules to be used and their parameters, using real market data. I use four pairs of currencies, two of them involving currencies bounded by bandwidths and two others involving only free-floating currencies. I compare the results obtained for the different pairs, both in terms of profitability and in terms of the types of the rules that are used. Evidence of profitability is more consistent for the only pair including a bandwidth bounded currency without a narrow price band, the USD/SGD. Trend reversing rules are preferred for this currency pair, while the preferred type of rule seems to depend on the pair of currencies when free-floating currencies are considered. In the case of the SD/HKD, the small number of price changes, as well as the price stability (possibly consequences of a narrow price band) do not allow me to obtain conclusive results.