Just had a read of an article from Rajiv Sethi who had this choice quote to make on regime switching:
"One ought not to expect markets to be efficient or inefficient, but rather to experience periods of relative efficiency that are interrupted from time to time by severe disruptions."
"Relative efficiency" I like that.
There is a real groundswell of academic/ industry and blogosphere research into the robustness of rotation/ switching strategies.
Only last night, I spotted another great piece of research from MarketSci, testing a Fidelity select rotation strategy. Adding a trend following filter as also suggested by Mebane Faber smooths and increases returns compared to the original idea.
There's a couple of missing pieces for me.
1. As already mentioned, trading costs have to be considered. The Fidelity select strategy makes use of the no switching fee facilities available. The good news is that this strategy is potentially workable from the UK thanks to the Fidelity Funds network. What I need to do next is research the number of funds available with no 'load' or switching fees to see if such a strategy would be workable within a tax wrapper. Another consideration is that although there may be no transaction costs, the funds still charge an annual fee and as I've blogged about here, this can materially eat into your returns. So your 'alpha' will have to be significant to overcome these annual fees. CXO address this issue from a US perspective here.
2. Secondly, there has to come a point when, as Sethi points out, the momentum in anomalies/ strategies/ markets becomes a bubble. If you're a momentum switching model, there is always the danger of catching something at the top of the market. So I found it interesting to read this paper from Trimtabs which indicates that cash inflows into ETF's can be a great contrarian indicator. I wonder if adding this as a filter could act as a good bubble avoidance strategy. I'd like to test this,but at the moment am unsure how to get the data without it costing the earth.