transitions
Today we return to our series on regime switching and the topic of managing portfolios of strategies. In particular, we build on the examples illustrated in sensitivity testing and steppin’ out, in which we showed historical and then real-time ‘forward-walking’ of strategies. The next step we’d described was to evolve the techniques illustrated to support the real-time management of a portfolio of strategies.
In the example below, we look at another ‘meta’ strategy named StrategyPortfolio which maintains a dynamic portfolio – P – of strategies which it will select from a set of strategies – S – running concurrently in simulation. The constituents of P as well as their cash allocations and parameterizations will be rebalanced/adjusted regularly after an initial ‘out-of-sample’ period during which only the S strategies are run.
Apart education, the intention of this strategy, as I’d originally suggested here, is to ‘back-into’ a regime-switching strategy without attempting to directly quantify the regimes explicitly.
This has proved to be even more interesting than I’d expected, not so much because it performs particularly well (though it’s promising), but because of all of the things it has taught us. In particular, the transitions are a killer and there are properties of strategies which (dis-)qualify them from being effective in such a scheme…
bad news good news 
As before, I’m not going to take the time to write-up my results in any formal manner but will again rely upon a quick screencast of the software running. The good news is that I’m figuring out how to edit these things, so it’s mercifully shorter than earlier screencasts…
Please click here to see the screencast.
I hope you find it interesting and will look forward to any comments or suggestions you might make.
EMS Internals, portfolio management, regime-switching, strategy development
Very interesting idea. A thorny problem to deal with is the one that makes regime-switching models hard for everyone: “In particular, the transitions are a killer and there are properties of strategies which (dis-)qualify them from being effective in such a scheme…”.
Often transitions are characterized by high volatility, and if your model doesn’t switch over quickly the losses in transition can be disastrous.
A couple of thoughts with no conclusion: There’s a lot of historical data, academic studies and common sense showing that capital markets are very sensitive to Fed policy – I know one successful quant who has three models – one for Easy, Neutral and Tight. Of course, the Fed Easy trigger didn’t kick off in 2008 & 2009 as it normally would – it took the launching of the massive QE program in late March 2009 to make a proper bottom. Did I call it? Of course not, I went super long in November 2008 and got pinched pretty hard.
One idea would be to have a TRANSITION definition in your regime switching routine – and you wouldn’t have to trade through the transition period – if you could just accurately and quickly define a transition period early on and cut the bleeding you’d be on to something.
Last, a good quant pal of mine in institutional equities has a factor efficient model he runs where he picks the 3 factors that should be most correlated with return going forward, and then he tries to find one more sensible factor but that’s negatively correlated with the output of his favored three factors and runs an optimization routine to determine the lowest risk combination of the weights of the four factors.
Keep up the interesting work.
Thanks for your interesting comments and kindly words.
“Often transitions are characterized by high volatility, and if your model doesn’t switch over quickly the losses in transition can be disastrous.”
Oh so true. But if the ‘right’ strategies are running during a transition, it can be very profitable. Ultimately, this example fails due to the fact that potential transitions are being scheduled on a fixed timeframe and the market isn’t so metronomic. Pity.
Another thing that may not have been clear in my description of the example is that the actual mechanics of the transitions are difficult. e.g, a strategy running in sim has a position in crude and it gets ‘promoted’ onto the live environment. Should we initialize the strategy to have the same position so that the two strategies are effectively in sync? This raises a question whose answer ’smells’ a lot to me like the idea of idempotence in computing… There are other similar issues here as well.
Your suggestions sound good to me, particularly the last. Alas, so many strategies, so little time… and each takes real effort to see through to a happy (or not) conclusion.