I’ve been reading a book of late that I will shortly be reviewing. There was a comment of the author’s that I wanted to address now, though. In talking about system testing the author made the statement that it’s better to run strategy tests based on position reversals (long to short, short to long) rather than position closures. It’s a statement with which I’m not very comfortable.
Now, to provide a little context, the focus here is mainly a longer-term investment one, so it’s generally looking at things from long-only perspective. That means a sell signal is an exit signal. It’s not really meant to be a short entry signal, though obviously there’s the potential for playing the short side as well. The author’s argument for testing a strategy as if shorts will be entered is that it will provide the best selling signals.
Here’s the thing, though. Good long exits are not necessarily the same thing as good short entries.
Pure traders, of course, have somewhat different considerations, but investors should be thinking about total return considerations. If the market is going to go sideways for a lengthy period, for example, it would be better to be out of your long near the start of the move into consolidation than it would be to wait until a short signal was triggered at the end of it. That would save a lot of emotional capital being wasted during a choppy back and forth period and provide the opportunity to put the money to work in a way which would actually provide a return (fixed income investments, for example).
My point is that optimal short entries do not necessarily make optimal long exits (or vice versa). Of course, it always comes down to testing.