The firm’s chief researcher Robert Rotella will go back and look at the models to see whether there is anything unsound about the way they are programmed. Generally, there is not.
Next, a check is made to see whether there is a way to improve the model. Such improvements have to made with caution, he says, because the whole idea of being a systematic trader is to have robust systems in place that don’t have to be tweaked all the time.
The more a trader tinkers, the more he has to worry about curve fitting. While Rotella, an absolute disciple of robust models, doesn’t like to fiddle with its models, there is sometimes no choice if the market sends a message that losses continue even after delevering.
“Obviously you don’t want to overhaul a program in response to one year just because something didn’t work. That’s when you’re almost guaranteed that it would have worked the next year had you kept it in there.”
If you keep it changing it, you have no system.