In a recent column for Canada’s Globe & Mail, Norman Rothery says that using simple numerical stock-picking models can lead to strong returns — but that sticking to those strategies is both critical and very difficult.
“A huge pile of research points to an array of simple numerical stock strategies that boosts returns over the long term,” Rothery writes. “Such methods range from low-ratio value strategies to momentum-oriented schemes. If the studies are to be believed, it should be easy as pie to make a small fortune on Bay Street. (Even when you don’t start with a large one.)” But, he says, numerous studies show that when humans try to tinker with statistical forecasting models, they end up hurting their results — whether it be in investing or other fields. “In other words, investors might get the best performance from following simple stock screens to the letter, rather than using them as a short list to pick and choose from,” he says. “The extra intelligence brought to bear on the problem might actually hurt returns.”
Rothery points to data from top investor Joel Greenblatt that seems to confirm this. “Mr. Greenblatt … offers investors two types of accounts,” he says. “In one type, investors follow [Greenblatt’s statistical stock-picking] method automatically. In the other, they start with his list and then pick individual stocks they want to buy.” Over a period of nearly three years from 2009-2012, the mechanical accounts produced total gains of 84% after expenses, handily beating the S&P 500’s 63% return. Those who picked and chose stocks from Greenblatt’s screens gained just 59%, however.