How computers can help account for human biases and nudge stock prices closer to their real value.
CAN COMPUTERS PICK stocks and optimize investment portfolios?
Yes, they can come close to doing what human portfolio managers do — and often better. That’s the view of Charles M.C. Lee and other proponents of active quantitative investing, who believe that investing is a scientific, quantifiable process that can be driven by technology. It involves the use of speedy computers, and sophisticated software and mathematics to find patterns in financial data that can be turned into successful investment strategies. “It’s like teaching the computer to think like an analyst, teaching it how to analyze and screen thousands of stocks quickly. You harness intuition, not just automate,” explains Lee, an accounting professor at the Stanford Graduate School of Business.
First, portfolio managers electronically comb through public and private databases of financial information on listed companies, such as the universe of more than 8,000 U.S. stocks. They quantitatively screen the stocks by sifting through millions of data points on earnings, cash flow, market-to-book ratios, market indicators, and other measures. The computer ranks the stocks in attractiveness on many attributes, based on mathematical models set up by portfolio managers. These rankings allow the system to produce return forecasts on thousands of individual stocks.
Computers then come up with a list of recommended stocks to buy and sell. For instance, Cornell’s Cayuga MBA Fund initially screened more than 6,000 U.S. stocks, yielding weekly lists of 560 recommended stocks that were the most and least attractive. The list was sorted by industry sector and assigned to MBA students acting as sector analysts. They winnowed the list to 120 stocks and prepared detailed presentations for classmates. After a class vote, 100 stocks were selected for 50 long and 50 short positions.
After target stocks are picked, the next step is constructing the portfolio while controlling its overall risk. Specialized risk management software calculates the appropriate weightings of individual stocks and industry sectors to ensure that the overall portfolio is sufficiently stable.
Finally, technology is used to forecast and minimize trading costs of individual stock holdings. Transaction expenses, especially for large funds or for those that trade frequently, can have a significant impact on portfolio returns.
For active quantitative managers, it takes a lot of science to get the process right.
— Maria Shao
Also on Stanford Knowledgebase: