A good deal of recent evidence suggests that retail investors consistently predict future stock returns. Are these retail investors merely good followers, buying stocks that are already on the rise? Or are they the leaders that other investors should follow, making well-founded trading choices that put them in front of the market?
Professor Paul Tetlock worked with Eric Kelley of the University of Arizona to better explain how it is that retail traders regularly predict stock price movements. “Past research focuses on how many dollars retail investors lose from trades and how many dollars institutional investors win,” Tetlock explains. “Instead, we look at the direction of trades to see if retail traders are buying in advance of price increases and selling in advance of price decreases.”
The researchers analyzed the largest and most recent US database of orders ever studied, made up of retail order flow routed to two large market centers from 2003–07, representing 225 million executed trades and $2.6 trillion in volume. The researchers first measured retail trader order imbalances, adding up all buys and subtracting all sells, and found that the net buying activity of retail investors positively predicts the direction of future stock returns for at least one month and up to three months. They then combined the order data with a database of 3.75 million financial newswires covering the same 2003–07 period to determine if any of three possible explanations account for retail traders’ ability to predict returns.
The private information hypothesis says that retail investors trade on information that others aren’t yet aware of, and that a stock’s price will take some time to reflect all available information. “Suppose there is a relatively large number of physicists in the United States who know a lot about microchips,” Tetlock says. “They may know something about AMD, the microchip producer, that stock analysts on Wall Street don’t know — perhaps AMD has developed a superior production process that will ultimately lead to big gains in market share for AMD — and those physicists buy ahead of everyone else.
“We should see newswire stories with many positive words about AMD start to appear in the weeks after the physicists started buying AMD, when the market and the financial press become aware of that private information.” Tetlock says. “And it does turn out that aggressive buying — market orders from retail traders to immediately buy regardless of price — usually precedes positive news.”
The liquidity provision hypothesis considers how to explain activity around retail traders’ so-called patient orders — limit orders to buy or sell a stock when it reaches a certain price. “Maybe AMD suffered a negative liquidity shock when a mutual fund sold it for reasons unrelated to AMD’s profitability. Retail traders who recognize this can step in and buy AMD stock cheap, which provides liquidity to the mutual fund. The traders eventually realize profits when the stock rebounds, once people see that AMD’s profits haven’t changed and there is nothing fundamentally wrong with the firm,” Tetlock explains. In short, limit orders are responses to non-news, correcting past errors in market prices. Consistent with this, the researchers found no relationship between retail traders’ limit order activity and firm-specific news.
The one explanation the researchers do not find support for is the autocorrelated flow hypothesis, the idea that some retail traders simply buy day after day, putting repeated upward pressure on stock prices even though they lack private information. According to this story, because prices rise for no real reason other than in response to the buying activity, prices should eventually fall when the market corrects itself. But this is inconsistent with the researchers’ finding that price increases occurring after retail investor buying activity continue for at least one month and as long as one quarter. Tetlock suggests that traders who buy day after day may be like the hypothetical AMD physicists, buying one after another acting on the same information that the market simply hasn’t figured out yet.
Tetlock’s results, along with other recent evidence, contrast with earlier findings about retail investors’ predictive abilities, differences that he attributes to recent changes in the trading population. “Traders disproportionately invested in Internet stocks in 2000 would have lost about 80 percent of their money over the following two years, whereas traders with more diversified investments would have kept most of their wealth,” he says. “It’s an evolution argument: survival of the fittest. Those who were actively trading and doing poorly simply lost their money. Who is left? People who didn’t do that.” He also points out that early adopters of online trading may have been disproportionately young and male, suggesting there may be a more prudent population of older investors and women investors reflected in recent data.
“Ultimately, knowing that today’s retail traders predict which way stock prices are going to move in the next month is valuable information for institutional investors, high-frequency traders, and market makers whose profits depend on monthly stock price movements,” Tetlock says. “It allows them to direct their portfolios the right way.”
Paul Tetlock is the Roger F. Murray Associate Professor of Finance in the Finance and Economics Division at Columbia Business School.