How to Trade Those Terrible Analyst Earnings Estimates

(Bloomberg) -- The prevailing view of earnings estimates is, to be blunt, dim. Many investors believe that companies and analysts conspire to keep estimates low to produce a "surprise" that drives the stock price higher.

Anecdotal evidence suggests that the practice is widespread. The 30 companies in the Dow Jones Industrial Average met or beat the so-called consensus earnings estimate an average of 17 times in the past 20 quarters, Bloomberg data show.

The result is a financial hall of mirrors. Investors can't tell if the analysts' estimates are credible or whether to trade on the "surprise" when actual earnings are revealed to be higher than the consensus.

A new study by Robert Gillam and Gregory Samorajski of McKinley Capital Management LLC and Leigh Drogen of Estimize found ways to avoid the estimate trap and trade on earnings profitably. Some of their conclusions are counterintuitive: The best opportunities ahead of earnings announcements are when sell-side analysts are too optimistic, not too conservative. And once earnings have been reported, it's often better to buy companies that missed the consensus analyst estimate, not the ones that beat it.

McKinley Capital is an international money manager based in Anchorage, Alaska, that uses quantitative tools to pick securities. Historically, the firm, which has about $7 billion in assets, has tried to identify mispriced stocks by comparing consensus estimates with the consistently more accurate estimates of certain individual analysts. Drogen's firm, Estimize, generates "crowdsourced" earnings estimates on more than 2,100 U.S. stocks from more than 50,000 contributors.

The authors compared the consensus estimates from sell-side analysts against the crowdsourced Estimize consensus over a six-year period, finding more than 3,400 divergences of more than 10 percent. While they did confirm the conventional wisdom that money can be made on lowball estimates from sell-side analysts, there are even better trades to be had.

Part of the study looked at stock performance in the days and weeks before earnings announcements. In cases where the Estimize consensus was lower than the sell-side consensus, prices tended to drop. Short positions on those companies generated an excess return of as much as 2.75 percent over the S&P 500, skewing higher for larger-cap stocks. Gillam, Samorajski and Drogen describe those results as "economically meaningful and statistically significant."

The other part of the study, focusing on price movements after earnings are reported, found that it's more profitable to buy small-cap stocks that missed the sell-side consensus than it is to buy those that beat estimates. On average, those "misses" generated returns of 1.83 percent 20 days later, whereas the "beats" returned only 1.25 percent.  

Why do analysts get it so wrong? Drogen cites herd behavior, regulatory compliance and the pressure to please investment-banking clients. That's why "forecasting ability is rarely ranked among the most important characteristics of sell-side analysts."

McKinley says it increasingly plans to use Estimize data in its algorithms and expects they'll produce better returns.

To contact the author of this story: Erik Schatzker in New York at eschatzker@bloomberg.net.

To contact the editor responsible for this story: Josh Petri at jpetri4@bloomberg.net.