W&L Professor Examines Stock Price Charts' Ability to Predict Future Patterns

Adam Schwartz, Lawrence Term Professor of Business Administration, Washington and Lee University

Adam Schwartz

Does a stock’s price chart tell us anything about how that stock will trade in the future? What would happen if you compared the stock price chart of a contemporary stock to historical stock patterns back to 1926?

Those were among the questions that Washington and Lee University business administration professor Adam Schwartz and colleagues from Auburn University and Virginia Military Institute wanted to answer with an experiment that tested so-called technical analysis — the method by which stock traders try to find undervalued securities by studying their price charts.

The result of their study is a paper that is forthcoming in the journal Financial Management. Jimmy E. Hilliard, Harbert Eminent Scholar and Professor of Finance at Auburn, and James C. Squire, professor of electrical and chemical engineering at VMI, authored the paper with Schwartz.

• Download a copy of the paper (pdf)

The study identified recent stock patterns for thousands of random companies.  The stock price patterns of the random firms were matched against a database of US stock prices going back to 1926. Using a pattern-matching algorithm, a "twin" or matching stock was identified. Once those twins were discovered, the next step was to test for whether or not the performance of those twins could predict how a stock might perform in the future.

"We began with a random stock pattern for a company over a five-year period. Then we ask if the pattern of that stock looks like the pattern of any other stock that has ever traded," Schwartz explained. "We start in 1926 and look at every stock that's ever traded using the CRSP (Center for Research in Security Prices) data base.

If the random stock were, say, Nike for a five-year period beginning in 2000, the computer program would search every possible stock’s five-year pattern back to 1926.   The program would save any stocks whose price pattern is comparable to that of Nike stock. Visually the patterns of the target stock and the matching stocks will look very similar.

"We know what Nike has done during a recent five-year period," Schwartz said. "We’ve found a matching stock that’s done very close to the same thing. We also know what the historical match did over the next two years. What we want to know is will the performance of the twin tell us anything about what Nike will do over the next two years?"

What they found, Schwartz said, is that the target stocks that had the best performing twins — the stocks that were in the top-performing 10 percent — continued to perform well outside the sample period.

"We tested 25,000 random 60-month stock patterns in the sample. After adjusting for risk, the 2,500 with the best performing twins outperformed the average stock in the sample by almost 10 percent over a two-year period," Schwartz said.  That's a pretty huge win."

Although the hope would be that stocks with twin performance in next 10 percent would come in second place, that was not the case. Decile Two came in third place. Instead, the second-best performance out-of-sample occurred from stocks with twins showing the worst average performance beyond the sample period.

"According to the efficient market hypothesis, no information from past stock price patterns will help us pick winning stocks. So technical analysts, who search for undervalued stocks using chart data, are wasting their time," said Schwartz. "When you have a lot of people doing something — that is, technical analysts using price data to find stock picks — you have to wonder why they keep trying. We offer a possible explanation.

"We had no preconception about what we would find. But we did this experiment 41,000 times using daily data and 25,000 times using monthly data to test our idea.  We found for both tests that the group with twin returns in the top 10 percent beat all the other groups out of sample.  We hope to extend our work in the future to test the value of information used by more main-stream Wall-Street analysts, such as earnings growth patterns."

News Contact:
Jeffery G. Hanna
Executive Director of Communications and Public Affairs
(540) 458-8459

This entry was posted in Faculty, Press Releases and tagged , .
Bookmark the permalink.
Log In | Log Out