Picking Industry Winners

Publish date:

Dramatic outperformance over a one- to three-month time horizon usually involves a significant industry bet. Although investment portfolios focusing on the long term can perform well without active industry selection -- the market is by definition neutrally weighted within industry groups, and the five-year performance of the

S&P 500

has been pretty spectacular -- it's tough to outperform over the short term without an industry overweighting.

Picking the right industry groups isn't easy. In 1990, a heavy overweighting in the banking industry almost gutted

Sanford Bernstein

, one of the best active money managers in the country. Bernstein had overweighted the banks because they looked cheap on long-term earnings power. They were dead right. First dead as their portfolios underperformed by more than 25% during 1990, then right as they held their position and rode the wave back up. Nowadays, many institutional investors avoid industry bets like the plague, keeping weightings within strict relative guidelines.

To punt on an investment call, however, is neither my job nor my Inclination. I have spent considerable time trying to develop a quantitative strategy for active industry selection and have created a database of 77 industry groups going back to 1978. It includes aggregate prices, earnings estimates and fundamental data on a monthly basis for all companies in the

Russell 1000

, including dead companies, over the last 20 years. Using this database, I have tried to find a quantitative strategy for industry selection that consistently adds value.

My backtests focused on the internal dynamics of prices and earnings estimates, rather than conceptual trends or economic data. Oil prices correlate directly with energy industry performance, for example, but in the database, the impact of oil prices shows up only within analyst estimate revisions. The Internet is today's dominant growth theme, but shows up only as price momentum for






and by extension the software industry.

The most important finding was that price momentum strategies worked far better than value or earnings revision strategies for industry selection. On a systematic basis, buying industries that were cheap -- by any definition -- was a losing strategy over the last 20 years. Earnings revision also added no value for industry selection. As the quant at a shop selecting companies through a combination of value and earnings revisions, this did not make me happy.

Price momentum strategies, on the other hand, did add value. I focused on breakout strategies, where the relative performance line made a new high at the end of the month relative to the historical range. This strategy worked best when the price breakout was measured on a very short-term basis, relative to the last couple of months, but the turnover was an uncomfortably high 328%.

If the breakout period was lengthened beyond the last couple of months to a full year, the turnover went down to 60%, but the excess returns decreased substantially. In other words, the further extended the price breakout, the worse the long-term performance for the price momentum strategy. As you can see in the chart below, buying industries that made new highs relative to their two-year price history actually resulted in below-market returns. The columns in red below show the excess return for buying industries with price breakouts, using a collar for the breakout of two months to two years.

A more stable strategy was afforded by focusing on relative price to earnings breakouts. The rationale behind this strategy is that investors are great at picking industry groups. In fact, they are so good that a rise in prices accompanied by falling earnings should be interpreted as a positive sign. The intelligent stock market is anticipating a turn in fundamentals in the months to come.

If you have a contrarian bone in your body, the statement above sets your teeth on edge. It is, however, supported by the data. As you can see in the chart below, buying industries with breakouts on relative P/E was a better long-term strategy than simple price breakouts, with higher excess returns and more stability.

Ted Murphy operates the

MarketPlayer Web site. Prior to MarketPlayer, he was a partner at Equinox Capital Management. Under no circumstances does the information in this column represent a recommendation to buy or sell stocks. While Murphy cannot provide investment advice or recommendations, he welcomes your feedback, emailed to