People buy stocks in anticipation of future profits. The stock market is, therefore, a leading indicator for the economy. In a similar vein, brokerage stocks are a leading indicator for the stock market. This is because profits at brokerage houses increase during bull markets because of higher volume (profitable customers are more active) and more underwritings (companies go public when the price is right).

One of my favorite arrangements in my eSignal application is to plot two 60-minute charts in a stacked format. At the top, I display the S&P Composite and, underneath, the world's largest brokerage firm -- Merrill Lynch (MER). You could substitute the Amex Brokers Index ($XBD) instead. The two charts can be conveniently and easily saved as a layout in the eSignal application.

A lot of the time, these series are moving in concert, which means that this arrangement is not telling us anything. It is when divergences arise that we obtain some useful signals.

A classic example developed at the October 2002 bottom.

The screenshot shows that the S&P made its low on the 10th of October, but MER did not confirm. This represented a positive warning and indicated that the technical position was not as weak as you might think. Then, note that MER broke above a down trendline in the final two hours of trading. Not only was MER not confirming the S&P on the downside, but it was also now leading on the upside. The S&P then confirmed with a similar breakout. The confirmation came at the opening of the 11th when the index also violated its down trendline.

The first screenshot shows that both series rallied in gear for the balance of the chart, indicating that we should hold on to our long S&P position.

The second screenshot shows that, on the 2nd of November, MER violated an 8-day up trendline, which was later confirmed by the S&P. In this case, there was no actual divergence, so the joint trendline break acted as the sell signal.

This relationship is not perfect, but it is amazing how often it ferrets out false rallies and reactions in the S&P and offers advance warnings of impending trend changes. The approach is not limited to hourly charts. I have used it successfully for many time intervals from 10 minutes to daily and weekly.