WHAT IS A DIVERGENCE?
A divergence occurs when two or more items: indices, prices, momentum, etc..., fail to confirm one another. In Dow Theory, you start to worry about a rally if the Industrials make a new high, but the Transports do not. Another form of divergence might be a new price high, but on lower volume. When I refer to divergence, I usually mean that although prices made a new high (low), momentum did not also manage a new high (low).

WHAT ABOUT MOMENTUM?
Of course, you need to understand momentum. Momentum is simply the speed with which prices are rising or falling. One common measure of price momentum is the Relative Strength Index (RSI), developed by Welles Wilder. A simple measure of momentum would be a rolling rate of change going back 14 days. I usually use RSI since it is bounded by zero and 100.

WHAT IS THE DIVERGENCE SHOWING ME?
Funny you should ask. I will use a recent example to highlight what I am talking about. Look at the chart of the Software HOLDR (SWH, $32.79) (see below). It shows prices from late-2001 until the current time period. Note how SWH made a fairly substantial new low in October 2002, but how momentum, as measured by RSI, did not even come close to its prior nadir. This means that although prices fell to a new low, in the last leg down, the sellers were not as aggressive as they were the first time down. This can be an early warning sign of a reversal in the trend. Note also that there is currently a very tiny divergence on the chart. SWH made a new closing high, but momentum is below where it stood two weeks ago.

SO, IS THIS THE HOLY GRAIL?
If it were, then I'd at least ask you to extend your subscription before I sent it to you! The truth of the matter is that no technical indicator works every time. Look again at the SWH chart. Prices made three new weekly closing lows before turning higher, and the losses were substantial. A momentum divergence is merely a warning. You still need to get real trend-changing action before acting, but divergences should certainly get your antennae up!

You can avoid trouble by not trying to anticipate the change in trend. Also, I have often found that if you draw a trendline against the supposed divergence, and if that trendline breaks, then there is a good chance that momentum will ultimately confirm, thus removing your signal.

Also, it's worth noting that divergences can last a very long time. The S&P 500's momentum peak occurred in June 1996 at 670.63, yet the index topped at 1552.87 nearly five years later!

Steven Poser
Editor
The ETF Authority
New York, NY