Wednesday, February 11, 2015

Breadth Divergences: More Noise Than Useful Signal Of A Market Top

It's conventional wisdom that new highs in indices should be confirmed by an expansion in breadth. In other words, you want to see the number of stocks trading above their moving averages expand as the index price moves higher.

That sounds intuitive. All else equal, having an increasing number of stocks above, say, their 50-dma as the index moves higher means that the index is supported by a growing number of stocks that are in an uptrend. Think of it as a wide foundation to support the index.

We find there is a practical problem with this, however. The index has better performance looking ahead when breadth is weak.

Let's take three situations: one where the percentage of SPX constituents above their 50-dma is 80% or more (strong breadth), one where it is less than 80% and one where it is less than 50% (weak breadth).

If greater breadth was a positive, then SPX should do best when 80% or more of its constituents are trading above their 50-dma. In fact, that is when the index does worst. SPX does best when breadth is weakest.



In the chart above, we show returns in SPX 10 days later when breadth is strong, medium and weak (middle two columns). SPX gains only 0.1% and is higher 62% of the time when breadth is strongest; but SPX gains 1.3% and is higher 75% of the time when breadth is weakest. These results look back over the past 3 years and come from indexindicators.com.

Within those 10 days (far right two columns), the average gain is less than the average loss when breadth is strongest (0.7x); but the average gain is much bigger than the average loss when breadth is weakest (2.4x).

In the next chart, we visualize these results. Highlighted in yellow are times when breadth is considered strongest. SPX is often near a top.



And in the next chart we have highlighted times when breadth is considered weakest. SPX often is near a bottom.



In summary, while it sounds intuitive that better breadth should be a positive, the returns and probability of a successful long are best when breadth is the weakest. Going long when the index is "oversold" is often a winning strategy.

All of this is related to the often mentioned "negative breadth divergence." The conventional wisdom is that investors should be wary when fewer stocks are trading above their averages. The foundation of the market is weakening.

The practical problem is that these negative divergences can persist for many months and conclude with indices continuing higher. Or there are divergences in, say, the 50-dma but none in the 200-dma. Which matters more?

The chart below compares SPX (top panel) with the percentage of constituents above their 50-dma (middle panel) and above their 200-dma (lower panel).



Let's look at four difference situations in the chart above.

In May 2006, there were big negative divergences in both breath indicators, with both near 70% (red arrows). SPX lost 8%.

In October 2007, SPX was on the verge of a 60% drop (green arrows). It was a major top. Yet 85% of the index was trading above its 50-dma. There was no divergence. The 200-dma had been diverging lower for 8 months and, in hindsight, was a terrific tell of trouble.

In April 2010, breadth was great: over 90% of the index was above both its 50-dma and 200-dma (blue arrows). Nevertheless, SPX dropped 15%.

In May 2011, SPX was on the verge of a 20% drop (orange arrows). This time, breadth in the 50-dma was slightly diverging; 80% of stocks were above their 50-dma. The 200-dma showed no divergence at all: over 90% of stocks were trading higher.

There is, in short, a practical problem with weak breadth: a fall in the index can take place when either the 50-dma or the 200-dma has a negative divergence. Or the drop can take place when neither has a divergence. Or, finally, divergences can last for many months, as they have since 2012 with only minor corrections taking place. Waiting for the downside has been a bad bet as the indices have moved substantially higher. Spotting a divergence in hindsight is easy but there seems to be no useful way to act on a divergence in real time.

Our conclusion is that it is not actionable to track these divergences. Instead, it is better to use breath as sign that the index may be overbought or oversold. The table at the top of this post is one example of how this can be done but there are doubtless others.


If you find this post to be valuable, consider visiting a few of our sponsors who have offers that might be relevant to you.