Despite extreme valuations, investors’ fear of missing out is looking increasingly desperate. In market cycles across history, that has been an unfortunate impulse. On Monday, the major indices “gapped” higher on relief about the first round of French elections, with another opening gap for the Dow on Tuesday. A gap is a point where the opening price on a given day is materially higher than the prior day’s closing level. The stock market also gapped higher on March 1, which marked the record closing high for the S&P 500 to-date, as the banner headline on the CNBC news ticker read “Best Move: Just Buy Everything?”
Gap openings are fairly common when a stock or the broad market breaks out from a significant low, or when crossing above some widely-followed moving average. These “breakaway gaps” are often on high volume, and reflect the coordinated belief among some set of investors that the overall trend has reversed.
In contrast, when gaps occur after an extended advance, they are called “exhaustion gaps,” and they can signal desperation among investors to chase the prevailing trend. For individual stocks, these gaps are sometimes just temporary areas of consolidation. For the overall market, however, they tend to have far more hostile outcomes, particularly when they are associated with record highs, rich valuations, lopsided bullish sentiment, and deterioration in the uniformity of market internals.
Last week, Bill Hester walked into my office with a version of the following chart, which he conceived while examining data on gap openings last week. The sheer simplicity of the conditions is striking, relative to the market outcomes that typically follow. The chart shows all days where the opening level of the S&P 500 was at least 0.5% above the prior day’s closing price, and the S&P 500 was within 2% of an all-time high. In some cases, those two conditions briefly preceded the final market high. In others, including August 1987 and October 2007, they occurred within a few days before or after the market peak.
Bill and I wondered whether the same regularity held for the Dow Jones Industrial Average. One problem, unfortunately, is that opening data for the Dow becomes increasingly spotty the further one goes back in history, and often disagrees between data providers. Another problem is that as a price-based index, daily movements in the Dow can be slightly erratic because they are strongly influenced by its higher-priced members. Still, the question was interesting. Despite the spotty opening data, it's clear that if the daily low of the Dow Industrials is at least, say 0.25% above the prior day’s high, there must be a gap as well. The chart below shows all of the points where this occurred within 2% of a record high. Clearly, the criteria are too tight. They only capture two instances in market history: a series of three daily signals in August 1929, and the signals we’ve observed in recent weeks.
A looser condition is to look for points where the daily low of the Dow Industrials is at least 0.25% above the prior day’s close, without requiring that the low be all the way above the prior day’s high. That condition captures several additional instances, but a handful of these are false signals, invariably occurring at points where “all time high” wasn’t really synonymous with rich valuations or lopsided bullish sentiment. So the chart below tightens the criteria slightly, showing all days where the daily low of the Dow Industrials was at least 0.25% above the prior day’s close, the Dow was within 2% of an all-time high, the Shiller P/E was above 18, and advisory bullishness was greater than 50%
Just a note on the choice of those additional criteria: we certainly think there are better measures than the Shiller CAPE, but it’s widely followed, and we’ve regularly used the 18 level in the past as a rough measure of overvaluation - the current level is close to 28. For bullish sentiment, a level of at least 47% (Investors Intelligence) tends to be an effective threshold of “overbullish” across history, but 50% seems less arbitrary. Those levels of bullish sentiment are usually associated with bearishness in the 20% range, the rest being in the correction category. We impute sentiment data prior to the 1960’s.
Below, it’s clear that exhaustion gaps, particularly in the context of overvalued, overbullish conditions, are generally not friendly indications.
The chart below combines the S&P exhaustion gaps with those for the Dow Industrials. Given such simple criteria, the outcomes are rather disturbing.
To offer a sense of the market return/risk profile that has typically been associated with exhaustion gaps at overvalued, overbought, overbullish extremes, the chart below shows the maximum gain and maximum loss in the market as measured from each instance to the subsequent bear market low. Multiple exhaustion gaps in the same market cycle are depicted separately.
I recognize that my regular comments about the likelihood of the S&P 500 losing half or more of its value over the completion of this cycle may seem preposterous. A review of market history may help to understand these expectations, which are consistent with both the valuation evidence later in this comment, and with the outcomes that have typically completed prior speculative market cycles.