Portfolio Strategy and the Iron Laws

A few weeks ago, we observed the single most extreme syndrome of “overvalued, overbought, overbullish” conditions we identify (see Speculative Extremes and Historically Informed Optimism) at a level on the S&P 500 4% higher than the syndrome we observed in July. The S&P 500 has climbed about 1.5% further since then, and all of the features of this syndrome remain in place.

As I noted in December, except for a set of signals in late-2013 and early-2014 (when market internals remained uniformly favorable as a result of Fed-induced yield-seeking speculation), an overextended syndrome this extreme has only emerged at the market peaks preceding the worst collapses in the past century. Prior to the advance of recent years, the list of these instances was: August 1929, the week of the bull market peak; August 1972, after which the S&P 500 would advance about 7% by year-end, and then drop by half; August 1987, the week of the bull market peak; July 1999, just before an abrupt 12% market correction, with a secondary signal in March 2000, the week of the final market peak; and July 2007, within a few points of the final peak in the S&P 500, with a secondary signal in October 2007, the week of that bull final market peak.

Two weeks ago, we observed a fairly rare set of “crash signatures” that we associate with the risk of market losses in excess of -25%, generally over a period of about 6 months. No single variable drives these signatures. Rather, they capture infrequent combinations of market conditions that may include offensive valuations, dispersion across market internals, credit market weakness, lopsided bullish sentiment, Federal Reserve tightening, or other features which, in combination, have historically preceded steep and compressed plunges in the market. These signatures are designed to identify the most hostile points in a market cycle. The last three times these features were in place to the same extent were: Apr-Oct 2008, Mar-May 2002, and Aug-Sep 1987.

Last week, an additional class of risk signatures, typically active in only a small percentage of historical data, shifted to warning mode. To offer some idea of the risk profile we are estimating here, the chart below reflects a blend of several types of classifiers we’ve developed over the years. These include estimates of the expected market return/risk profile on horizons ranging from 2 weeks to 18 months, as well as signatures we associate with specific “events,” such as air-pockets, panics and crashes. The bars show points where a blended composite of these estimates was at least two standard deviations below average, as it is presently. There's no assurance that future outcomes will be similar, but most of these instances were rather challenging for investors.

Still, as I noted several weeks ago, it’s important to recognize that these are not forecasts. They are signatures of causes and conditions that create a permissive environment for what would otherwise be highly unusual events. The difference is that a forecast says “we expect this particular outcome in this specific instance,” while a classifier says “we identify the same signature of conditions that has regularly preceded this particular outcome in the past.” It’s a subtle distinction, but an important one. We needn't rely on forecasts. Rather, we continue to align ourselves with the prevailing evidence at each point in time, and our outlook will shift as the evidence does.

On the advancing half-cycle since 2009, and the completion ahead

Having anticipated both the 2000-2002 and 2007-2009 market collapses, and having advocated a constructive or even leveraged investment outlook after every bear market decline across three decades in the financial markets, we’ve come out admirably over the course of complete market cycles - the notable exception being the advancing half-cycle since 2009. It’s impossible to appreciate present market risks without recognizing the difference between present conditions and those that prevailed during most of the period from 2009 through 2014.

As I’ve frequently noted, our own difficulty in the recent half-cycle followed my insistence in early 2009, after a market collapse that we fully anticipated, to stress-test our methods against Depression-era data. The resulting methods improved the robustness of our discipline to Depression-like outcomes, but they also captured a historical regularity that turned out to be our Achilles Heel in the face of the Federal Reserve’s zero interest rate policy. In prior market cycles across history, the emergence of extreme “overvalued, overbought, over bullish” syndromes was regularly accompanied or quickly followed by a shift toward risk-aversion among investors (which we infer from the uniformity of market action across a wide range of securities and security types). Because of that overlap, these “overvalued, overbought, over bullish” syndromes, in and of themselves, could historically be taken as reliable warnings of likely air-pockets, panics, or crashes. Unfortunately, the Federal Reserve’s zero interest rate policies disrupted that overlap. With the relentless encouragement of the Fed, investors came to believe that there was no alternative to speculating in stocks, and even obscene valuations and overextended conditions were followed by further speculation. In the face of zero interest rates, it was necessary to wait until market internals deteriorated explicitly before adopting a hard-negative market outlook. We implemented that restriction to our approach in 2014.