The greatest miscalculation in my career as an investor has been to underestimate the lengths to which the miscalculation of speculators can extend. I’ve had to correct that error twice, and even if the completion of the market cycle ultimately made the error irrelevant, the challenge was excruciating in the midst of the market cycle, at least until it was fully addressed. The fact is that valuations drive long-term returns, but over shorter horizons, stock prices are the result of whatever investors collectively believe, however reckless or detached from historical evidence those beliefs may be. As long as enough market participants are attached to the idea that risk is their friend (or enemy) regardless of the price, there is no natural limit to how overvalued (or undervalued) stocks can become. There is only one way to address this: measure investor risk preferences directly through observable market internals. Don’t expect an overvalued market to crash until internals deteriorate; don’t embrace an undervalued market too aggressively until internals improve.
It’s a lesson that value-investors have learned and re-learned throughout history. Even the legendary value investor Benjamin Graham discovered it, in his case by becoming constructive far too early during the market collapse of the Great Depression. The collective risk preferences of investors rule the short run, but valuations ultimately rule the long run. Graham famously summarized that lesson in one sentence: “In the short run, the market is a voting machine but in the long run, it is a weighing machine.”
Investment successes and stumbles largely mirror how closely one's discipline captures that framework. For example, by the mid-1990’s, I had established a reputation not only as a value investor but periodically as a leveraged, raging bull. Unfortunately, my value-conscious bent left me unprepared for the tech bubble, as valuations moved beyond every post-war extreme, and eventually beyond those observed at the 1929 peak. Though value investing had served me well over time, something was missing.
As the tech bubble continued, I shifted my thinking. If overvaluation itself was able to bring the market down, then stocks could never become severely overvalued in the first place. From a research standpoint, the right question wasn’t “how much more overvalued can stocks become?” but rather “What distinguishes an overvalued market that continues to advance from an overvalued market that collapses?”
The feature that distinguishes an overvalued market that continues higher from one that collapses is the preference of investors toward risk. Across a century of historical evidence, the most reliable observable measure of investor risk preferences proved to be the behavior of market internals based on prices, trading volume, and risk-sensitive securities. Put simply, when investors are risk-seeking, they tend to be risk-seeking in everything, which results in a measurable uniformity of speculation across a wide range of risky assets. The best indication of a shift toward greater risk aversion is a gradual deterioration in the “uniformity” of price action and the appearance of "divergences" across a wide range of individual securities, sectors, and risk-sensitive asset classes.
Fortunately, we learned the right lesson, instead of the wrong one that others “learned” during the tech bubble. Too many investors mistakenly “learned” that valuations didn’t matter and that the economy had entered some “new era” where old metrics no longer applied. But recognizing that the uniformity of market internals merely postpones the consequences of extreme overvaluation, I wrote at the March 2000 peak: “The inconvenient fact is that valuation ultimately matters. Trend uniformity helps to postpone that reality, but in the end, there it is… Over time, price/revenue ratios come back into line. Currently, that would require an 83% plunge in tech stocks (recall the 1969-70 tech massacre). The plunge may be muted to about 65% given several years of revenue growth. If you understand values and market history, you know we’re not joking.”
As it happened, the S&P 500 lost half of its value, and the Nasdaq 100 Index lost… well, 83%, by the October 2002 market lows.
Over the following decade, that distinction between a voting machine and a weighing machine was central to successfully navigating both bubbles and crashes. It allowed us to remain modestly constructive until September 2000 (noting the shift to negative uniformity in October 2000), to move back to a constructive position in early 2003 as a new bull market took hold, and identifying the shift back to hostile market internals in July 2007 as the market was peaking before the global financial crisis.