In a 2001 Fortune story, Warren Buffett called this metric "probably the best single measure of where valuations stand at any given moment." The so-called "Buffett Indicator" has been flashing red for so long now as stocks keep soaring that you might think it's an out-of-date relic that's worthless in forecasting where the market's heading.

Think again. In the past, when the S&P's market cap leaps far above this benchmark, it's always retreated back to the normal range. This time, prices are flying free of the Buffett Indicator at altitudes never before witnessed. That doesn't mean the formerly reliable yardstick is finished. More likely, it signals that the crazier the party, the worse the hangover.

The Buffett Indicator measures the ratio of stock market's total value to U.S. economic output. The idea is that abnormally huge valuations relative to GDP can only stay at those levels, let alone wax from there, if future earnings also soar to devour a much-bigger-than-normal slice of national income. That can happen for a few heady years. But over a longer span, outsized profits trend back to their traditional share of GDP. As the late economist Milton Friedman used to tell me, "Over time, corporate earnings can't absorb more their traditional share of national output in our competitive economy." When that "reversion to the mean" in profits occurs, valuations also drop.

We've witnessed that pattern many times. The difference now is that stocks have never defied the Buffett rule for so long, nor gotten so incredibly out of the whack with this heretofore gravity-like force. Since the S&P 500 companies hold 70% of the value of all U.S. equities, we'll use it as a proxy for the overall market cap. For the past quarter century through the close of 2020, measured at year end, the median S&P market cap to GDP is 85.65%. The number ranges from 54% in 2008 in the depths of the Great Financial Crisis to 131% at the apex of the tech bubble in 1999. From the end of the dot.com craze, the ratio didn't reach 100% again until the close of 2016.

From there, the Buffett Indicator took off, jumping to 122% in 2019, 147% in 2020, and as of mid-day on April 5, with the S&P hovering at a record 4071, 156%, based on the CBO's GDP forecast for this year. Since 2016, the S&P's total valuation has risen 100%, far outstripping the 19% increase in national income projected through the end of this year. The current 156% number is 80% above the quarter century median, and almost a fifth higher than the ratio at the peak of the tech frenzy.

How has the S&P fared in future years when the Buffett Indicator hit its summits? In the five years from the end of 1999 to Q4 of 2004, the S&P 500 slid 17%. From a relative high point of 90% at the close of 2006, it dropped 11% over the next half decade.

But when the Indicator hit 100% in 2016, the usual reversal didn't occur. Since Q4 of that year, the S&P has exactly doubled, lifting the ratio of valuation to national output to that incredible 156%. Today's market features plenty of yardsticks pointing to inflated prices that would require a big drop to shift back between the normal guard rails. The Buffett Indicator may be the scariest. Even if it fell to the previous, tech bubble peak, shares would fall 16%. A return to 100%, regarded in the past as a sign of a market too-pricey-to-last, would mean a decline of one-third to 2700.

It was a great time to buy stocks in 2008 when the ratio was 54%, one-third of today's reading, or in 2011 at 72%. Get ready for what the novelist John O'Hara used to call the day-after-bash "sermons and soda water."