Image Credit: Crowd outside the New York Stock Exchange, October 1929, Library of Congress
By Jason Zweig | Oct 27, 2017 11:08 am ET
At the end of October, the stock market crashed.
By Oct. 26, the Dow Jones Industrial Average had already fallen 13% for the month. On Oct. 28, it dropped 13% more. On Oct. 29, it collapsed another 12%.
That crash was in 1929, of course, not 2017. Did anyone see the Great Crash coming, and what can we learn from looking back?
Then, as they still do today, market pundits claimed to have seen a crash coming. Speculator Joseph Kennedy, later the first head of the Securities and Exchange Commission, is said to have gotten out of the market when he noticed a shoeshine boy bragging about his stock picks.
One forecaster, it might seem, did call the crash almost perfectly. In his book The Stock Market Crash — and After, published in 1930, economist Irving Fisher credited Karsten Statistical Laboratory of New Haven for demonstrating mathematically that the stock market was up to 25% overvalued by early 1929.
Fisher, then the most eminent economist in the U.S., remains notorious for having proclaimed in a speech in mid-October 1929 that stock prices had reached “what looks like a permanently high plateau.” In his book on the crash, Fisher declared that with hindsight, “it is easy to appreciate” that Karsten’s numbers foretold a crash.
Karl G. Karsten, the statistician behind those forecasts, wasn’t so sure.
In a book, Scientific Forecasting, published in 1931, Karsten ripped his own prediction methods to shreds.
He had analyzed data back to 1866, Karsten wrote, but his forecasts had two fatal flaws. First, the stock market’s huge rise in the late 1920s swamped everything that had come before, blurring the guideposts of the past. Second, market psychology was “a potent factor and one which no statistical series could be found to reflect in advance.”
So Karsten renounced the work that Fisher had praised. And Karsten laced his book with warnings about placing excessive confidence in any forecasts, including his own.
Diving even deeper into the data, however, he made discoveries that anticipated many of the ideas behind hedge funds and so-called smart beta, or mechanical strategies to earn excess returns, that are so popular today.
By 1928, Karsten was investing based on his theories. He wasted two years trying to combine subjective judgment with statistical analysis, but on Dec. 17, 1930, he launched a small fund run with real money and nothing but math.
Under what Karsten called “the hedge principle,” his “Demonstration Fund” appears to have bought the three biggest stocks in the industry sector whose share prices had been rising the most; at the same time, it sold short, or bet against, the rest of the stock market. The fund rotated from one sector to another based on whichever had the best price momentum.
By June 3, 1931, Karsten wrote, the Demonstration Fund was up 78%, net of trading costs. The Dow fell 21% over the same span.
Karsten warned, presciently, that techniques like his couldn’t work if too many people tried them. Such an investing approach, he wrote, “can be used only by a limited amount of capital when a very much larger amount of capital is ignorant of this system, and willing to be exploited.”
Born in 1891, Karsten entered college before his 16th birthday, graduating from the University of New Mexico in 1911. He was a Rhodes Scholar, did graduate work at Columbia University and, in 1917, founded Karsten Statistical Laboratory.
It isn’t easy to say what happened to the Demonstration Fund. Walter Friedman, a historian at Harvard Business School who wrote about Karsten in his 2014 book Fortune Tellers: The Story of America’s First Economic Forecasters, says Karsten’s papers at the Library of Congress are so moldy that Mr. Friedman had to handle them wearing special gloves and a respiration hood. The Demonstration Fund seems to have dwindled by 1937, and there doesn’t appear to be any record of a fund after 1942.
Karsten also wrote an unpublished novel, “Horse in a Limousine,” about a future so prosperous that even horses would get to ride in chauffeur-driven vehicles. He died in 1968.
In the essay “Characteristics of Scientific Method,” published the same year as Karsten’s book on forecasting, the great philosopher Bertrand Russell pointed out a paradox. “All exact science is dominated by the idea of approximation,” wrote Russell. “It is characteristic of those matters in which something is known with exceptional accuracy that, in them, every observer admits that he is likely to be wrong, and knows about how much wrong he is likely to be.”
If Karsten were here today, I think he would have two warnings for investors.
When someone uses historical data to forecast a cataclysmic market crash with near-certainty, bear in mind that the patterns of the past may no longer hold. And remember that trendy strategies like “smart beta” are likely to work best when most investors doubt they will work at all.
Source: The Wall Street Journal, http://on.wsj.com/2gGEvmy
For further reading:
Definitions of BACKTESTING, CRASH, FORECASTING, OVERCONFIDENCE, PANIC, RISK, in Jason Zweig, The Devil’s Financial Dictionary
Chapter Four, “Prediction,” in Jason Zweig, Your Money and Your Brain
Walter A. Friedman, Fortune Tellers: The Story of America’s First Economic Forecasters
Bertrand Russell, The Scientific Outlook
Frederick Lewis Allen, Only Yesterday: An Informal History of the 1920s