Random thoughts of snow on Wall Street

I visited New York City a few years ago. I sat for over an hour as the plane was prepared and de-iced in the middle of what seemed to be unending snowfall. As I watched the snowflakes dance about outside my window, I was struck by their quick and random movements. Although a part of the same storm, each individual snowflake seemed to have a mind of its own.

The great composer Tchaikovsky captured this sense of randomness well in the “Waltz of the Snowflakes,” part of his monumental score for “The Nutcracker.” The seemingly haphazard sounds of various instruments invoke a lightness and surreal quality, as runs of harps, flutes, bells, and violins dart in and out of the listener’s ears.

In the more modern world we live in, a popular screen saver features a winter scene with a gentle snowfall. If one looks behind the scene to the coding of these visual masterpieces, the heart and soul of the imagery is a simple random-number generator. Each new number thrown out by the device sends the snowflakes darting in unpredictable directions, reinforcing the illusion.

For most of my 50-plus-year career as a financial analyst, I have been told that stock prices follow this same random behavior. When I started in 1969, the die was already cast. Randomness was already an underlying rationale for Professor Harry Markowitz’s 17-year-old mean-variance passive-allocation methodology that still holds sway over most financial advisers. It held that since stock price direction was not determinable, the best result could be accomplished by mathematically achieved diversification between asset classes.

Backup for this way of thinking arrived from others in academia. Professor Paul Cootner released a book called “The Random Character of Stock Market Prices” in 1964, followed by Eugene F. Fama’s 1965 paper “Random Walks in Stock Market Prices.” Soon after I entered the field, Princeton economist Burton Malkiel wrote his seminal work, “A Random Walk Down Wall Street.” The 1973 book is now in its 12th edition.

This formidable array of academics all reached the same conclusion. They all believed that the movement of stock market prices was, like the snowflake’s, random.

Despite its credentials, the theory has taken a lot of hits in the more than 50 years since it was first pronounced. While charts of daily returns over time do appear to approximate the single hump of a random curve, the devil has once again proven to be in the details. Many researchers following up on the early studies discovered on detailed examination that the so-called random curve of stock prices had two “fat tails.”

The existence of these “fat tails” meant that stock price data exhibited more negative and positive returns at the extremes of the curve than a curve would exhibit if the behavior was truly random. These fat tails provided both higher-than-expected opportunities and more devastating losses than index investors would anticipate if prices were truly random.

Then along came Warren Buffett. Following the work of Benjamin Graham and David Dodd in the 1930s, he demonstrated in real time that stocks could be chosen with results that consistently defied randomness. Then, of course, came the work of Professors Andrew W. Lo and Archie Craig MacKinlay, which questioned the very concept of randomness of stock prices and led to the inevitable counterpunch book, their 2002 “A Non-Random Walk Down Wall Street.”

Bringing the story more up to date, recent research from Eugene Fama (yes the same Eugene Fama that said stock prices were efficient and random) and Kenneth French demonstrated that certain factors could explain market returns. Soon there were three-factor, four-factor, and five-factor models. The pièce de résistance for active investors like me, however, came when French showed that momentum, or the price behavior of stock prices, might be the most significant “factor” determining stock market returns. (In the interest of full disclosure, he denied this reported mathematical conclusion when I talked with him personally a few years ago, citing the volatility of momentum strategies.)

The investment world had gone full circle, like a winter storm that just won’t advance. Stock prices were determinable, then they were random, and now it could be argued that they were the strongest factor in determining a stock’s return. Yet, while the theory had folded back on itself, the practice of most financial analysts remains locked in the 1950s world of Markowitz and its “random walk” assumptions.

I have had plenty of time to consider my airport observations on the randomness of snowflakes. I can’t help but notice that while the motion of individual snowflakes is random, masses of snow picked up by the wind are often blown (relentlessly and for long periods of time) in the same overall direction—much as individual stocks, once caught up in a rally or a correction, move together in the same direction, higher or lower.

I also notice that weather forecasters have the ability to nail the timing of storms. Snow begins to fall right in the middle of their projected window. A single snowflake’s movement is still random, but the path and timing of the storm that contained it was predictable.

But “predictable” does not imply “certainty.” When I flew to NYC a few years ago, the city was shutting down as it responded to a winter weather warning. Predictions were for a city shutdown the next day due to a huge snowstorm expected to arrive that Monday morning.

Needless to say, the snowstorm, while it enveloped much of the East, did not reach the city. Temperatures stayed up, and the worst bad weather we experienced was rain.

Stock market predictions and quantitative methods live and die by the same rules as weather forecasts. We listen to them because most of the time they are right, but we know that many times they are wrong.

It is important to remember that these methodologies are all about putting the odds on your side. Whether it’s about wearing your snow boots, gloves, and heavy winter coat or becoming defensive in a raging bull market, these systems use market history to discover rules to put the probabilities of success in your corner. As such, the results they can produce are anything but random.