The Globe and Mail

  Sometimes stocks will move by sheer luck alone


Saturday, October 28, 2000

It's amazing how every little bump and dip in the stock market demands an explanation. There are plenty of good reasons on offer: Portfolio window dressing, relief rallies, profit taking, housing starts, employment reports et cetera are all part of a lexicon to provide justification for the day's market action. But you will never hear an analyst say, "The market went up 50 points due to a random statistical blip." The idea that chance is a factor in stock prices seems to fly in the face of common sense.

Thomas Gilovich, a professor of psychology at Cornell University in Ithaca, N.Y., illustrates the non-intuitive nature of randomness to his students this way. He asks each member of his statistics class to write down a mock series of 20 random coin flips and to represent that series on a piece of paper using X's and O's to designate heads and tails. One student is told to actually flip a coin and write down the results.

Mr. Gilovich is always able to ferret out the one with the real-life coin flips. He is able to do so because his students always underestimate the likelihood that chance will result in a bunching of heads and tails. Like most people, they imagine that a 50-50 probability will even out in a fairly small sample. (Try this for yourself. We did and got XOXOXXOXXOOXXOOXXOXX, a 12-8 majority to heads!)

Our daily lives are peppered with such misunderstandings of the effects of randomness, and we insist on seeing patterns where none exist. A particularly pervasive example is in political polling.

These surveys have become more and more common since the days of George Horace Gallup. In this year's U.S. presidential election, pollster John Zogby is producing a new poll each and every day from Sept. 29 until election day. And every day this close contest carries the same disclaimer, "Since the daily tracking poll began, the race has never been outside the survey's statistical margin of error of plus or minus three percentage points."

Okay, so the election is too close to call. But it doesn't stop there. The daily headlines from sponsors MSNBC and Reuters rhapsodize like sportscasters calling the Queen's Plate. Every nuance of the candidates' performance is examined and a line is drawn to today's numbers. Al Gore sighs too loudly and George W. Bush goes in front. Mr. Gore increases his lead with Soccer Moms and closes the gap.

Every one of these embellishments can be more plausibly explained by simple random distribution. A tiny wobble here and there and the script would be completely different. It gives journalists something to write about, but it is no more relevant than the chat on the psychic channel. At least there the disclaimer at the bottom of the screen is clearer: "For entertainment purposes only."

The ebb and flow of the business cycle and investor psychology create many patterns that are useful to investors for stock selection and helping to time buys and sells. The enormous challenge is to eliminate all of the quirks of normal statistical randomness and focus on those connected to fundamental trends. That's easier said than done, for if these could be easily separated, making money would be no harder than printing tide tables.

By paying attention to the "signal to noise" ratio when absorbing information, it is easier to filter out the meaningless jabber and concentrate on the fundamentals. Avoid the tendency to overanalyze today's news, and remember, sometimes things just happen, because they happen.