©2006 All content on site protected by copyright
The Influence of Anecdotes on Markets, by Victor Niederhoffer
Reading through the book Chance in Biology by Mark Denny and Stephen Gaines, one sees numerous examples of randomness and anecdote in biological systems. One of the key exhibits in the book is how spider webs, which are pound for pound four times stronger than iron and can absorb three times as much energy as Kevlar, derive their properties. The secret is that crystals of ordered proteins are aligned and these are joined together by randomly arranged protein chains.
There is a comparable resilience and strength that randomness provides in markets. And on Father's Day, I feel I should provide some helpful examples of how this combination can be used properly and understood to everyone's benefit.
Let's start with one of the gravest fallacies out there today. The midterm elections are very bearish -- four-year bearish cycles in the years 1998, 2002, 2006 et al. thing. On the surface, if you look at the last four mid-year elections, and you stop in September, it's possible to come up with an average decline in those months that is somewhat improbably say, one in ten in isolation as follows:
Year First 9-months move Last 3-months move
2002 -30% 8%
1998 +6% 21%
1994 -1% -1%
1990 -14% 8%
1986 ... ...
That's it. But, but, but. How many degrees of freedom, how many implicit hypotheses are in there. There's the years mid two, the four years, the nine months and the stopping in 1990 knowing that the move in 1986 was +6% in the first nine months. You have four observations. Pretty sickly. Given that you knew that 2002 was a big down year, it was almost impossible not to find something that would show three of four up, a negative average chance with literally hundreds of hypotheses. And yet, this type of selective stopping and starting, retrospective picking of years that rhyme or sequence with the last bad years and selecting of tops and bottoms underlies so much of the work in using anecdotes to predict markets.
Let's turn to former Governor Poole, who has detected anecdotal signs that energy prices are creating spillover price increases in other markets. Yes, there are prices that are being increased. Are there any that are being decreased? For example, the 50% reduction that Intel announced on old microprocessors?
I am reminded of the company I had for sale that was doing everything right and had a million ingenious techniques of manufacture and consumer benefit for its tools. Sears was their main customer and I asked them: why not put through a price increase to them? They pointed out that because of Sears's 2000 stores, they couldn't accept a price increase without notice of at least six months before they could update the computer systems they all used. I asked them what about a price reduction. "Oh, that they can put through immediately."
I turn to quarterly earnings. It's been financial dogma for 25 years that when a company misses its forecasts, it performs worse in the future. But is this true any more? Many retail companies find that they have a bad mix of goods on occasion and they clear everything out on those occasions. I have a friend who runs what I believe to be the most successful retail fund of all time, and he points out that he makes money by buying the companies that report the bad quarters because it makes the comparisons much better in the future, when good new stuff comes into their stores.
Another example of the improper use of anecdote in markets comes when earnings warnings come. These are always much worse than expected for the company. But of course if you announce the worst things that are happening to any company, and you don't announce the good things, then you are always likely to be anecdotally noting the bad, and inferring that these ephemeral announcements for a given company are representative of the totality.
Gaines, in his magnificent and insightful book, uses the proper treatment of randomness, to explain the proper size of arteries, why horses hold their heads up, how scallops swim, what the extremes in sound are at a cocktail party, how to predict the size of waves, the life expectancies of jet engines, the limits of human longevity, and the likelihood of the next .400-hitter in baseball. The market person can use the techniques of randomness to explain the proper size of positions, why it is good to buy markets with positive drift, what the proper value of derivatives should be, what the size of likely declines and rises will be in a day, and the likelihood of the next 10% drop or rise in a market.
I would be interested in other examples of the improper use of anecdote in markets and proper remedies and techniques for dealing with it and I call on readers for mutual enlightenment.