Asymmetrical Information

From the desk of Mindles H. Dreck:

Mending Lines

Something there is that loves a line. That something in human nature also loves central tendencies. In all sciences, hard and soft, we seek linear relationships and "mean reverting" patterns in all events, man-made or "natural."

In all likelihood, any asset allocation advice you have received is driven by a mean-variance analysis of the past behavior of the market over some period. Theories of global climate change revolve around the idea that a "true mean" temperature has been identified, and recent trends reflect a linear deviation from that mean temperature. In social sciences you hear of "cycles" of public thought or behavior, as if we swing back and forth like a pendulum over some identifiable center. One of the silliest example is Arthur Schlesinger's Cycles of American History, in which he posits a 15-year cycle from liberal to conservative in order to conveniently wish away Reagan (I have mentioned this book before).

Why should we be able to explain behavior around us this way? Did god command that all the world exhibit linear behavior and central tendencies? Should all phenomena exhibit "normal" distributions?

I think the reason we like to believe the world works this way is because it seems to provide a measure of certainty. We are not comfortable that the universe, and time itself may have no beginning or end, and we are not comfortable that certain behaviors exhibit something less than perfect randomness..and yet far from perfect predictability.

In truth, measures of central tendency can be useful to describe behavior, but often give the user a deceptive comfort that events will unfold in a neat pattern towards their expected result. THis is a big problem when attempting to predict the dollars and sense outcome of market behavior. In fact, one of the problems with Markowitz analysis, which is behind most asset allocation analysis, is that it doesn't describe the messy path an investment might take to its expected return. That path can dramatically alter the amount of money available at the end of the investment period. The more volatile the investment, the more true this is. (and some guy has written on this).

John Allen Paulos provides a good hypothetical example of this problem in his column "Average Riches, Likely Poverty", in which he describes playing a hot IPO market:

Hundreds of IPOs come out each year. In the first week after the stock comes out, the price is usually extremely volatile. It’s impossible to predict which direction the stock price will move, but assume that for half of the companies’ offerings the price will rise 80 percent during the first week and for half of the offerings the price will fall 60 percent during this period.

The investing scheme is simple: buy an IPO each Monday morning and sell it the following Friday afternoon. About half the time you’ll earn 80 percent in a week and half the time you’ll lose 60 percent in a week for an average gain of 10 percent per week — (80 percent – 60 percent)/2. Note to the mathematically savvy: in this instance, you are allowed to average percentages because of the 50/50 split.

Ten percent a week is an amazing average weekly gain, and it’s not difficult to determine that after a year of following this strategy, the average worth of an initial $10,000 investment is more than $1.4 million! Imagine the newspaper profiles of happy day traders, or week traders in this case, who sold their old cars and turned the proceeds into almost $1.5 million in a year.

But what is the most likely outcome if you were to adopt this scheme and the assumptions above held? The answer is that your $10,000 would likely be worth all of $1.95 at the end of a year! Half of all investors adopting such a scheme would have less than $1.95 remaining of their $10,000 nest egg.

It's actually worse than Paulos suggests, because at some level of investment - say $1,000, you don't have enough money to continue to invest in the next IPO (imagine calling your broker for $500 of a hot IPO...). For more than half of the investors, it results in "game over".

An article by Malcolm Gladwell in this week's New Yorker (no link available yet, but here's an abstract) takes on these ideas, in a way. It describes a hedge fund manager, Nassim Taleb, who has supposedly built his investment philosophy around the observation that large, random movements in the market happen more frequently than our analysis tools predict. In the article's words, market behavior has "fat tails" - there is higher frequency at the extreme's than a normal distribution implies. Options are conventionally priced using the Black-Scholes method, which assumes a more normal distribution to the potential price outcomes. Mr. Taleb's fat tails suggest that options covering extreme contingencies are therefore underpriced. He is long out-of-the-money options.

Unfortunately for Mr. Taleb, this means he loses money most of the time. Taleb is OK with this because he knows, when the shit hits the fan, he's gonna make a lot of money.

Gladwell contrasts Taleb with his mentor, of sorts, Victor Niederhoffer. Niederhoffer is famous for making piles of money, but ultimately blowing up two hedge funds by primarily selling options into a large market event. Niederhoffer makes money most days, but puts himself out of business when those "fat tails" kick in.

As readers, we are asked to consider whether we would want to have our money with Taleb, who "never blows up but can only bleed to death" and promises to make a pile when the market goes haywire; or with Niederhoffer, who has made money consistently, except on the two occasions he lost it all.

Options are like insurance. They provide a payout based on a contingency, charging a premium for the privilege. My first reaction is that nobody makes money buying insurance (Taleb) and very few people make money selling it (Niederhoffer). In fact, almost every non-life insurance company in the world loses money on its insurance activities. Most of them only make it back on the float (the invested premiums/reserves). AIG is famous for maintaining a 99% "combined ratio", which means it makes all of 1% of its premiums in profit on pure risk underwriting activities.

My second reaction is that just because you understand a phenomenon, doesn't mean you have found a way to make money consistently employing your understanding.

Path risk is the problem with both strategies, namely you may lose all your money before you win the big one with Taleb, and you may be wiped out by Niederhoffer. So both Niederhoffer and Taleb base their investment theories on the mispricing of risk on average across the market, but must deal with highly volatile specific "game over" outcomes along the way to their expected (average) result. You pays your money and...

I'm quite sure there is more to the investment strategies of Taleb and Niederhoffer than Gladwell indicates. But I'm also glad to see the mainstream press attacking these issues in a lucid way.

Hey, Megan asked me what I thought.

UPDATE: Since there are some readers still bewitched by expected values and underappreciative of the distribution of potential outcomes, let me offer a few jokes to illustrate the point:

Two actuaries are out hunting. The dogs flush a grouse who flies directly over the happy duo. The first actuary fires and misses 10 feet to the left. The second actuary fires and misses 10 feet to the right. They look at each other, smile, high five and say "Direct Hit!"

An actuary has his head in an oven and his feet frozen in a block of ice. "On average", he says, "I feel pretty good!"

Posted by Mindles H. Dreck at April 23, 2002 10:20 PM | Technorati inbound links