View Full Version : Another reason markets look random...

12-20-2001, 05:45 PM
Suppose you wanted to learn to time the market. You might look at all the days it went down, and see what news caused it. Then you might wait for that same news to happen, and sell.

But the news reported is not just a product of 1) in what direction the market went - as everyone knows - but also, 2) of what people want to look at.

Every news outlet knows it's their job to provide an answer to the question "Why did the market go down?" And the answer is never, as the legendary speculator (the original Baron Rothschilde?) said, "Because there were more sellers than buyers."

If there are two competing news outlets, the one that throws some "hidden" reason at you will get your subscription. Moreover, if people have, after a series of such events, "learned" that a particular peice of data - whether it be Cisco earnings, semiconductor book-to-bill, or the "napalm" number - is often the "cause" of a market decline, the first place news outlets will look for a purely coincidental scapegoat is that piece of data.

If they give you a piece of data that says the market should have risen, they go about of business. If they lead with a piece of data other than the one you are on the lookout for, they will go out of business. In other words, not only do news outlets know what news you have learned to look for, but also what news you have learned to look for is itself originally a nonsense.

It's like when I lived in Ft. Lauderdale, my neighbor was a basketball handicapper. Each week, he'd get X new clients, and tell half of his clients one team, and the other half the other. Each week, he'd lose, at most, half his clients. The 1/32 of clients who hit five times in a row would never leave.

So when you go to the Yahoo stock screener, or something, understand that the data you have learned to look at, and the data they choose to report as relevant, is nothing more than the product of this same sorting mechanism by which my neighbor, Doug, ended up with a certain group of clients.

And yet, there are still people who try to watch CNBC, and read the WSJ, and time the market based on making some sense out of the news, or P/E ratios or something. And then when they fail they think the market is random! They have failed to realize that the refined subset of news availabe to them, and what they think to look at, is not the beginning, but rather the very end of a process of statistical winnowing.

Sure, people think they have found a pattern. And they have! But the pattern has been manufactured for their consumption - because they are looking for it - and is utterly divorced from anything going on in the stock market! People who think the market is random have never really tried.


01-01-2002, 01:38 AM
Good post! I agree. But I find it hard to fathom how you can accomplish the short term plays you describe. Add up all the transaction costs and I think you have a hard time beating an index. Good luck anyways.

01-01-2002, 12:29 PM
Your comment reminds me of one of the better known of these S&P system day-traders.

He had, maybe, two or three coefficients on his breakout system. He'd buy when the market got to X times the previous day's high plus, like, Y times the previous day's low, plus like Z times the close minus the low - or something.

He eventually got to so many millions of dollars he had to move in and out of the market when these numbers crossed, that the best he could try to do was try to get his position to match his "theoretical" within a five-minute window.

Needless to say, his returns flattened out. But he still didn't lose money.


01-13-2002, 06:01 PM
Let me guess...he did this during a bull market. Lot of bull market geniuses out there.