Tuesday, September 05, 2006

Patterns in Minds and Markets

What seems significant can be purely random, and sometimes what seems random is quite significant.

An interesting discussion on a mailing list rekindled my memory of having tried a little experiment. I went to a website that performs runs tests on number sequences to test whether or not those sequences are random. I did my best to construct a random series of numbers, moving my fingers at will around the keyboard. The test found that the number sequence I entered showed probable evidence against randomness. In fact, my best effort at being random was significant at the .03 level.

Later, I entered the daily price changes for the S&P 500 Index for August, 2006, a period of seeming positive trend. Nope. Little or no evidence against randomness.

It turns out that human beings are quite patterned in their efforts to produce randomness. Their random sequences have fewer runs of numbers than are found in true randomness. If we're tossing a fair coin, for example, we should get runs of five consecutive heads about 3% of the time. Interestingly, my attempt to generate a random sequence with 1's and 2's didn't even have runs of three (which should occur about 12.5% of the time). It was statistically significant at the .01 level!

What does this mean for trading? Quite simply, unaided human judgment is poorly positioned to detect what is significant and what is random. We see a string of rising prices and conclude it must be a "trend". Intellectually, we know that any pattern of six dots positioned on a piece of graph paper must form certain shapes occasionally as a matter of chance. Nevertheless, we persist on finding meaning in such "chart patterns".

We assume that sequences are significant and that randomness lacks sequence. And that is all wrong.

It's also the source of many psychological problems in trading. When we have sequences of winning trades, we view this as significant and conclude that we have skill and a hot hand. Not infrequently that affects our next series of trades. Conversely, when we have sequences of losing trades, we imbue that with similar significance and lament our slump. Often our next trades reflect our diminished confidence.

We'll buy a trading system after it's had a hot period--and then get rid of it after a series of flat or losing periods. After extended drops, traders reliably withdraw their money from Rydex funds and their equivalents; bullish sentiment rises with a rising market. The human mind seems wired to detect patterns, and even when we try to be random, we find ourselves patterned.

When asked about his cigar as a possible symbol of a sexual organ, Freud replied, "Sometimes a cigar is just a cigar."

And sometimes a pattern is just a pattern.

Our ability to detect patterns is at once the source of trading skill--and weakness.