Noise

More than once on this blog I’ve written about things I know nothing about. At least I know that I don’t know, I vet my sources to be reasonably certain that they do know, though if they don’t I’m in trouble. And, of course, if you rely on my second-hand knowledge you’re also in trouble. Well, here I go again. This time I’m trying to glean some insights from the world of sound synthesis.

Many traders try to devise ways to filter out market noise to produce tradable signals, so I thought that the world of electronic music might offer some clues. When I found a paper that talked about Brownian motion, the random walk, and algorithms, I figured that it was worth summarizing ever so briefly on the off chance that one or more of my readers might have a eureka moment.

Market noise is, of course, not same as noise in a musical setting. Let’s start with a definition of the latter. In its broadest form, “noise is an audio signal that consists of an accumulation of sinewaves of all the possible frequencies in the hearing range and with all possible amplitudes and phase relations.” Noise stands in contrast to sound, which is a sinewave signal with a single frequency. Is it possible to filter noise to produce sounds that should be hidden somewhere in the noise? The answer is no, and not just for technical reasons. “When the frequencies are filtered out correctly the amplitudes still vary wildly, making it virtually impossible to create steady tones.”

But noise signals can be differentiated according to their statistical distribution profile. Although noise has no apparent pitch, “when each possible frequency has an equal chance of occurrence . . . it sounds like a very bright hissing sound” and is what we know as white noise. We can filter white noise to produce a dull red noise, a pleasant pink noise akin to the sound of a distant ocean surf, or a very dark brown noise that is actually derived from Brownian motion. But we can’t get sound.

We know that high frequency traders function in the world of noise and presumably act on algorithms that filter cacophonous market noise into actionable “colored” noise. Noise is not, however, the exclusive property of high frequency traders. There is market noise on all time frames. Normally traders are told to stay clear of noise, but perhaps it’s time to embrace noise and simply try to filter it. One standard piece of advice is to drop down to a smaller time frame. What is noise on a 60-minute chart might appear to be a clear trend on a 5-minute chart. But are we deluding ourselves? Can we really filter 60-minute noise in such a way as to produce 5-minute “sound”? If the analogy to electronic sound holds, what we see on the 5-minute chart is still noise. The lengths of 5-minute price bars “vary wildly”; there are no steady tones here. Perhaps what we describe as a trend is really colored noise.