The Stock Market's Feedback Loop

I have a computer science and physics background. This is what I think of when I think of this week's stock market activity.

Positive feedback

1. Why is positive feedback important? The key feature of positive feedback is that small disturbances are amplified. When positive feedback is present, there is some causal loop where a small change creates an effect that causes an even bigger change -- like a ball rolling down an increasingly steep hill.

2. Systemic risk is the risk that an amplification or leverage or positive feedback process is built into a system, this is usually unknown, and under certain conditions this process can amplify exponentially and rapidly lead to destructive or chaotic behavior. A Ponzi scheme is a good example of a positive-feedback system, because its output (profit) is fed back to the input (new investors), causing rapid growth toward collapse.

3. In sociology, a self-fulfilling prophecy is a positive feedback loop between beliefs and behavior: if enough people believe that something is true, their behavior makes it true, and observations of their behavior in turn increase belief. A classic example is a bank run.

I found it hard to believe that one trader could cause such carnage by simply typing a "b" instead of an "m".

I did not find it hard to believe that several competing computer algorithms might indirectly explore each other's black boxes and find neither willing to support prices. If that was indeed the case, then these competing computer algorithms would simultaneously come up with the same conclusion. Sell!

Black box

In science and engineering, a black box is a device, system or object which can (and sometimes can only) be viewed solely in terms of its input, output and transfer characteristics without any knowledge of its internal workings. Almost anything might be referred to as a black box: a transistor, an algorithm, or the human mind.

That is exactly how program trading works. What goes on within each firm's algorithms is a closely guarded secret.

One firm can't know what the black boxes of other firms are doing except by watching them. Watch a black box long enough and you'll start to figure out how they will probably react to changing inputs though.

Let's say there are two firms and two black boxes. We'll call them BlackboxA and BlackboxB. In the early days BlackboxA and BlackboxB were simple algorithms. They didn't know or care about each other's existence.

Some bright guy at Firm A decided that his black box could be smarter if it started watching BlackboxB. If BlackboxA knew BlackboxB would sell, then BlackboxA would sell first. This made BlackboxA smarter.

Meanwhile...

Some bright guy at Firm B decided that his black box could be smarter if it started watching BlackboxA. If BlackboxB knew BlackboxA would sell, then BlackboxB would sell first. This made BlackboxB smarter.

See the problem here? BlackboxA relies on information from BlackboxB. BlackboxB relies on information from BlackboxA. BlackboxA therefore indirectly relies on information from BlackboxA. It's a form of circular logic. That's bad.

Circular reference

Circular references can appear in computer programming when one piece of code requires the result from another, but that code needs the result from the first.

The entire set of functions is now worthless because none of them can return any useful information whatsoever.

A circular reference represents a big problem in computing.

Picture two cells (A1 and B1) in Excel each representing one of the black boxes above. You can type "=B1-1" into cell A1 in Excel. A1 now relies on the information from B1. There will be no complaints. Once you type "=A1-1" into cell B1 there's a big problem though. It makes no sense. Excel now has no way to determine what the value of either cell should be.

Microsoft cannot calculate a formula. Cell references in the formula refer to the formula's result, creating a circular reference.

You can still do it in Excel though if you turn on the maximum number of iterations feature. It will be amusing if you do. Every time the spreadsheet recalculates, the value of those two cells will get smaller (as each cell tries to be smaller than its neighbor).

Computers, Not Human Error, Likely Caused Market Meltdown

Market sources say the confluence of events likely triggered electronic selling programs. As that selling commenced, the human element of the markets quickly evaporated and computerized trading programs wantonly sold stocks at whatever price was available, creating a feeding frenzy.

That seems to make the most sense to me. It's just an opinion though.