PsyFi Search

Monday 21 September 2009

Sexual Trading

Contagion in Markets

It’s often remarked that stockmarket manias and panics are contagious, as though there’s some virus spreading through the markets, infecting the participants and causing their irrational behaviour. Of course, this is usually meant as a metaphor rather than something to be taken literally – it’s a nice conceit that there’s a disease “out there” causing us to all do stupid things all together.

However, humans are social animals, we live in social networks and networks are prone to attack at their weakest points. Models of how real and virtual diseases spread – the study of epidemiology – can give us clues as to what’s really happening when everything goes screwy. And it turns out that the more connected we are the more danger we’re in.

Rod Steiger’s Network

Most people are aware of the Kevin Bacon game which measures actors by how close they’ve come to acting with the prolific star of Flashdance. Someone who’s acted in the same film with Kevin has a Bacon number of 1, someone who’s acted with someone who’s acted in the same film as him has a number of 2, and so on. By tracing out this network researchers have been able to piece together the network of social connectedness amongst Hollywood actors. In fact it turns out that Mr Bacon isn’t anywhere near the most connected actor, and the thesps should actually be linked by the Rod Steiger number.

What’s really interesting, however, is what happened when the researchers looked at how the actors were connected. It turned out that the average actor has an average number of connections but that a few, key, thespians held the network together. These critical “nodes” had far, far more connections than most of their fellow luvvies. If you think of this as a network, then if you could somehow remove the key players then suddenly the whole thing would fall apart.

Scale Free Networks

It turns out that this is a decent model for the way most social networks link and, in particular, for the way diseases spread. Epidemiologists have a real interest in understanding this because they’re interested in figuring out how to prevent the spread of diseases like AIDS. When a bunch of researchers from Stockholm and Boston studied the Web of Human Sexual Contacts they discovered a network that looks an awful lot like that of the actors. Preventing the spread of sexually transmitted diseases turns out to be difficult because not all of the people have the same importance within the network: the especially promiscuous individuals – the critical nodes – were the key. Targeting these people successfully could stop the disease transmission, but a failure to do so would see re-infection occur again and again.

Networks like these are known as scale free, because there’s no pattern to how connected each node can be – to be precise, the distribution of connections follows a power law. It seems that the worldwide web has the same pattern. A few, very popular websites, are linked to with a frequency that’s way in advance of the average site. These sites are our critical points for disease transmission when some problem sweeps the information highway. And herein lies the next clue to the epidemiological model of stockmarket madness.

Markets Are Social Networks

Markets are inherently comprised of traders, agents communicating with each other, doing deals. Some of these agents are our supernodes, the critical people holding the vast majority of the network together. If these nodes get infected then we can surmise that all of their unimmunised contacts also get the virus. Which is why we have nothing to fear but fear itself.

Interestingly this research suggests an amendment to efficient markets theory rather than an alternative to it - a reminder that econophysics research like this comes from a background of rational theory rather than empirical observation. The development of scale-free networks from models of interacting traders has been observed by Tseng, Chen, Wang and Li, who posit a concept of zero intelligence traders in their models. This, of course, may be rather closer to the truth than they imagine, but the argument is that the scale-free nature of the network emerges from market structure, not trader interaction - aka, the Efficient Market Hypothesis.

On the other hand a paper from Rossitsa Yalamova suggests that extreme market events are a result of the coordination of trader activities due to synchronisation of trading rules through a relatively small number of highly connected individuals. As she states:
"I argue that stock price correlations actually reflect coordination of traders activity that might be initiated by different factors related to news or noise. At the end what determines price changes are trading orders. Therefore, I propose that price changes actually measure the interaction strength among traders especially in extreme events situation."
Infection and Connectedness

So, if the key market nodes get infected then all we need to assume is that other people will copy their behaviour. From this it’s easy to deduce the spread of waves of buying and selling or, at extremes, manias and panics: the contagion that can spread through networks of this type is greatly exaggerated by these critical points getting infected. The same is true of the internet where computer viruses can survive long after they should have been destroyed – the ability of any critical node to re-infect swathes of the internet is the problem. It’s like one of our promiscuous lovers refusing to use any protection despite the clear evidence that they’re a danger to their partners.

So it is in markets, and modern technology makes it worse. The days when the average non-professional trader, working from home, was disconnected from the critical nodes has long gone. The critical nodes are out there on the internet, like giant spiders, spinning their webs, entrapping their clients and claiming their victims.

Immunise Yourself or Perish

Of course, there’s no getting away from the promiscuous purveyors of popular trading ideas, and the ever present global media, which likes nothing more than to feed off the extremes of human reaction: there's no exciting copy in news that the stockmarket isn't crashing or booming. So each of us has to take action ourselves, to seek out and immunise ourselves.

Worse, we need to desensitise ourselves to the opinions of those we work with or discuss ideas with. Danger lurks in listening to the authoritative ideas of anyone, but especially in social situations where herd like behaviour can develop: internet bulletin boards are a particular danger, although even simple investment clubs may be risky.

Developing a sceptical frame of mind when it comes to do with anything about investing is probably a good idea. As the old adage has it, you’ve got to buy when the blood is running in the street, even when it’s your own. You can’t bet against the market when all you’re doing is listening to what the market’s telling you: as anyone who sold stocks in early 2009 can tell you.


Related Articles: Newton's Financial Crisis, The Limits of Quantification, Capitalism Evolving, Be a Cockroach Not a Dinosaur, Bacteria, Boids and Market Instability

6 comments:

  1. This is the best written money blog I know of. Quality is high here and none of the articles I have read just parrot conventional wisdom (the most common failing in this field).

    We reached valuation levels never seen before in the late 1990s (which means that we are likely going to see the biggest economic crisis in U.S. history before this is over). I have wondered to what extent the problem is the internet communications medium. The wonder of the internet is that it so greatly facilitates the spread of information. But what happens when the "information" spread is mostly rationalizations for investing in insanely overpriced stocks? In that circumstance a medium that facilitates "information" sharing is a National Suicide Pill.

    Rob

    ReplyDelete
  2. Hi Rob

    Thank you, you're very kind.

    On network effects I'm not sure, really. We had plenty of bubbles and crashes before the internet. Is it really the case that 1999 saw overvaluation greater than 1929 (I don't know, I'm just curious)?

    However one might expect greater volatility if virus like information is more easily distributed. That's something I'll look into when I get some time.

    ReplyDelete
  3. Is it really the case that 1999 saw overvaluation greater than 1929 (I don't know, I'm just curious)?

    The valuation metric I use is P/E10. This is the price of a broad index over the average of the last 10 years of earnings. P/E10 was recommended by Benjamin Graham and is now used by Robert Shiller and many others.

    The fair-value P/E10 number is 14 or 15. Things get a bit dangerous above 20. Insanely dangerous markets are those that go above 25. The average loss on the four occasions when we went above 25 is 68 percent.

    The highest we ever went before the invention of the internet was the 33 we reached in the month before the 1929 crash (that led to the Great Depression).

    In January 2000, we went to 44.

    We were above 25 for the entire time-period from 1996 through 2008 (except for about two months when we dropped to 23).

    We are today sailing in totally uncharted waters.

    Rob

    ReplyDelete
  4. I'm trying to figure out where the promiscuous nodes might be during a housing bubble?

    Newspapers? TV? Real Estate Agents? I don't think it's blogs, or anything on the Internet, because these trades do not happen on the Internet.

    ReplyDelete
  5. Hi Alan

    I don't think it's necessarily that the trades are internet mediated that's the problem - although instant access to share dealing facilities is likely to be a contributing factor. For housing more likely it's standard social networks probably with a bit of a boost from the media.

    Property booms tend to be a cultural thing as well, though: countries with an Anglo background are more focused on property than many others. I suspect that this is a historical issue with its roots going back to the Magna Carta: a man's home is his castle and all that.

    Which, as it happens, is the subject of a future post I'm currently researching ...

    ReplyDelete
  6. Simply brilliant!

    I discovered your blog a couple of weeks ago. I picked some past posts and with almost every one of them my excitement grew until I came across this post. Now I am ecstatic!

    Great work! I simply love it.

    ReplyDelete