Flickr / smartfat.
There are countless methods of predicting what will happen next in the stock market. Technical traders have their charts, quant funds use specialized algorithms, and psychics have crystal balls. Many of their methods have failed quite spectacularly in the past, especially in the 2008 financial crisis.
With that in mind, researchers have turned to other methods of understanding the vagaries of financial markets. One idea is to model the market's activity in the same way geophysicists model earthquakes. It seems to work, but what if it ends up working too well?
Seismic financial activity
The idea is that a financial market crash looks -- mathematically speaking -- kind of like an earthquake. Both tend to be self-perpetuating, meaning that they build on their own momentum, and are often followed by aftershocks. Indeed, the researchers found that between 84% and 88% of extreme market drops were caused by this "self-excitation."
Also common to both is that extreme negative events are more common than normal statistical models would suggest, and in both cases, it's these extremes that beget the biggest risks. In other words, it's not the 3.0 magnitude quake that's going to worry you, it is, as we Californians like to call it, The Big One.
Does a model like this make sense?
The benefit to models that are built to fit data, rather than an assumption, is that they can constantly update and presumably become more accurate over time.
The authors of this study found that their early warning system models were pretty good at predicting sharp drops in the S&P 500, and that the false positives tended to outweigh the false negatives (there were differences based on the model used -- if you crave math, you can download the paper here). In other words, this concept can still be thrown for a loop, but it seems pretty promising so far.
But maybe it could work too well
The major difference between earthquakes and financial markets is that one of them can be affected by the actions of humans while the other one cannot.
If you have an early warning system for earthquakes, all you can really do is get out of town or bunker down in an appropriately outfitted building. But what happens if the financial early warning system warns of a crash and a lot of people get the signal?
In this case, I'd argue that the warning could become a self-fulfilling prophecy, making a modest crash into a really, really big one. Think about it: One of the features of crashes, even according to this model, is that they are self-propagating. Could a system like this just make things worse?
The jury is out, but I'd argue that a reliance on highly accurate early warning systems could lead to even more swings in the market than we've seen in the past. If you're the smart money, maybe you pay attention to such warnings so you can laugh hysterically and buy while everyone else is packing their metaphorical bags.
But doesn't that mean everyone else get hosed? I'm not sure. Either way, it makes me think that predicting the future is not an enterprise for the faint-hearted.
