The stock market crashes all the time without anyone even noticing. That’s because it happens so fast human eyes can’t see it.
That’s the conclusion of a paper from researchers at the University of Miami, the University of Vermont, the MITRE Corporation, and financial-tech firm Nanex. The study uncovered 18,520 stock market crashes or spikes between January 3, 2006, and February 3, 2011. That’s an average of nearly 10 a day.
Co-author and physics professor Neil Johnson at the University of Miami says the flash crashes are the result of an “ultrafast algorithmic arms race.” Some new information sets off a chain reaction, where computer programs each try to exploit each other’s reactions. “When the information flowing to them is of a certain type, many of them may adopt similar strategies and so they end up fighting for the same scrap of food (i.e. same profitable opportunity),” writes Johnson.
The resulting spikes go unnoticed because they’re very fast, happening in less than 1.5 seconds, and also because they’re relatively small. To qualify as an “ultrafast extreme event” according to the researchers’ methods, a stock price needed to fluctuate in that 1.5 second period by an amount exceeding 0.8% of the stock price.
But if the fluctuations are so easy to overlook, why should anyone other than high-speed traders even care? For one, patterns in the micro-events may help predict big ones. “These fleeting spikes and crashes start to escalate well before the financial crash in 2008, and this is suggestive of some kind of causal relationship to macro global market health,” says Johnson. “In particular, they are most prominent in the financial stock (e.g. JP Morgan and Lehman Brothers) leading up to the collapse, and this is the sector where the system-wide collapse is attributed.” Johnson thinks the evidence is good that the short-term events are “inextricably linked” with the long-term (and large-scale).
But he also thinks there may be insights that generalize outside of the stock market. “I think that the real-world example of the markets actually gives a remarkably strong insight into the shape of things to come in the future for cyberwarfare,” says Johnson. “The old image of cyberattacks being a human sitting at a terminal and hacking a password is one version, but much more likely in the future in my opinion is the unleashing of swarms of simple but ultrafast predatory algorithms that are in and out in the blink of an eye.”