Risky 'Robotic' Market Algorithms?
Karel Čapek, Playwright
I read an alarming column by Mark Buchanan that riffs off Ray Kurzweil's concept of the "singularity" -- the point at which technology surpasses us in intelligence -- which might be sooner than we generally believe. The column, titled "The Reign of Robots May Be Closer Than You Think" (Bloomberg, March 7, 2012), warns of the danger lurking in our increasing tendency to automate processes that had formerly been subject to human judgment. Why a danger? Buchanan points to evidence of a shift toward automaticity in market trading and explains its risky implication:
[F]inance is flirting with a similar transition [through a singularity of its own], as ever-faster computing and communications technology takes high-frequency trading into a regime of speed where human beings can no longer keep up.Indeed, Buchanan thinks that "we may have already arrived," and offers some disturbing evidence:
In a recent study, physicist Neil Johnson and colleagues found more than 18,000 instances over the past five years where markets, in about a second and a half or less, either ticked up or down at least 10 times in a row, making prices rise or fall in that span by more than 0.8 percent. Many of these mini-crashes and mini-booms took place in well under a tenth of a second, effectively instantaneous from a human perspective . . . . [and] have been happening roughly 10 times per day . . . . The same study looked at the incidents on different timescales, both above one second and below, and found a striking difference. Over periods of one second or longer, the distribution of events by size has the familiar "fat tailed" distribution -- the norm for markets, . . . which reflects their pronounced susceptibility to large price changes . . . . [F]or events that last less than one second, . . . the distribution is "fatter than fat" and shows an even greater than normal tendency for Black Swan-type upheavals . . . . [But what's] so special about one second? . . . . [T]he researchers point out . . . that one second . . . [is] around the speed limit for fast human decision making . . . . [A]re the markets at this timescale showing the signs of an emerging all-machine phase of trading over which human decisions have little influence or control?Buchanan would answer yes, markets are showing signs of machine control. Why is this happening? We first have to understand a distinction between uncrowded and crowded markets:
Mathematical studies . . . show that one of the most fundamental factors influencing . . . basic [market] dynamics is how "crowded" the market is . . . . If the participants in a market use a wide and diverse range of trading strategies, then the market is uncrowded . . . . A healthy diversity of participants earns profits in different ways -- thinking and acting on different timescales, taking different views on the future and so on . . . . Real markets . . . look a lot like this uncrowded phase, with highly irregular market fluctuations and fat tails . . . . [I]f a market becomes overcrowded -- that is, if many traders chase few opportunities and use very similar strategies to do so -- then the continuity of the market tends to break down. In this regime, the market becomes prone to . . . sudden moves up or down much like those now observed in the sub-one-second trading regime.So, the market is crowded, according to Buchanan. What, specifically, is driving this shift toward a crowded, rapid market?
[The use of] high frequency algorithms [that] . . . compete on speed and have to act extremely quickly . . . . [Such algorithms] must be relatively simple, and can't waste time analyzing too much information about the past. Given these constraints on the range of possible strategies, and given the number of traders operating within them, overcrowding is quite likely -- as are the . . . troubled market dynamics arising from it.Not only is the market crowded, it's overcrowded! Now comes the scary part:
[W]e . . . should actually expect to see increasingly frequent Black Swan events in microscopic timescales. They may well be the natural consequence of machine trading . . . uncoupled from the strong influence of conscious human decision making . . . . [A]s Johnson and colleagues put it, [we're moving] "from a mixed phase of humans and machines, in which humans have time to assess information and act, to an ultrafast all-machine phase in which machines dictate price changes."Concerning such a shift to machine control, Buchanan concludes:
We're crossing a boundary into a trading twilight zone, and doing so without much thought or awareness of the potential dangersI think of the differences between the physical laws that describe the behavior of macroscopic and microscopic particles -- except that the quantum realm doesn't destabilize the macroscopic realm, whereas such destabilization seems the very real danger in markets as the algorithms begin to drive not only the sub-second market processes, but by implication the supra-second ones as well!
Do these automatic processes offer an intimation of what is to come? Will we enter a truly post-human world in which robots -- as in Karel Čapek's play, Rossum's Universal Robots -- vanquish humanity?