A black swan is a rare and beautiful sight. Here, where I live, I can go to the park and admire these elegant creatures. I could go daily, but I won’t. Most likely most of us haven’t seen a black swan ever. The world of the stock exchange is oft perceived as a fast and exotic place. They are absolutely right! Because between 2006 and 2011, a black swan was sighted there on average more than 10 times a day.
Research by Neil Johnson at the University of Miami shows that in a 5 year period black swan events, or big wild swings in an ultra-short period of time, took place a whopping 18,520 times. These sightings passed by so fast that in the time for a human to become aware of this rare creature it had also disappeared from sight, only leaving a (digital) paper trail. These ultrafast hiccups are actually algorithms besting it out in less than a second.
Compare that to a brawl in a bar. Your annoyingly drunk friend gets in a fight, thrown to the ground, and helped back up and, in good measure, ordered a beer, all in under a second. Are you a bad friend by not noticing? Are the folks of Banking Supervision bad friends of ours, because they are not stepping in fast enough to stop our algorithmic friends beating up an innocent stock?
For these kinds of questions, we need a dead French philosopher to help us out. Baudrillard is just the guy to channel in. We believe that the stock exchange in the long term will show us how companies are really doing, right? Right now Baudrillard is laughing his French ghost laugh right in your face… Because, according to him, reality can no longer be produced. We try to map it, but our doodles are not resembling reality, nor the truth. It looks worse than the self-portrait of your 4-year-old nephew.
So what on earth has this French philosopher to do with our question about ninja fast algorithms behaving badly in Wall Street? Hold on to your brain, because now the tough cookie part begins. First, a Baudrillardian question: If algorithms are not producing the real then what are they producing?
Second, we can talk more about your friend who gets really annoying after a certain amount of alcohol and the correct input. Because, yes, that is exactly how an algorithm works. Really, only here alcohol equals time. The more alcohol our mutual friend consumes the more annoyingly unpredictable he gets. The faster an algorithm works the more annoyingly unpredictable it gets.
Just listen to the frantic screams from Wall Street after a bunch of our algorithmic friends roamed through Wall Street, casually causing the flash crash in 2010, also known as the four minutes of apocalyptic financial doom.
And let us be honest, our mutual friend is not a daily bad drunk, if he was then we would unfriend him in an instance but met with the right circumstances, he occasionally just goes bonkers. Our algorithmic friends however, just happen to be alcoholics who go violent at least 10 times a day. And I almost forgot, they don’t smash up barstools, but violently distort a stock price. Of natural gas, for example. And that, ladies and gents, can have a nasty effect on your pension fund, your bank, our economy. You know, the stuff that is harder to replace than a barstool…
Sobered up, we return to our first question. What are the algorithms producing if not the real, or the actual state and value of companies? That question I will leave you with for now. But ask yourself; would you accept this behaviour from a friend? Or would you prefer friends that do not get into ninja fast brawls 10 times a day, every day, without even being able to notice the start, let alone being able to step in and try to soothe everyone, before that one big brawl starts that destroys the entire bar. O, and where the bar is actually our financial world…
Time for an intervention?