May 6, 2010 started out as an unusually turbulent day for Wall Street. Debt crisis in Europe triggered a decline in the Euro against the Dollar and Yen, and an uneasy feeling amid investors caused volatility in the Dow Jones Industrial Average. Against this backdrop loomed a massive sale: $4.1 billion in futures contracts, which would be pushed into the market by a super-powered computer in a span of 20 minutes.
At 2:32 P.M., the super computer at financial firm Waddell & Reed initiated the sale. Within moments, hundreds of super computers, programmed with lines of code called algorithms, picked up on the sale and began to react. They bought and sold the futures contracts at high speeds, passing them back and forth in what was called a “hot potato.” At the peak of the swarm, over 27,000 exchanges had taken place–in 14 seconds. The Dow Jones spiraled downward, shed 573 points, and then quickly rebounded. It left some small companies and individual investors at bottom and everyone involved in the markets scratching their heads.
The event came to be known as the “flash crash,” a phenomenon that Jeff Augen, an investor who builds algorithms and recently authored a book called Trading Realities, said left the market plagued with uncertainty and unfairness.
“No one really knows what happened in the flash crash,” Augen said. “Studying it is like examining a nuclear explosion and trying to study the path taken by every single particle in the explosion.”
What is more unsettling, Augen explained, is that the mechanisms set in place by the government following the crash could worsen the problem if it happens again.
The government’s reaction was, first, to unwind several trades that occurred as a result of the crash; they returned money to those who lost it. Afterward, the Securities and Exchange Commission set up circuit breakers–mechanisms for freezing the market in the event of extreme price fluctuations.
“I would be very afraid right now to own a portfolio of stocks,” Augen said. “The government doesn’t really know how this happened. There’s nothing to stop it from happening again because it was driven by algorithmic trading. And now, they’ve demonstrated that when an event like this occurs, they are willing to step in and selectively unwind some but not all trades.”
The idea of the SEC’s intervention in the event of a future crash spooks investors who hedge their investments with short positions designed to profit from a sharp decline. The concern arises because the SEC has demonstrated that they are willing to step in and selectively reverse trades that were profitable. In many cases their effect is to erase an investor’s hedge, leaving only the losing trades.
Furthermore, the circuit breakers, which halt the market if a stock’s price fluctuates by more than 10 percent in five minutes, could kick in just as the market is finding its bottom and rebounding.
“Usually it’s the short sellers that stabilize the market,” Augen said. “If you freeze the market, that disrupts all the activity that would cause it to rebound.”
The cause for all of this uncertainty is the recent influx of algorithms in Wall Street. The super-fast machines, which analyse data streams via code, arrived on the scene in the 1980′s, but they didn’t come to dominate the market until the last five years. Today, their influence is so great that, according to some estimates, 80 percent of the activity on the market is conducted by algorithms.
According to Khaldoun Khashanah, professor and program director of financial engineering at Stevens Institute of Technology, the machines analyse vast data streams, spanning from natural disasters, to pipeline explosions, to Twitter streams, and they filter the information to make one decision: buy, hold, or sell. The analyzation process used by the machines is far from that used by humans. The algorithms calculate expansive data sets, make rapid trades, analyze earning statements and even newsfeeds. But they lack common sense. They make dangerous automatic decisions that escape their form of rationality–like swarming on indicators to sell and causing the market to plummet, as evidenced in the flash crash.
“Like the May 6 event,” Khashanah said. “The effects of an isolated trade can easily lead to a cascading effect.”
Despite these dangers, super computers that trade using algorithms are magnets to those who can afford them. The amount of capital they can trade and generate on a daily basis is beyond anything Wall Street has experienced. And, according to Khashanah, the risk will become too great only if events like the Flash Crash start cropping up repeatedly.
“It’s a matter of emergence of patterns,” Khashanah said. “We need to look at the probability of a systemic risk in algorithmic trading, and we have to be careful to balance those risks against the open market and its ability to achieve its desired objectives.”
But Augen argues that the increased volatility in the market over recent years is testament to the algorithms’ effects. For example: between 1990 and 2008 there were 16 days with intraday high-to-low swings larger than 5 percent. By contrast, between 2008 and 2010, high-to-low events happened on 48 days.
“This is the era of algorithmic trading,” Augen said. “It’s the type of trading that shakes the individual investor out of the markets.”
The individual investor is further disadvantaged because of his or her inability to afford the expensive trading machines, Augen said.
“It just puts money in the pockets of the people who own $100-million-dollar super computers. But it doesn’t help anyone else,” he said.
That is because the machines allow institutional investment firms like JP Morgan Chase or Goldman Sachs to react and trade immediately following the change of trend. Regular investors do not have access to these machines, and are therefore left to contend with the markets only after they have been manipulated.
For Augen, the days of successful individual investors in the market are of the past. “A company’s business performance and its stock are connected by a 100-mile rubber band.”