Too Fast for the Market

Morgan O'Rourke


June 6, 2013

On Tuesday, April 23, just after 1 p.m., the Associated Press posted a shocking tweet that read, “Breaking: Two Explosions in the White House and Barack Obama is injured.” Given that the news came from such a respected, mainstream source on the heels of the Boston Marathon bombing, it was hard to blame any of the AP’s 1.9 million Twitter followers for fearing the worst. Almost immediately, the Dow Jones Industrial Average plummeted more than 140 points, temporarily wiping out some $136 billion in market value.

But, of course, the tweet was a fake.

Minutes after the phony news went out, the AP’s corporate communications team rushed to assure the public that the @AP Twitter account had been hacked and that no explosions or injuries had occurred. White House spokesman Jay Carney also confirmed that the president was fine, and the Dow rebounded to its pre-tweet levels. The entire incident lasted only about 10 minutes.

Cybersecurity implications aside, the larger issue is how quickly the news moved stock prices. In recent years, financial markets have started to see more and more of these fleeting and unpredictable market collapses, known as “flash crashes,” that come and go within minutes, or even seconds. The original flash crash happened May 6, 2010, when the Dow dropped more than 600 points in five minutes, erasing $850 billion in market equity, before gaining it all back 20 minutes later.

But smaller crashes happen all the time. A day before the AP Twitter hack, for example, Google experienced a flash crash that saw its shares drop 3% in less than a second before recovering, and days later, software maker Symantec watched its market value plummet by $1.7 billion when a three-second crash caused stock prices to drop 11% before it too rebounded. According to a CNN report in March, most stock traders claim that dozens of these mini-flash crashes happen every day.

On the surface, these kinds of crashes seem relatively harmless. No matter how large the loss, the market usually gains back its value rapidly. But investors can be on the hook for devastating losses. During the May 2010 crash, one investor reportedly lost $17,000 because of bad timing when he decided to sell his Procter & Gamble shares just as the company’s stock price was dropping more than 30%. Other investors have fallen victim to stop-loss orders, which are ironically set up to automatically sell stocks when their prices fall to a certain level in order to protect investors from major losses. But in a flash crash, the stock rebounds so quickly that investors who sold low are left holding the bag, unable to buy back the stock without incurring a hefty loss.

At issue are these automatic, computerized transactions and what is known as high-frequency trading. High-frequency trading uses computers to make precise calculations designed to carry out thousands of trades per second in order to take advantage of even the smallest fluctuations in the market. The algorithms that guide these trades are programmed to take all sorts of market conditions into account, including news reports and tweets, which helps explain why the markets crashed after the AP hack. Words like “explosions” and “White House” can immediately influence stock sales until other algorithms kick in to automatically prompt buying again. It is not hard to imagine how a hacker could cause chaos in the market, and possibly make millions doing it with some well-timed stock trades, just by taking advantage of algorithms that can’t tell the difference between fake news and the real thing.

To help mitigate the problem, the SEC and the major stock exchanges have developed “circuit breakers” that pause or stop trading in the event of large, quick-moving stock-price fluctuations, giving the market a chance to restabilize. But these measures can only reduce the damage, not stop it from occurring in the first place. Flash crashes can still happen fast enough to cause serious damage, and the risk will remain as long as software and algorithms have a role in high-speed stock trading.

The knee-jerk reaction would be to ban all automatic, high-speed transactions, but for better or worse, technology is a Pandora’s box that doesn’t close once opened. However, in order to preserve the integrity of the market, regulators may have to enact even stricter measures to limit these transactions and reduce the impact of these arbitrary swings. After all, real people, not algorithms,  are at risk here, and as any driver knows, there’s no sense in speeding if you can’t stay in control.

Morgan O’Rourke is editor in chief of Risk Management and director of publications for the Risk & Insurance Management Society, Inc. (RIMS)