Quantcast
Motherboard Blog

When Wall Street Robots Go Rogue, Banks Lose $10 Million a Minute

Wall Street robots went rogue on Wednesday unleashing a torrent of trades that wreaked havoc on the markets and cost one firm nearly half a billion dollars, or about $10 million a minute. It was a “mini flash crash,” experts called it, bringing back...

ALEC LIU

ALEC LIU

Wall Street robots went rogue on Wednesday unleashing a torrent of trades that wreaked havoc on the markets and cost one firm nearly half a billion dollars, or about $10 million a minute. It was a "mini flash crash," experts called it, bringing back dark memories of 2010, when a similar type computer glitch brought the Dow Jones Industrial Average tumbling 1000 points — or nine percent — in a matter of minutes.

This crash wasn't as severe, but for Knight Capital Group, the firm whose trading software went haywire, it could be a fatal blow. In about 45 minutes, the firm lost $440 million, nearly four times the revenue it brought in the first quarter of this year. "With the events of yesterday, you have to question if this is the beginning of the end for Knight," said Christopher Nagy, founder of the consulting firm KOR Trading. Knight Capital employees nearly 1500 people in 21 locations around the world.

Knight may very well recover, but the events are a bitter reminder of the fragility of the stock market, long taken over by Wall Street's robot overlords. In banker parlance, it's known as "High Frequency Trading" or HFT, where banks trade gigantic volumes using sophisticated computer algorithms so that they can jump in and out of the market at lightning speed, taking advantage of microscopic changes in order to make a quick buck.

Wall Street first started to use computers to assist with trading in the 80s — even then, computerized trading was blamed for amplifying the effects of the Crash of 1987 — but in the last ten years, the practice has accelerated exponentially as firms embarked in a cosmic race for size and speed, in a world where nanoseconds can determine the difference between winner and loser. These changes have forever warped the nature of the market.

When people think of the stock exchange, many still picture the floor of the NYSE, a scrum of maniacal brokers busting ass to get their trades in a la Trading Spaces:

Exchange floor specialists are ancient history.

In reality, the floor is much quieter these days, if nothing more than a historical artifact. Human traders have long ago been replaced by emotionless artificial intelligence. The stock market as you know it is dead. In its place are a bunch of deadly efficient supercomputers powered by black-box algorithms trading amongst each other.

No mob necessary. Just one guy to make sure things are run smoothly (or not).

At the start of the millennium HFT comprised just ten percent of the equities market. By 2010, more than half of all US trades daily by volume were being done by HFT operations, according to a report from the TABB Group, a financial services industry research firm. It's easy to see why. By utilizing their proximity to the market, banks like industry leader Goldman Sachs made more than $100 million in trading revenue on a record 46 days during the second quarter of 2009, aided in part by their use of HFT bots, according to Bloomberg. TABB estimates that the 300 securities firms and hedge funds that specialize in this type of trading took in roughly US$21 billion in profits in 2008. And that was four years ago.

HFT firms make money in a myriad of ways. The most conventional is market-making, by taking orders for securities and cashing in on the bid-ask spread, the difference between the cost of buying and selling. This was the strategy employed by Knight Capital Group.

Other strategies however, are a bit more nefarious, such as the vaguely termed "event arbitrage." With these, finance "quants," or the industry math whizes, create algorithms that can predict the occurrence of certain events, such as when a pension fund might buy or sell assets. Using their technical speed advantage, these firms can jump in first, before and after the trade, and cash in on the difference. With the kinds of volume involved, even tiny differences can equate to big profits.

As Cathy O'Neil, a math-professor-turned-hedge-fund-quant-turned-Occupy-the-SEC activist, noted in episode 4 of Frontline, her job as a quant at hedge fund D.E. Shaw was to predict when pension funds would buy or sell assets so her traders could "frontrun" the trade, essentially skimming off the top of some elderly couple's retirement money.

"I just felt like I was doing something immoral," Cathy O'Neil, a math-professor-turned-hedge-fund-quant-turned-Occupy-the-SEC activist, told PBS's Frontline earlier this year during their special on Wall Street. "I was taking advantage of people I don't even know whose retirements were in these funds. We all put money into our 401ks and Wall Street just takes this money and skims off a certain percentage every quarter. At the every end of someones career, they get some of that back. This is this person's money and it's basically just going to Wall Street. this doesn't seem right"

But like all profitable financial innovations, those with any skin in the game have been quick to laud their merit. Technology has helped create a more efficient market, they say. "Generally speaking, the evolution of high frequency trading has helped limit spreads and increase market liquidity — efficiencies that benefit all market participants," the Chicago Board Options Exchange calmly states on its website.

They're not entirely off base. The advent of computerised trading, especially when it comes to market-making activity, has reduced spreads and thus trading costs while also providing liquidity, academic papers have shown. But as this last mini-crash, the recent Facebook IPO debacle that cost UBS hundreds of millions — also due to a computer bug — and the infamous crash of 2010 indicate that while we might be getting bigger, faster and cheaper, we're also infinitely more prone to sudden catastrophe. Since these are recent developments, it's impossible to fully grasp the implications of what this might mean (although one research firm is now reporting that HFT may cost average investors $2.5 billion a year).

And like many trendy financial innovations, they can quickly mutate into unruly untameable, uncontainable beasts. "Technological advances have outstripped our ability to regulate them," says Andrew Lo, director of Massachusetts Institute of Technology's Laboratory for Financial Engineering and chief scientific officer of quantitative-analysis hedge fund AlphaSimplex Group LLC.

"It's like the Wild Wild West."

Follow Alec on Twitter: @sfnuop

Connections: