Technology

SEC Chief: Tech Glitches Triggering Wild Market Swings

Basic errors in software testing and deployment, maybe as much as newfangled algorithms, have caused the trading malfunctions in the stock markets,...
Vincent RyanOctober 5, 2012

Another technical malfunction on the Nasdaq exchange that happened this week may get regulators to pay more attention to core technology and infrastructure issues in the stock markets. In particular, the failure to bulletproof new trading platforms and the lack of robust risk-management systems seem to be behind some of the problems. 

On Wednesday Kraft Foods was completing its switch from the New York Stock Exchange to Nasdaq as part of a split with its snacks business (the newly named Mondelez International). But in the first minute of the trading of Kraft shares on Nasdaq, something went awry. Kraft shares skyrocketed 29%, to $58.54. The stock quickly settled down to the mid-$40 range, but the large price move, absent news about the company, led Nasdaq and other electronic-trading brokers to cancel some of the trades that occurred in those first few minutes.

Nasdaq OMX has only said the event was caused by an error by a broker. But speculation is that algorithms from high-frequency traders also contributed.

The Kraft shares mishap is only the latest hiccup to cause wild swings in a company’s share price. But while the occurrences have largely been attributed to the complexity and fragmentation of the stock markets and the lightning speed with which trading orders travel across exchanges ­some of the troubles, in part, may be simpler, says Securities and Exchange Commission chair Mary Schapiro.

At an SEC roundtable a day before the Kraft Foods glitch, Schapiro said there are technology infrastructure problems on exchanges and with electronic-trading platforms, and that fact may be getting lost amid all the arguments over high-frequency trading, market liquidity obligations, and market manipulation. “Matters of infrastructure are essential to any holistic approach to improving how our markets operate,” she said.

From Schapiro’s point of view, glitches like the failed initial public offering of BATS Global Markets, the extended delay when Facebook shares opened trading, and the order problems that forced Knight Capital to scramble for funding were, in part, “basic technology 101 issues.”

“We have a very competitive market environment in which rapid innovation and speed-to-market may compete with diligent testing and validation of the technologies that support such innovation,” Schapiro said.

The BATS, Facebook IPO, and Knight Capital incidents all involved flaws in software programs or electronic-trading systems; in BATS’s case, a single line of code from newly installed software was the cause, said the firm’s chief operating officer at the SEC roundtable. The Knight Capital incident involved a trading program that had just been installed.

Those technology glitches get magnified by the current market structure. Multiple, interlinked trading venues mean infrastructure failures in one market can “cascade” into other venues, Schapiro pointed out, and trading’s inherent speed “means that even small, short-lived infrastructure issues can cause drastic harm.”

And newly installed trading platform systems are not the only areas of tech vulnerability. Broker-dealers and high-frequency traders are spawning algorithms that contributed to events like the May 6, 2010, “flash crash” and other smaller, less-publicized instances when the markets went haywire, experts say.

Algorithmic traders use financial models and computers that analyze mountains of market data to make transaction decisions. A study released last month by the Federal Reserve Bank of Chicago found that  some broker-dealers and proprietary trading firms don’t have stringent processes for developing, testing, and deploying the code in their trading algorithms. They also, according to the study, “deploy new trading strategies quickly by tweaking old code and placing it into production in a matter of minutes.”

While the best practice is for a firm to perform pretrade “risk checks” to prevent a misbehaving algorithm, some “are relying solely on risk controls set by the exchange,” the Chicago Fed wrote in its report. But the exchanges’ risk systems don’t always catch the erroneous orders.

Firms must have robust processes for developing and testing new software. But the industry also needs “advanced risk-management systems to limit the risk of unintended trading activity by a firm or its client,” said Chris Concannon, executive  vice president of Virtu Financial, an electronic market maker, when testifying before the Senate last month.

While regulators have some preventive measures in place  like single-stock circuit breakers that are supposed to prevent a stock’s price from collapsing when trading glitches occur  the SEC roundtable discussed introducing “kill switches.” Kill switches could be flipped by exchanges quickly to block a broker’s trades when errors are discovered.

But better processes, software safety switches, and testing of new code also need to be supplemented by human monitoring, said TABB Group CEO Larry Tabb at the Senate committee hearing last month. If the equity markets want to insulate themselves from the potentially dangerous systemic effects of a rogue algorithm, “every trading machine needs a stop button,” he said. “An electronic-trading problem is only an electronic-trading problem for a minute  after that it is a human problem.”