WASHINGTON/NEW YORK (Reuters) - U.S. securities regulators have taken the unprecedented step of asking high-frequency trading firms to hand over the details of their trading strategies, and in some cases, their secret computer codes.
The requests for proprietary code and algorithm parameters by the Financial Industry Regulatory Authority (FINRA), a Wall Street brokerage regulator, are part of investigations into suspicious market activity, said Tom Gira, executive vice president of FINRA’s market regulation unit.
“It’s not a fishing expedition or educational exercise. It’s because there’s something that’s troubling us in the marketplace,” he said in an interview.
The Securities and Exchange Commission, meanwhile, has also begun making requests for proprietary algorithmic trading data as part of its authority to examine financial firms for compliance with U.S. regulations, according to agency officials and outside lawyers.
The requests by SEC examiners are not necessarily related to any suspicions of specific wrong-doing, although the decision to ask for it can be triggered by a tip, complaint or referral.
According to interviews with attorneys, traders, industry executives and regulators, the unusual requests for algo code and other computerized trading strategies really ramped up this year and have targeted stock-trading firms such as broker dealers and hedge funds.
It has alarmed some traders who are afraid their “secret sauce” — intellectual property sometimes developed over years and at great cost — could get into the wrong hands, especially when SEC and FINRA examiners leave for the private sector.
“I’d be disappointed and upset” if they asked for code, said a high-frequency trading firm executive who declined to be named. “I mean, are these people all going to work at the SEC forever?”
The SEC’s new focus on algo strategies will likely help inform any new structural rules the government agency applies to an electronic market, criticized by some as unstable or unfair, especially after the “flash crash” on May 6, 2010.
While anything the regulators find could lead to legal action such as market manipulation suits, FINRA’s effort appears more targeted at wrong-doing.
FINRA, which reports to the SEC, usually focuses its requests on flawed codes in an effort to better understand how they are constructed, operate, and how they are supervised, Gira said. An unusually large wave of orders for a lightly traded stock, for example, could lead to a request, he said.
Trading code is a high-stakes secret for high-frequency firms that battle each other to earn razor-thin profits on tiny price imbalances in the market. Such firms can make thousands of trades per second and provide much liquidity to the market.
High-frequency trading is estimated to be involved in more than half of all U.S. stock trading. Regulators have said the algos behind such trading were a factor in the flash crash, but that they did not cause it.
Carlo di Florio, who heads the SEC’s Office of Compliance, Inspections and Examinations, said the agency started asking firms for proprietary algorithmic trading data over a year ago, and has since more broadly incorporated such requests into its risk-based exams.
Most of the algo-related requests, he said, have been made to hedge funds that use quantitative trading strategies.
Although some lawyers and industry sources have said the SEC has asked for the actual computer code itself, di Florio said such a request is “very rare.” Instead, most of the time the SEC has been asking for research papers containing sensitive information about trade reasoning and proprietary formulas.
“When we go in ... we are thinking about what is the most critical information that will give us the insights we need, and often times, that is not the code itself,” di Florio said in an interview. He said so-called white papers, which detail the purpose and strategy of a trading model, are often most helpful.
SEC examiners want the information to ensure that hedge funds are actually using the strategies they market to investors. They also review it to make sure that algos are not being used to manipulate the market.
An industry attorney said that FINRA and the SEC have also been asking firms specifically how their algos react to different market conditions, and what data feeds they use.
“They’ve certainly taken this to the next level,” the attorney said.
Last year, SEC Chairman Mary Schapiro said regulators were investigating whether traders manipulated prices, encouraged volatility, or committed fraud by flooding the market with rapid-fire orders that were almost immediately canceled.
FINRA, meanwhile, has made market manipulation a high priority since it fined a small firm called Trillium Brokerage Services $1 million last year for “baiting” other traders with a high volume of “illegitimate orders” in 2006 and 2007.
The requests by regulators for what is often considered intellectual property are making some firms nervous.
Since Schapiro took the helm of the SEC in early 2009, she has pushed to revamp its enforcement and examination programs. Part of that effort is hiring outside industry experts — some of whom will likely one day return to the private sector to work for these firms’ competitors.
Some high-profile industry hires include Rick Bookstaber, a former hedge fund manager who works in the SEC’s Risk, Strategy and Financial Innovation Division, the SEC’s “think tank” unit that often assists with exams and inspections.
Other agency hires have included Erozan Kurtas, a former Standard & Poor’s staffer with multiple degrees who has algo design experience, and Tim Techathuvanan, who previously worked at a hedge fund coding quantitative trading models.
Although the agency currently only has about a half-dozen algo experts on staff, di Florio said he hopes to hire more people with these kinds of backgrounds — especially as the number of firms using such automated trading strategies grows.
Gira said FINRA also recently beefed up its staff to add people who better understand codes and their market impact.
Underscoring the sensitivity of the matter, a former Goldman Sachs Group Inc programer, Sergey Aleynikov, was sentenced in March to eight years in prison for stealing code from the bank as he left for a job at start-up trading firm Teza Technologies.
Both the SEC and FINRA said they understand that firms are concerned about confidentiality of intellectual property, noting they have policies barring employees from using sensitive information like codes to their advantage. Criminal laws also serve to deter people from stealing such data, they said.
Still, one industry attorney said his client has lingering concerns that a staffer might remember something he saw in an exam and use it down the road.
Even though the SEC believes it needs this algorithm information to help it police the market, many on Wall Street are still not convinced the agency will know what to do with the data.
“Let’s just say the good developers in the industry are being hired by the industry — not by an SEC salary,” a trader said.
Reporting by Sarah N. Lynch and Jonathan Spicer; Editing by Tim Dobbyn