By Emily Flitter and Sarah N. Lynch
NEW YORK/WASHINGTON Feb 26 (Reuters) - The U.S. Securities and Exchange Commission is beefing up its capacity to detect insider trading and other illegal activity by engaging with Palantir Technologies, a software company originally set up to help foil terrorists.
The SEC has embarked on a multi-year deal, worth more than $13 million, to use Palantir's technology to help the agency crunch massive amounts of data, said John Nester, a spokesman for the U.S. securities regulator.
Its partnership with Palantir is one of multiple "big-data" projects the agency has initiated as it tries to narrow the technology gap between regulators and Wall Street.
The SEC, with its $1.35 billion budget, has always grappled with trying to keep pace with the industry, which is able to outspend the regulatory agency on technology and pay higher salaries to its experts.
Closely held Palantir was co-founded by Peter Thiel, the billionaire tech mogul who also co-founded PayPal, the online payment service now owned by eBay Inc.
The Palo Alto, California-based company's software has its roots in technology that PayPal uses to detect fraudulent transactions, and it takes its name from a magical stone in J.R.R. Tolkien's epic "The Lord of the Rings" that can see things not immediately apparent.
Officials expect Palantir's platform to help the SEC find evidence of illegal activity more quickly and easily by linking trading records and personal contact information from paid databases with tips, complaints and referrals the agency has received.
The Palantir platform, while up and running for several months, has yet to be rolled out to full capacity throughout the SEC.
Another big-data project designed to help SEC examiners detect illegal trading activity -- the National Exam Analytics Tool, or NEAT plus -- is scheduled to go live on March 1.
The NEAT plus tool uses a free database created by a U.S. Navy contractor for missile defense systems and gives SEC examiners the power to quickly analyze massive amounts of trading data for brokerages and other Wall Street firms.
"Plus is the secret sauce - how we convert the data to information," said Erozan Kurtas, an SEC quantitative analytics expert who gave a presentation about the tool on Saturday at the Practising Law Institute's annual "SEC Speaks" conference.
"All the hedge funds always talk about their secret sauce. We finally have one, too."
The SEC declined to discuss more details of the project beyond what has been said publicly, saying agency officials still need to work through the details to make sure both the systems and exams are not compromised.
THE POWER OF PALANTIR SOFTWARE
Members of the SEC's Division of Enforcement are using Palantir's technology to look for incidents of insider trading, pump-and-dump schemes in penny stocks, accounting fraud and violations of the U.S. Foreign Corrupt Practices Act, among other activities.
The agency is already using analytical models and algorithms developed in-house and run on software provided by SAS to find suspicious activity in stock trading and other activities it regulates. Palantir's contribution, according to the sources, is to strengthen the connections between different sets of data.
While Palantir is a platform linking data sets, NEAT is an analytical tool that can be applied to sift through the data for patterns and other important observations.
With the NEAT project, Kurtas declined to say which contractor created the database being used.
But he noted that because the database is free, the SEC was able to essentially build NEAT in-house by utilizing existing resources and avoiding spending millions of dollars for new software.
The SEC is one of a host of government agencies turning to "big data" to improve their detection of suspicious activity. Palantir works with other U.S. government agencies, including the National Counterterrorism Center.
A spokeswoman for Palantir declined to comment.
The new platform is designed to let people working in different divisions of the SEC more easily share discoveries about companies and people that could help lead to more cases.
Sofia Hussain, a senior forensic accountant in the SEC's Enforcement Division in Boston, said pattern-recognition tools already have led to several new cases.
Palantir allows the SEC to compare the records of every trade made in a particular stock, along with the identities of those who made each trade, with addresses, phone numbers and other personal information from paid databases. That enables it to see more clearly relationships between all those trading in the same stock.
The trend toward big data in white-collar enforcement has reached the private sector, too. A new software vendor is selling technology to hedge funds, pension funds and brokerage firms that profiles traders, keeps track of their normal trading habits and reports any suspicious deviations.
The company, AIMPaaS, began marketing its software three weeks ago and is currently negotiating its first deal. Its co-founder, Leslie Seff, said the company is talking to large hedge funds. He said AIMPaaS software can be customized to change the threshold for suspicious activity, which is determined by changes in order size, trading venues and other features.
EXAMINERS EYEING INSIDER-TRADING
Like the Palantir platform, the National Exam Analytics Tool will assist the SEC with tracking potential illegal activities, such as insider trading, front-running or violations of "Regulation M, which bans traders from short-selling ahead of an offering and then buying the same stock in the offering.
SEC Chair Mary Jo White, in a recent speech, said the insider-trading analysis is done by comparing one database of information on merger deals and other corporate activities against trading activity data.
Andrew Bowden, the head of the SEC's Office of Compliance, Inspections and Examinations, on Saturday called the tool "revolutionary," saying it gave the SEC an edge over Wall Street. "They will have to raise the bar in order to keep up with us," he said.
Kurtas said on Saturday that he has several other projects in the works.
One prototype would help the SEC detect possible money-laundering activity, and another is designed to flag riskier firms so examiners can prioritize their inspections.
Kurtas said his team was also building a high-frequency trading analytics laboratory. It will collect and sort through voluminous computer files of transactional trading data, including order cancellations and other types of message traffic.