Global bank regulators call for more risk controls around algorithmic trading

By John McCrank

NEW YORK (Reuters) - Banks need to put in place more stringent standards around algorithmic trading to help reduce risks to the financial system as trading becomes more electronic and increasingly complex, a group made up of global regulators said on Thursday.

Most trading firms are highly automated and use pre-programmed instructions, known as algorithmic trading strategies, to make lightning-fast decisions on which securities to buy and sell, with little human intervention. Banks regularly use algorithmic trading strategies and have high-frequency trading firms as clients.

The concern is that risk controls at financial institutions may not have kept pace with advancements in algorithmic trading, said the Senior Supervisors Group, which made up of members from bank supervisory authorities from several countries.

"An error at a relatively small algorithmic trading firm may cascade throughout the market, resulting in a sizable impact on the financial markets through direct errors or the reactions of other algorithms," the group, which includes the U.S. Federal Reserve and the European Central Bank, in a briefing note.

The 2010 "flash crash" that temporarily wiped out $1 trillion from U.S. stock markets was an event that regulators highlighted as a shortcoming in risk controls. Another example was the 2012 Facebook Inc initial public offering, when market makers lost around $500 million due to a glitch at Nasdaq OMX Group Inc's exchange. Regulators also noted the 2012 trading error that cost Knight Capital Group $461 million and forced the sale of the firm.

The group recommended that banks put in place transparent, firm-wide risk management practices that includes testing during all phases of an algorithms lifecycle, because as automated trading grows in number and complexity, so too may systemic risk.

Algorithmic trading began in the U.S. equity markets, where it now accounts for around 50 percent of the volume, but has become a common feature of other markets and asset classes, including futures, foreign exchange, and fixed-income markets.

As these markets become more interconnected due to algorithmic trading, the effects of errors or attacks could amplify risk in the financial system, the report said.

"The immediate need for stronger controls is paramount," the group said.

(Reporting by John McCrank; Editing by Lisa Shumaker)