BATS Crash: Problems with High-Frequency Trading?

On March 23, the Better Alternative Trading System (BATS) exchange saw a crash due to technical issues on the same day its own company made its initial public offering. The crash precipitated inquiries into high-frequency trading and the automation of financial markets.

BATS Global Markets planned to offer over seven million shares, at $16 apiece, on its own exchange, a high-frequency trading venue. After the shares opened at $15.25 under the ticker “BATS,” within nine seconds they fell to four cents per share, at one point spiking below one penny per share, forcing a halt and the subsequent withdrawal of the company’s IPOs from its own listings.

“Although our affected market has reopened, in the wake of today’s technical issues, which affected the trading of certain stocks, including that of BATS, we believe withdrawing the IPO is the appropriate action to take for our Company and our shareholders,” wrote Joe Ratterman, BATS chairman, in a statement that afternoon.

It turns out that the crash at the third-biggest U.S. stock-exchange venue was due to a coding error that disrupted trading in firms with ticker symbols from A to BFZZZ. Along with BATS itself, Apple was also affected, with a slump of 9.4% within about five minutes before trading was suspended by circuit-breaker rules.

The BATS exchange’s largest investors include trading firm Getco and investment banks such as Citigroup, Morgan Stanley, Credit Suisse, and Bank of America. These banks, in addition to Deutsche Bank, Sandler O’Neill, Wedbush Securities, Raymond James, Rosenblatt Securities and Nomura were to be lead and co-managers of the exchange’s IPO. The underwriting agreement, however, was canceled on March 23 according to BATS.

BATS appeared to be quite undamaged by the crash as a venue for trading. By March 27 the exchange’s market share has recovered to 11%, about two-thirds of NASDAQ, and close to the New York Stock Exchange – the only two exchange venues surpassing BATS in volume. However, BATS’s credibility as a venue for IPOs was hurt. It is rare for a company to withdraw its IPO after the IPO has been priced, as BATS did.

More importantly, the crash renewed alarmed inquiries into the use of high-speed trading and the automation of financial markets, and recalled attention to the May 6, 2010 flash crash, where the Dow Jones Industrial Average crashed around 700 points before recovering the losses within minutes. During the 2010 flash crash, bursts of giant stock orders had flooded the high-speed equity trading markets. It was after the flash crash that the Securities and Exchange Commission and Commodity Futures Trading Commission launched circuit breakers for single stock and index product trading, which was triggered in halting Apple’s trades on March 23.

The SEC has announced a new probe into high frequency trading. In particular, the SEC plans to probe into whether high frequency traders used their relationships with automated stock exchanges like BATS to conduct “quote stuffing” and other practices, potentially leading to market drop, according to the Wall Street Journal. The inquiry may extend to the May 23 glitches at BATS.

Georgetown University financial markets professor James Angel, who had testified before Congress on the risks of high frequency trading, noted that the stoppage in the trading of Apple shares on March 23 indicated the problem with circuit breakers, according to the International Financial Law Review.

“If there was a systemic thing hitting Apple you want the circuit breakers to kick in, but there was no change in supply and demand [of Apple stock],” Angel told the IFLR. Angel supports use of limit up – limit down bands, proposed last April by a group of national securities exchanges that included BATS, NASDAQ, NYSE and the Chicago Stock Exchange. According to Angel, such bands would have prevented the halting of the market by introducing a band outside of which trade cannot happen for individual securities.

However, the calculations involved to determine the price bands would be very complex, as Angel acknowledged. He advocates a simpler limit up – limit down plan that has a less complicated algorithm, so that the risk of a bandwidth overload computing error could be lower, while trading stoppages of equities, such as happened with Apple on March 23, could be prevented. As BATS recovered from the crash, there does seem to be a need to probe further into the infrastructures of high-frequency trading.