James Rundle: Imperfect Solutions

Just two years ago, the word “Knightmare” inspired memories of a British children’s show from the 1990s, in which a team had to guide a blindfolded friend through a series of virtual mazes by verbal direction alone. Now, of course, the word has taken on a more sinister connotation, being the algorithmic trading glitch that practically bankrupted broker-dealer Knight Capital in the summer of 2012, leading to its eventual acquisition by high-frequency trading (HFT) specialist Getco, and the formation of securities giant, KCG.

Knight’s troubles have certainly been the most high-profile example of algorithmic glitches, but by no means the only one. Just last year, Goldman Sachs erroneously entered hundreds of millions of dollars in options trades, an event that was obscured shortly afterward by the pricing engine fiasco at Nasdaq OMX, which shut the exchange for the afternoon. This followed on from the Flash Crash of 2010, which saw the Dow Jones Industrial Average lose over 900 points in a matter of minutes, only to recover quickly, a move that may not have been caused by algorithmic trading, but was certainly aggravated by it.

Pulling the Plug
These events have led to widespread calls from regulators and industry bodies for the implementation of “kill switches”—a mechanism by which trading is halted if prices for a security or an index swing aberrantly, and a topic covered in depth by my colleague, Marina Daras, in this issue of the magazine (see page 14). Some exchanges, such as the London Stock Exchange and the Chicago Mercantile Exchange (CME), have had such mechanisms in place for years already. Indeed, in a recent conversation, a CME executive credited the exchange’s controls with actually allowing the markets to recover after the Flash Crash.

In the Asia-Pacific region, though, controls such as kill switches aren’t as uniform or as widespread as in Europe or the US. This is partly due to the region’s fragmented regulatory regimes, but it’s also because algorithmic trading and HFT are still nascent in some markets. The presence of these trading models, though, is starting to introduce the same story.

For instance, after placing thousands of incorrect trades in Kospi options, South Korean derivatives broker HanMeg Securities faces a KRW42.6 billion ($4.3 million) loss, while Chinese brokerage Everbright Securities was fined RMB532 million ($85 million) by the Chinese Securities Regulatory Commission after erroneous electronic trades caused huge price discrepancies on the Shanghai exchange.

Though nascent in some Asian markets, the growing use of HFT models is introducing the potential for new “Knightmares.”

Wrist Slaps
These events have led exchanges in the region to hasten the implementation of circuit breakers and kill switches, but while they’re a sensible move to at least try to restrict the effects of technological disaster, they can’t provide a full solution. Even more advanced risk management mechanisms such as cancel-on-disconnect, single-stock breakers or thorough pre-trade risk checks aren’t 100 percent effective.

These problems, generally, start at the source. The exchange doesn’t build the trading engines used at brokerage houses to pump thousands of orders into an open market, every day. It doesn’t employ the software engineers and the market-structure experts, or the algorithmic designers. The race to be the fastest, to provide the most efficient execution and get the best price often introduces a tunnel-vision mentality of design, build, deploy, with a minimal test element.

Safety is becoming paramount thanks to enhanced regulatory attention in the US and Europe, but in the Asia-Pacific region, individual regulators need to pay closer attention to the technology entering their markets, and adjust their policies accordingly. It can’t be left to the exchanges to pick up the regulatory slack, and clearly, it can’t be left to the trading entities themselves, either.