A series of recent market glitches has raised questions about the increasingly common practice of algorithmic trading. According to Thomas Peterffy, who introduced the first automated trades by hooking an IBM computer up to a Nasdaq terminal in the 1980s, computerized trading systems need more layers of control to mitigate the risk of causing market abnormalities.
Peterffy told Traders Magazine that he believes all trading will eventually be done electronically, with algorithms handling any exchange based on technical analysis. The role of human traders will shift to overseeing broader portfolio management and managing the algorithms that handle the actual trades, he said.
“They will feed the order into the algorithms and the algorithms will execute the order,” he told Traders Magazine, adding that human eyes will still be necessary to manage programs and catch algorithm errors before they cause significant damage.
Concerns with algorithmic trading
Several recent market events have brought algorithmic trading practices into the spotlight. In the last month, software glitches led to shutdowns on the Moscow and Stockholm Exchanges. In the latter case, the exchange’s order book software failed to catch an abnormal futures purchase valued at more than 130 times the total Swedish GDP.
In October, Kraft Foods stock prices skyrocketed following a switch from the New York Stock Exchange to the Nasdaq index, causing some onlookers to blame high-frequency algorithmic trading, CFO.com reported. Such occurrences have raised questions about the integrity of electronic trading infrastructure.
“Matters of infrastructure are essential to any holistic approach to improving how our markets operate,” Securities and Exchange Commission (SEC) chairwoman Mary Schapiro said in a recent SEC roundtable, according to CFO.com
Automatic trading glitches have also had substantial financial consequences. A software update at Knight Capital on August 1 set off a flood of mistaken orders, ultimately resulting in losses of $456.7 million in 45 minutes, Traders Magazine noted, adding that the company is expected to field offers for its takeover in the coming weeks.
Improving algorithmic trading controls
Some industry voices have suggested improving control over algorithmic trading by slowing down trades or charging a transaction tax, Traders Magazine noted. However, according to Peterffy, a more effective solution might be to establish new layers of protections against erroneous trades.
Some existing layers of protection include single-stock and market-wide circuit breakers established by the SEC following the market flash crash in 2010 and a 2011 market access rule. Peterffy suggested adding a third layer of software on brokers’ sites that would control against unwanted trades.
“It would evaluate each order in terms of the resulting position as part of the pre-existing position and if such position could not be supported within the broker’s or its customer’s capital then the order would have to be rejected,” he told Traders Magazine.
He also recommended instituting an added layer of safety on the side of the stock exchanges. By clarifying the controls for what constitutes an erroneous trade, many current trading problems could be eliminated.
Implementing such controls begins by instituting a more robust development process, CFO.com suggested. By using techniques such as static analysis or peer code review, developers can improve controls to anticipate erroneous trades and introduce a greater level of human oversight into automated trading software. Adding this extra layer of testing and judgment may have significant effects in the long run.
“An electronic-trading problem is only an electronic-trading problem for a minute,” TABB Group CEO Larry Tabb at an October Senate committee hearing, according to CFO.com. “After that it is a human problem.”
Software news brought to you by Klocwork Inc., dedicated to helping software developers create better code with every keystroke.