When the Dow Jones stock market index , fingers were soon pointing at the high-frequency traders and the computer programs they rely on. Are computers now too powerful to be allowed in financial markets?
Could computer traders bring down Wall Street?
It certainly seemed plausible at 2.30聽pm in New York on 6聽May. Traders went into a panic as the Dow Jones index, which follows 30 large publicly traded companies plunged an unprecedented 6聽per cent in 20 minutes 鈥� for no apparent reason.
Advertisement
Such a drop represents billions of dollars being wiped off a company鈥檚 value, although in this case prices quickly bounced back. The reasons for the plunge remain unclear, but high-frequency traders 鈥� who use powerful computer algorithms 鈥� are in the frame.
The traders use computers to profit from short-lived fluctuations in markets. An algorithm might, for example, watch for large transactions from institutional investors that could affect a stock鈥檚 price and make trades before the rest of the market has time to react. That and similar strategies have created a market for high-frequency traders in which transactions amounted to around $8聽billion last year.
These near-instantaneous trades make some economists nervous. They fear that the algorithms could interact to create a feedback loop of continuous selling, driving market prices off a cliff. Although the cause of last week鈥檚 volatility is still unclear, it seems that algorithmic trading played a role.
But computer trading has been around for years. Why the growing concern?
Algorithmic trading is well established, but the speed at which trades are executed 鈥� usually milliseconds 鈥� is shrinking fast: by a factor of 10 since 2007, says at Tabb Group, a firm based in Westborough, Massachusetts, that studies financial markets.
These lightning-fast speeds are due to investment in dedicated optical-fibre networks and faster routing devices. Speed is so critical that high-frequency traders must have their computers physically as close to markets as possible. It takes over 0.01聽seconds for a signal to travel to Chicago to New York and back via optical fibre, a delay that Chicago-based high-frequency traders cannot afford.
鈥淓verybody has relocated,鈥� says , a finance professor at Baruch College in New York. 鈥淣ot doing it means you鈥檙e behind the curve. Nobody will do business with you.鈥�
Should high-frequency traders be banned?
In any market, buyers need to find sellers and vice versa. And if traders can鈥檛 find a partner, funds stay locked up in existing investments when they could be used elsewhere instead. That鈥檚 why all markets want more of what economists call 鈥渓iquidity鈥� 鈥� the more liquidity there is, the easier it is for traders to do business.
High-frequency traders are valued by the companies who run markets because they buy and sell large volumes every day, which adds liquidity to the system. The number of trades on the New York Stock Exchange rose from 3聽million per day in 2005 to 22聽million by 2009, largely thanks to the rise of high-frequency trading.
Meanwhile, transaction times have been dropping: the average execution time for one class of small trade on the New York stock exchange fell from 10聽seconds to 0.7聽seconds between 2005 and 2009.
What can be done to improve market stability?
鈥淭he big problem,鈥� says , who studies financial markets at Georgetown University in Washington DC, 鈥渋s that US exchanges have no real-time safeguard against extreme malfunction.鈥� If a mistake in an algorithm 鈥� or even deliberate sabotage 鈥� were to set off a wave of selling, the people tasked with suspending markets in the event of dangerous trading would not be able to react quick enough.
鈥淲e could see a major crash within milliseconds,鈥� says Angel. 鈥淚s this a high-probability scenario? No. But can it happen? Yes.鈥� He proposes that the require that all markets incorporate a 鈥渃ircuit-breaker鈥� which would suspend trading on signs of an algorithm-driven crash.
High-frequency trading should be allowed to continue in the meantime, say most financial researchers. But much about the practice remains to be understood. Donefer backs a proposed SEC rule that would require firms to identify the type of strategy 鈥� the market condition the algorithm is designed to exploit 鈥� behind each trade. The information would be seen only by the commission, which would use it to study the positive and negative effects of different trading strategies. Armed with that information, regulators might be able to do a more thorough post-mortem next time the Dow takes a plunge.
