Stock trading performed almost entirely by computers is on the rise, provoking some weighty questions when it comes to markets: Will investing in shares become like chess, where only the very best humans can beat machines? Or is a human touch always superior to cold calculation?
Algorithmic trading — the use of computer programs for trading stocks — is a simpler method of trading with very little human interaction. High-frequency trading, the more robust of computer-driven trading methods, is much faster and sophisticated. In high-frequency trading, computer programs analyze market data to capture trading opportunities that may open for only a fraction of a second.
Progress Software , which develops algorithmic systems, is working on applications that can “self-evolve” to find trading patterns "in a genetic way, in a Darwinian sense, seeing which can be the most profitable", said Dr. John Bates, chief technology officer at the company.
Humans will become “more the validator than the coordinator” of such programs, he added. In Bates’ view, it will be the machine, rather than the human, that takes the lead on the trading process.
But more traditional thinkers hold that trading is an art, not a science. Renee Coyler, founder of research and consulting firm Forefactor, has spent much time reviewing competitive trading technologies, exchanges, alternative execution venues and trends in electronic and algorithmic trading.
She disagrees that machines will supplant humans in trading stocks.
“We have to remember that machines are created by man and that there is always someone at the end of the line creating algorithms, creating all the various technologies. It will always come back to man, no matter what. Man will have built in his or her own set of trading strategies, usually based on market analytics,” Coyler told CNBC.com.
"Individuals trade based on their comfort level with risk; they trade based on the market events occurring at that particular time,” she said.
Computer-based trading is always changing for the markets, ferreting out new patterns in stock movement. Today, the average shelf life of a trading algorithm is three months, Bates told CNBC.com.
“The shelf life starts when you discover a trading pattern — for example, a statistical arbitrage where two stocks are correlated, and when one goes up the other goes down — you can buy one and sell the other to make a profit. The strategy may make money today but in three months it might not,” said Bates.
But automated trading can cause serious concerns, as the so-called "flash crash" of May 6, 2010 proved, when a single computer-driven trade is believed to have been behind an almost 900-point intraday plunge by the Dow Jones Industrial Average .
Regulators such as the Securities and Exchange Commission have warned firms of the dangers of automated trading.
Surveillance systems and circuit breakers have also been introduced and put in place by many exchanges to prevent an outright market collapse. Meanwhile, trading continues rapidly changing and new, innovative trading technologies continue to be invented.