When Bruce Bittles first started trading in the 1960s, the US stock market was a largely human affair. Exchange floors were the chaotic maelstrom of shouts, frantic phone calls and finger waving, later made famous by 1980s films such as Wall Street and Trading Places. But now the machines have taken over.
Nasdaq became the world’s first electronic stock market when it opened its doors in 1971, but since then, the trading world has been revolutionised several times over. The old bourses and trading pits now are largely shuttered. Virtually all stock trades are done electronically in data centres.
The rise of modern electronic markets has also led to the birth of a new breed of traders: algorithms that either execute transactions on behalf of investors, scan markets for profitable opportunities or even buy and sell securities systematically — and largely autonomously from human control.
This is a development that has been under way for several decades but trading “algos” have grown increasingly complex, sophisticated and fast in recent years, sometimes even using artificial intelligence techniques to reach their decisions and help them adapt dynamically to shifting market conditions.
While this has hammered down trading costs to nearly zero, enabled the rise of popular vehicles such as exchange traded funds (ETFs) and allowed asset managers to deploy increasingly sophisticated strategies, it also means that markets can sometimes move in mysterious ways that can befuddle and frustrate human traders and investors.
“For much of my career, I had a pretty good idea of what would happen the next day. That’s no longer true,” laments Mr Bittles, now chief investment strategist at Baird, a wealth management firm.
Trading algorithms come in many shapes, from the deceptively simple to the staggeringly complex, and are used in many ways. But the advantages are clear. Trading has never been easier and costs never lower thanks to human intermediaries being rendered obsolete.
Algorithms have also nurtured developments such as exchange traded funds, which have revolutionised the industry and brought advantages to millions of ordinary investors.
In addition, the US stock market bid-ask spreads — the difference between the price for selling and buying a security and a handy gauge of transaction costs — have collapsed by 95 per cent since 1994, the Managed Funds Association noted in a report on algorithmic trading last year. “Over the last five decades, technology and automation have brought significant benefits to investors, including greater accessibility, lower transaction costs and fairer markets,” the association argued.
These days, virtually every money manager, broker or day trader uses algorithms in some form. Most of the bond market remains the domain of human traders, but algorithmic trading has become increasingly important in the trading of US Treasuries, too. Indeed, algos have rendered human traders obsolete in many areas.
“I think the most obvious advantage of algorithmic trading is the reduced cost and scalability,” says Christina Qi, who works at Domeyard, a high-frequency trader.
“Some people think that your execution costs will magically go down if you are faster,” she adds. “What I mean is that hiring a trader costs a lot and, more importantly, adds very little value. You can back this up with data. We’ve seen many companies replace traditional traders with computers.”
But as a result, the stock market has grown increasingly complex and often confusing, with millions of algos sparring for advantages in the electronic markets of today. This may be leading to negative side effects.
For example, to counter HFT firms that have largely supplanted the old “market makers” of Nasdaq and the New York Stock Exchange’s “specialists” — dedicated intermediaries that facilitate trading — and to avoid moving markets too severely, many asset managers splice and dice big buy and sell orders into smaller bites and drip them into markets at random intervals.
“In the old days we could do ‘big tickets’, but we have to be much more careful these days,” says Patrik Safvenblad, chief investment officer at Harmonic Capital Partners, a hedge fund.
“Technology is both helping and hurting us. It means we can seep out orders more gradually, but because everyone does it the liquidity suffers.”
Despite assiduous risk management controls, things can go wrong. Most memorably, a poorly executed automated sell order by a big US asset manager triggered the 2010 “flash crash”, and in 2012, Knight Capital, a high-frequency trading firm, imploded after an errant algo lost it about $440m in a 30-45-minute trading frenzy.
“Sometimes all computers do is replace human stupidity with machine stupidity. And, thanks to speed and preprogrammed conviction, machine stupidity can devour markets far faster than any human panic can achieve,” Gavekal, a brokerage, noted at the time.
The market ripples and waves caused by automated investment flows can also frustrate investors. Nevsky Capital, a London-based hedge fund, earlier this year closed down in part because the “current algorithmically-driven market environment is one which is increasingly incompatible with our fundamental, research-oriented investment process”.
In its final letter to investors, Nevsky wrote: “Butterflies flapping their wings now regularly create hurricanes that [hurt] fundamentally driven investors who cannot remain solvent longer than the market can remain irrational.”
Even some of the proselytes of the revolutionary benefits of technology fret that this complexity makes modern markets vulnerable to glitches that can have devastating impacts at high speed.
The benefits are real, but so too are the risks.