A software glitch nearly brought down market broker Knight Capital. Is it time to curb computer-driven high-frequency trading? Matthew Partridge reports.
What happened at Knight Capital?
An error in a line of software code caused mayhem for Knight Capital, which executes an estimated 10% of US equity trades. During the 45 minutes it took to fix the error, it traded, seemingly randomly, in and out of stocks and accumulated huge positions in several companies that it couldn't subsequently fund. The firm tried to get regulators to cancel the trades. However, as The Wall Street Journal reports, the Securities & Exchange Commission refused (ironically, the rules limiting a regulator's ability to unwind trades were brought in after the flash crash' two years ago, which briefly caused the Dow to fall by 1,000 points) and only voided trades in six out of the 140 firms involved. Knight Capital then offloaded the shares to Goldman Sachs at a massive discount, taking a loss of $440m. As a result, its own shares plunged and it ended up selling a large equity stake in itself to a consortium of firms in return for cash.
Exactly what is high-frequency trading'?
Knight Capital's near demise thought to have been caused by a supposedly dormant piece of software has thrown the spotlight on high-frequency trading. This is where banks and hedge funds use automatic computer systems, rather than human judgment, to trade. Programs can be designed to implement a variety of strategies, but many have a common aim: to take advantage of very short-term price movements. As Jerry Adler of Wired puts it, high-frequency trading "makes money a fraction of a cent at a time, multiplied by hundreds of shares, tens of thousands of times a day". Thanks to advances in cable technology and signal speeds, the time needed to communicate an order for, say, listed shares is just 13 milliseconds (0.013 seconds). However, newer systems are expected to cut this down to nine milliseconds.
Try 6 free issues of MoneyWeek today
Get unparalleled financial insight, analysis and expert opinion you can profit from.
Start your trial
Sign up for MoneyWeek’s free twice-daily newsletter.
Join more than 165,000 subscribers and keep yourself informed with latest financial news, insights and expert analysis to help you understand what really matters when it comes to your finances.
Join more than 165,000 subscribers and keep yourself informed with latest financial news, insights and expert analysis to help you understand what really matters when it comes to your finances.
Doesn't this make markets more efficient?
Princeton Professor Burton Malkiel says the incident proves the resilience of modern markets. "When Knight's flawed sell orders hit the market, legions of floor traders and other market participants were happy to buy. Market prices quickly returned to normal." Besides, human traders can also make costly errors. As eFinancial News's Tara Loader Wilkinson notes, seven years ago a typing error by a trainee cost the Japanese bank Mizuho Securities $224m and triggered a 2% fall in the Japanese stock exchange. Alexander Green of Investment U thinks that computers, along with competition and regulation, "have combined to make trading faster, easier, cheaper and more transparent than ever before". He notes that, "in the past decade, spreads were reduced to pennies and some stock spreads today are a fraction of a penny". So all good news then? Hardly.
Latest Videos FromMoneyWeek
For starters, as Professor Lawrence Harris of the University of Southern California notes in The New York Times, trading times "have reached the point where the competition is measured in microseconds and there are essentially no benefits to the public at that level". Indeed, some critics believe that an obsession with speed is actually now pushing up the average cost of a trade. Worse, Felix Salmon of Reuters thinks high-frequency trading has made the stockmarket a lot more risky. "The stockmarket is clearly more dangerous than it was in 2007, with much greater tail risk. At some point in the future significant losses will end up being borne by investors with no direct connection to the high-frequency trading world, which is so complex that its potential systemic repercussions are literally unknowable."
What can be done about it?
Salmon calls for a transaction tax to make high-frequency trading uneconomic. This would give it "the funeral it deserves". In The New York Times, the Manhattan Institute's Nicole Gelinas argues instead for better enforcement of the existing rules. "Market manipulation and front-running other participants' trades is illegal and programs that aim to manipulate prices should therefore be found and shut down." She also suggests that "companies whose programs run amok should be banned from trading and should bear the cost of restitution to other investors, even if the cost forces them into liquidation". In the meantime, regulators are starting to act. India is considering limiting the speed at which trades can take place. European lawmakers are also weighing up curbing several of the most popular strategies. As regulators attempt to restore faith in battered markets, high-frequency trading is likely to remain firmly in the spotlight.
Are we too dependent on software?
Professor James Kwak of University of Connecticut thinks this scandal reveals the extent to which many economic and industrial decisions are underpinned by software. Such software also features in "the huge, creaky applications that run Walmart's supply chain or United's reservation system or a Toyota production line". The key problem is that "most software isn't very good". This makes it prone to errors that can have a big impact, a problem that could get worse. "As computer programs become more important to the financial system and hence the economy, there is insufficient incentive for trading firms to make sure their software works properly". Yet he concedes there are no easy solutions. "How can you write a rule saying that companies have to write good software?" In the end, "the only real solution is to acknowledge that computer programs are going to fail and try to minimise the damage they can cause in advance".
