My colleague Paul Britt’s blog has generated an interesting debate on the pros and cons of high-frequency trading.
Britt refers to the dangers of placing a market order—that is, an order to buy or sell a security, say an exchange-traded fund or share—at the best prevailing bid or offer price, when it’s possible that the true market lies elsewhere. Nothing to worry about, says a poster called Kid Dynamite, commenting on the blog. Such “concealed” orders may well be within the published market spread, allowing even better execution. Another commenter, Vance Harwood, is more concerned, referring darkly to the “many bad things” that can happen if you trade “at market”.
Britt surely has a point in arguing that an exchange on which 98 percent of orders are cancelled before execution resembles a gigantic game of bluff more than a real marketplace. No-one can dispute that headline bid-offer spreads have declined markedly since the proliferation of automated trading, to the apparent benefit of all. At the same time, the 2010 “flash crash” also left no doubt about the inherent weaknesses of large, unsupervised, complex networks of interconnected computers. As Britt points out, mini-flash crashes, such as the recent one in Apple shares, continue to occur.
It’s unsurprising that exchanges are beginning to have a rethink of the way they incentivise high-frequency traders, moving away from bulk fee rebates paid to those who post orders to discriminating more between orders that actually result in trade execution and those that don’t.
Despite such moves, some expert observers remain concerned. Dr Christopher Clack, director of the financial computing programme at University College London, is an expert on system design. Clack doesn’t pull his punches when it comes to describing current market structure, calling it “an accident waiting to happen”.
Worryingly, argues Clack, while existing investigations into the flash crash have focused on finding a root cause or an explanatory timeline of events, it’s likely that such endeavours are missing the point. We’d all like a neat, Sherlock Holmes-type answer, but it’s probable that an explanation of A caused B caused C simply cannot be found.
Instead, he says, what may be a completely minor triggering event can give rise to a non-linear, ultimately catastrophic response, given the way trading networks are currently put together.
This is food for thought, of course, not just for ETF users, but also for anyone transacting in securities and on exchanges. But Clack’s arguments also suggest that the much-vaunted intraday tradeability that ETFs offer may not always be as much of a virtue as ETF exponents claim. If things go wrong and markets hit an air pocket (or gain a rocket boost) as the result of problems in the way computers communicate with each other, you could end up losing a lot of money, and through absolutely no fault of your own.
Clack will be talking in more detail about his latest research—both in a lunchtime address and in a panel discussion immediately afterwards—at our forthcoming Amsterdam conference, Inside ETFs Europe. ETFs, high-frequency trading and their joint effect on market structure and index correlations represent one of the hottest topics in finance. To debate whether or not anything is wrong, find out what might need to be fixed and agree how that should be done, come and join us on May 16.