Stricter risk controls are necessary at all levels of the high-frequency trading cycle, according to the Federal Reserve Bank of Chicago this week. HFT firms aren’t employing risk controls endorsed by industry and regulatory bodies, the bank’s study found, and algorithms run amok more often than expected.
Kill switches that terminate trading at certain levels, as well as limits on order quantities within a certain period were among the risk controls endorsed in Chicago, as were restrictions on how much money one could lose.
Meanwhile lawmakers across the Atlantic want to set a speed limit on HFT by requiring participants to hold positions for at least half a second, according to a position paper on which the European Parliament’s economics committee is set to vote on Wednesday. Such a mandate would doubtlessly blunt electronic traders’ capacity to rapidly exploit unexpected market volatility, but using compliant algorithms could also help provide liquidity, inhibit added volatility and avert another 2010-style Flash Crash.
All of this amounts to perfect technological solutions for HFT skeptics. There’s just one problem:
“When it comes to building software,” Joshua Walsky noted last week, “perfection is not possible.”
Parts of the Problem
Accepting that there can be no reasonable expectation of perfection frees us to decide how to move forward, according to Walsky, CTO of New York-based financial solutions provider Broadway Technology. Moving forward would include treating HFT as system risk, as he said in his op-ed in The Wall Street Journal last month.
“System risk is the possibility of financial loss or gain resulting from systems performing in an unexpected way,” Walsky said. The idea is that people will be responsible for managing system risk, armed with tools, methods and mechanisms.
But people have also been part of the problem. A former CME Group software engineer could spend as much as 10 years in prison and $500,000 in fines for downloading thousands of CME source code files from the electronic trading platform’s operating system, all while planning to help a chemical electronic trading bourse in China.
To solve more conventional problems, the British Financial Services Authority recently announced it would use Nasdaq OMX Group technology to try curbing market abuse among high-speed traders. And the U.S. Security and Exchange Commission contracted with Red Bank, N.J.-based HFT firm Tradeworx earlier this year to monitor the markets.
Traditional trading firms aren’t just watching HFT firms, they’re adapting to them, according to new U.S. trading data.
Trading before their low-latency counterparts during closing auctions helps the less-frequent traders avoid the trap of final-millisecond trades, when high-frequency trading firms lock down closing prices.
Others call for banning HFT, as ECB Governing Council member and Austrian National Bank chief Ewald Nowotny did during a panel discussion at a regulatory conference last week.
What’s the Frequency, Kenneth?
Of course, an all-out prohibition of HFT would be tough to enforce internationally, as every nation would have to prosecute the ban with equal gusto.
A more reasonable solution would be to eschew continuous trading in favor of a universal frequency, according to Chris Sparrow on TABB Group. Instead of getting every region or nation to dial in, it would only require harmony amongst the trading venues.
“Establishing a standard frequency of trading would put an end to the technology arms race by removing the marginal benefits of squeezing an extra microsecond of latency out of the process,” Sparrow said. “If the frequency were 1 Hz (i.e., trade matching occurred at one-second intervals) then participants would be able to use equipment that could process orders within this time window.”
Sparrow predicts such synchronization would:
- Remove the opportunity for latency arb
- Contain the market data explosion and quote stuffing
- Reducing systemic risk
- Increasing confidence of investors
- Level the trading field
But exchanges, firms and regulators would all have to get onboard, which isn’t likely to happen any time soon. Just look at how long it’s taken legal entity identifiers (LEIs) to get off the ground.
Supporting Big Data
For now, the market is moving toward HFT or co-location as more of the norm, according to SAP’s Stuart Grant, who presented at last week’s “HFT Under Attack: Using Big Data to Find New Trading Correlations and Opportunities” webcast.
Grant stressed the significance of a technology strategy that supports vast data volume because it is increasingly important that firms never discard anything. Big Data requires both static storage and streaming technology capabilities.
“Even if volumes move toward dark pools, the strategies are still relevant as they allow you to improve your analysis of the market,” Grant told me after the webcast. “Traditional technologies won’t get you there.”
Perfect is in the eye of the beholder. Different ideas of basic matters, such as how people should make money and where the market is going, make perfection all the more elusive.