Taming Runaway Algos in the Wake of Knight
As regulators go about piecing together the events that led to last week’s Knight Capital trading fiasco, one of the pillars of pre-trade risk regulations—namely, that they are aimed primarily at hedge funds and other buy side institutions—appears to have been undermined.
“Last week’s events at a major market maker were no black swan—this kind of failure is rapidly becoming commonplace, as algos become faster, more autonomous and more sophisticated,” said Michael Chin, CEO of risk management software provider Mantara.
Knight has brought attention back over to the sell side, and has made regulators aware that the same types of risks can occur with automated market making strategies executed by the sell side.
“Following a series of mishaps which include the Flash Crash, the Facebook IPO, and now Knight Capital, its electronic trading platforms and operations that are about to receive the most attention,” said James Frischling, president and co-founder of NewOak Capital.
The SEC’s rule 15c3-5 applies to broker-dealers with market access to an exchange or alternative trading system, as well as to broker-dealers that provide customers or other persons with access to trading securities directly on an exchange or alternative trading system.
Under the rule, the broker-dealer is responsible for having risk management controls and supervisory procedures reasonably designed to ensure compliance with all laws, rules and regulations.
But these controls are aimed primarily at preventing runaway algos used by buy side firms, not at the broker-dealer itself.
“So much of the focus on automated trading and preventing runaway algos has been focused on the buy side, and as a result of the 2010 flash crash and all the buzz around HFTs, the buy side have been implementing risk controls,” said Michael Chin, CEO of risk management software provider Mantara.
It seems obvious to state that systems must have failsafe mechanisms, Chin said, but the obvious was apparently not included in the design spec.
“Effectively managing an automated trading strategy requires the ability to monitor it and the ability to kill it at a moment’s notice,” said Chin. “This monitoring mechanism must sit outside the trading architecture, lest a cascading failure take it out as well. The kill switch must be able to kill the strategy, ideally at the order level and at the network level.”
Mantara’s E-Brake is a hardware-based switch that sends kill messages directly into network switches, such as Arista or Cisco, immediately terminating the session.
Firms can set kill parameters for single, multiple or all strategies based on a variety of risk parameters, such as order size and frequency average daily volume, and duplicate order checks.
“If Knight had visibility into average daily volume and duplicate orders, it would have been able to detect that something had gone awry with its market making strategy,” said Chin. “But even if it had this visibility, it still would have needed a mechanism to shut the trading strategy down, which is what the kill switch does.”
Algorithms have become more prevalent in the spot FX market.
QB’s Algo Suite for futures market trade execution is also being co-located to HKEX.
Breaking data silos is key to deploying automation beyond 'nuisance' orders.
They can be used on quantum hardware expected to be available in 5 to 10 years.
Streaming blocks change the basis of matching and price discovery so institutions can find new liquidity.