On 6 May 2010, the Dow Jones Industrial Average suffered a ‘Flash Crash’, dropping almost 1000 points in five minutes. On 1 October the Securities Exchange Commission and the Commodity Futures Trading Commission published a report on the causes of the crash and confirmed the role that was played by high-frequency algorithmic trading.
The ‘flash crash’ suggests that post-trade surveillance is no longer adequate for modern markets – exchanges need to employ real-time pre-trade checks and alerts.
Algorithms have the ability to determine and place trades by the millisecond allowing participants to disguise, manage, and profit from their trading activity at lightning-fast speeds.
Ever more sophisticated algorithms have increased trading volumes and created demand for ever faster exchange infrastructures. Exchanges have reacted by increasing execution speeds and reducing network latency to retain market position.
Controlling machines with machines
To prevent runaway losses or rogue behaviour, trading firms are beginning to implement systems to validate, monitor and manage their own algorithms.
Exchanges and market places must also play their part by monitoring participants’ algorithms and halting them if they threaten the orderly functioning of the market. While not mandated by the regulator, it is the duty of any responsible trading venue to invest in their own algorithms to monitor and inhibit questionable trading and market patterns.
Although exchanges have not matched the risk management investments of trading firms, they carry a reputational risk that could see an erosion and possible exodus of participant trading. An exchange’s failure to protect its reputation as a paragon of fairness and control may lead both to a loss of confidence and, in turn, liquidity. With new trading venues springing up every day using the latest technology and most streamlined market practices, there is likely to be great competition for the business of anxious participants of other exchanges.
Where an exchange does not actively constrain the use of algorithms, it must consider pitting machine against machine. An exchange’s algorithm strategy should ask:
Should algos be regarded as an opportunity or a threat?
How will increased order volume and reduced order-to-trade ratios effect operating costs and fee models?
What operational procedures and technical systems can be employed to manage the strategic response?
Can current procedures react to rapid sub-second trades and complex trading patterns?
Can crude, market-wide circuit breakers that close markets for hours be avoided by using more sophisticated algorithmic monitoring solutions?
Exchanges must understand the algorithmic maturity and goals of their market’s participants to assess the suitability of their existing market infrastructure and control procedures. Unless trading venues focus on the impact of the algos, the exchanges’ human-centric risk management approaches may not be able to mitigate against the rise of the machines!
PA has been successfully predicting the impact of algorithmic trading on exchanges for several years. To speak to one of PA’s experts about developing a strategy to protect against algorithms please, contact us now.