In today’s volatile trading conditions the useful life of quant-driven automated trading strategies has become shorter, according to John Bates, founder and general manager of the Apama division of Progress Software.
In a study released in July, research and consulting firm Aite Group commented that short-term quantitative trading strategies only remain effective for between three or four months. But the recent ups and downs are creating opportunities for profit that can disappear as rapidly as they materialise, meaning that yesterday’s strategy may no longer fit the bill, and needs to be revised or replaced.
“In this climate, the average life cycle of a trading algorithm is probably one day, given the amount of volatility and strange things that are going on!” says Bates. “The period from concept to profit is being squeezed from months into days and possibly even hours.”
Bates, whose firm offers a complex event processing platform for designing, testing and implementing automated strategies, argues that traditional development methods are simply not quick enough to cope. “Are you going to get a team of 20 developers to design and implement a trading algorithm in 24 hours? It’s not going to happen,” he says. “This is why new approaches are going to continue to gain momentum, such as rapid development through graphical modelling and business users being directly involved in development.”
Complex event processing and graphic modelling allow users to develop strategies from ready-made ‘building blocks’ of code that can be arranged accordingly. “You have in your toolkit a lot of reusable components that you can plug together and put custom rules on top so that you can build a strategy out of the order management pieces, analytics, moving averages etc. and get something working quickly,” Bates explains.
He acknowledges that the newer, faster tools for developing trading strategies are not perfect yet. “I don’t think we’re at the ultimate stage yet,” he says. “But firms are buying into the concepts and seeing the value in being able to build strategies quicker rather than using old development methods.”
With both trade generation and execution growing ever faster and more automated, an awareness of risks is also growing. “We have seen an increase in the real-time risk management aspects of algorithmic trading,” says Bates. “People think algorithms are great to capitalise on opportunities faster than a human can manually, but what about the exposure?”
In the past, he argues, some firms have been relatively relaxed about the risks, arguing that if they had one bad trading day, the profits from the rest of the year would make up for it. “But in this climate, that attitude has significantly changed,” says Bates. “Now everybody is interested in what their exposure is at a given point in time and what the safeguards are. That’s a really big focus.”
To allow users to monitor risks and exposures from their automated trading models, Apama has developed what it calls a ‘risk firewall’. “Just as a network firewall monitors packets of data as they are coming in, this firewall is looking at trades going out and is constantly monitoring exposure and what’s happening in the market,” he explains. “It will stop trades in real time should the trader’s exposure exceed a certain level. It also provides a real-time view of risk for the whole institution, not just one trader, and multiple asset classes.”