Testing, testing and more testing. That’s the theme being echoed by both the buy and sell sides as they seek to hone their algorithmic trading strategies.

“A lot of people feel there’s a magic bullet, but throughout the past decade the ability to experiment with data analytics to judge the outcomes of different approaches has proven to be of utmost importance,” said Nick Nielsen, head of quantitative trading at Marshall Wace, at a panel discussion in New York last week.

“You need to create repeatable and sustainable processes to obtain a statistically significant sample and constantly improve on one aspect, however small. It’s a continual experimentation game.”

Marshall Wace is an equity long/short hedge fund that employs a process-driven and fundamental-driven trading strategy across multiple geographies and time horizons, Nielsen said.

For the sell side, creating and packaging algorithms has become an industry unto itself and quality assurance has taken on increasing importance.

For example, a new pre-market trading algorithm released by ConvergEx Group, a technology firm, was developed using ConvergEx’s Automated Testing Facility, which rigorously back-tested over 20,000 different trading scenarios using years of pre-market trading data.

These regressions allow ConvergEx to ensure the algorithm functions as it should in any number of market conditions and allows the company to roll out reliable products much faster than those testing code using older QA.

“Traders have generally had a difficult time executing efficiently during pre-market hours because, until now, it could only be done manually,” said Gary Ardell, head of the financial engineering and advanced trading solutions group at ConvergEx, in a statement. “Now they have an intelligent tool at their disposal that automates an otherwise labor intensive and tricky process.”

The algorithm, said Ardell, is another example of ConvergEx’s “commitment to provide unique solutions for our customers’ trading challenges”.

The pre-market trading algorithm allows customers to capture liquidity outside of traditional market hours. This is particularly valuable during earnings seasons or during market-moving news when traders are more frequently looking to buy or sell positions before the markets open.

“Pre-market trading is highly complicated,” said Scott Daspin, managing director of the global electronic execution group at ConvergEx. “Unlike normal market hours, liquidity tends to be sparse, prices can be less rational and certain orders can cause major movements. We have created an algorithm that addresses these complexities and effectively trades in the high volatility and thin liquidity that characterizes pre-market hours.”