As algorithmic trading and machine learning take a bigger role in the financial world, the importance of high-quality data has never been more critical. That was one of the themes that emerged from the discussions at the Equity Leaders Summit (ELS) in Miami this week.
With the increasing complexity of financial markets, the need for accurate, real-time, and reliable data has never been more crucial. The panelists also discussed the dangers of relying on poor-quality or delayed data.
“For traders, we’re seeing an ever-increasing requirement for high quality data,” Dr Elliot Banks, Chief Product Officer at BMLL, told Traders Magazine.
Currently, trading firms are spending enormous resources on cleaning and scrubbing data to make it usable for traders, he said.
“Firms should understand the usability, reliability and ease of use when evaluating data sources. Otherwise, you’ll spend four times as long cleaning the data as deriving useful value from it,” he commented.
Banks explained that when it comes to historical data, all facets of quality are important.
That means data that is: accurate – interpreted correctly with edge cases in the data handled correctly, with no gaps or crossed books; consistent – there are 16 venues across the US equity market alone. Having a consistent, normalised schema that is global in nature is essential; and complete – no missing fields that have been dropped during capturing processes.
Banks emphasized that one of the biggest challenges traders face is accessing clean, high-quality historical data for backtesting and execution analysis. With poor-quality data—whether outdated, incomplete, or misformatted—traders risk making decisions based on inaccurate information.
He noted that quants spend up to 80% of their time cleaning and preparing data, and as data volumes grow, the need for reliable historical data has never been more critical.
“And with ever growing volumes of data, access to good historical data is becoming ever more important,” he commented.
“We’re seeing an increased demand for high quality data from traders, whether that’s for traditional backtesting or TCA, or to fuel innovative new solutions leveraging the latest techniques in machine learning and AI,” he said.
“We’re most excited by the fact that high quality, consistent historical data means that firms can spend more time on these solutions, rather than data cleaning,” he added.
In the age of algorithmic trading, high-quality data is crucial, confirmed Rob Laible is BMLL’s Head of Americas.
Speaking on the sidelines of the conference, he said that to build effective models, you need a large, well-maintained data set (or data lake) because if there are gaps or anomalies in the data, the machine learning model won’t recognize those issues. So, the foundation is a solid database with reliable data, and once that’s established, you can start applying machine learning to it.
Complex markets will continue to require ever more complex decision making around routing and trading decisions. There are 16 US equity exchanges now, but more on the horizon, Banks said, adding that’s the same in the US options markets.
“And demands on firms to continue to prove execution quality in this increasingly complex environment will only increase the demand on firms to have high quality, consistent historical market data,” he said.
In today’s fragmented equity markets, with multiple exchanges and many different ways to execute, having high quality historical data is essential, Banks said.
“Understanding which venue has the highest probability of being filled, or what exchange is setting best prices most often, is a critical question for anyone making routing or execution decisions,” he said.
“And these questions are impossible to answer without extremely high quality historical data. One firm told us that “if historical data quality is only at 99%, you might as well not bother”,” he said.
However, not all data sources are created equal. The accuracy of the data provided can vary, which is why selecting reliable sources is paramount.
“There are some potentially nice regional players or single market players, or people go to particular exchange. You get a fee, but if you’re a global player there are not a lot of avenues you can go down if you’re not going to do it yourself,” Laible said.
As the demands for better execution and complex decision-making continue to grow, it’s clear that traders must prioritize high-quality data to stay competitive. Ensuring data integrity not only improves performance but allows firms to focus on innovating with AI and machine learning rather than getting bogged down by cleaning and correcting datasets.
