CausaLens, a provider of signal testing technology, has released an update to its platform to incorporate more robust cause-and-effect relationship mapping.

The three-year-old company uses proprietary automated machine learning (AutoML) to build predictive models at scale and claims to shorten the time data users need to evaluate different sources of data. It says its platform also cleans data and can independently discover the data sources or signals that are best suited to a particular strategy.

Being able to map causal relationships among data is extremely important for asset managers. Traditionally, machine learning algorithms (like deep learning) work by discovering correlations between data points, which are then used by the machine to make predictions on future outcomes.

Finding those correlations in the data can help asset managers make investing decisions, but discovering causality (i.e., a data point that causes a specific effect on another data point) is widely thought to represent a new frontier for the investment management community. Financial markets are more complex than academic use-cases, and therefore correlations can be a less effective tool for predicting future outcomes than causations can be.

CausaLens is adding its new research to its existing platform, an automated machine learning tool for time-series data. The company says the updated product will retain its ability to clean, sort and monitor hundreds of datasets at the same time, but will layer on the new research, which “teaches machines to understand cause and effect.”

The launch represents a “step towards true artificial intelligence,” according to Alejandro Ortega, the firm’s director of scientific communications.

He added that the updated platform could be used to understand cause-and-effect relationships impacting business KPIs, including trading strategy optimisation.

Moving forward, the CausaLens team plans to devote more time to improving the causality capabilities of its platform, as well as personalising the KPI optimisation functions for a wide range of industries.

Photo by Josh Rose on Unsplash