Skip to content

Finding alpha with machine readable news

Eric Fischkin
Eric Fischkin
Proposition Director for Machine Readable News, Refinitiv

Machine readable news based on Reuters' extensive global coverage is providing quant investors with powerful signals in the #SearchForAlpha. When news and sentiment can take hours, days or even months to play out, what are the benefits of using Refinitiv's natural language processing tools for quantitative research.

  1. Refinitiv’s Machine Readable News leverages Reuters’ extensive global coverage and natural language processing to create actionable signals for quant investment strategies.
  2. Opportunities for quant investors using machine readable news exist in the slower reactions to news and social media that can take hours, days and even months to play out.
  3. Machine readable news data is now attracting regular interest from market surveillance, counterparty risk and sell-side research, among other groups.

From Reuters’ early days, investors have valued swift access to the general and business news that drives their trading and investment strategies.

The media carrying news may have changed but its power has not, especially now in the era of data-intensive quant investing. As investors continually seek new paths to alpha, they increasingly view news as a timely and essential multi-faceted data source.

Refinitiv’s Machine Readable News leverages Reuters’ deep news experience, depth, and global coverage into actionable signals informing investment strategies.

The service has its roots in the days when customers traded on Reuters-based news and economic indicator releases over their desktop applications.

From there, an up-to-the-millisecond event-based trading business was born, followed by strategies that took a longer view, based on automatically scored sentiment.

Today, the larger opportunities for market participants reside in the slower reactions to news and social media that can take hours, days and even months to play out.

These non-event-based reactions are driven by the speed at which people process information and incorporate it into complex models, as well as by emotions that drive irrational transactional behaviour.

Machine readable news data sees regular interest from. Finding alpha with machine readable news

Screening for ‘greed’ and ‘fear’

Quantitative research has demonstrated that news flow and sentiment enhance standard quantitative anomalies such as the post-earnings announcement drift.

It also has shown that market-level news sentiment can help indicate when the risk-return relationship predicted by the Capital Asset Pricing Model holds true or when low-risk stocks may outperform their higher-risk peers on a risk-adjusted basis.

The convergence of asset management wisdom, behavioural finance and quantitative modelling has given birth to a new breed of “quantamental” investors who screen for “greed” and “fear” with definable measures and history suitable for backtesting.

Quantitative research

As a result, large asset managers such as NN Investment Partners and Union Investment Group have embraced these datasets in adapting their investment strategies to uncorrelated sources of alpha.

Financial companies with a mandate for innovation and backed by data scientists are approaching news and metadata with creativity and fervour.

Having established a solid footing in market surveillance, the use of machine readable news data is now attracting regular interest for new uses such as market risk, counterparty risk and sell-side research.

How does Refinitiv stand apart?

Refinitiv has stood at the forefront of the machine readable news business since its inception well over 10 years ago. Not only can Refinitiv provide advanced natural language processing (NLP), metadata, and time-stamping, it offers the holistic experience one would expect.

Rather than parse text themselves, quants can rely on measures such as sentiment, company identifiers and event tags to filter for news-based signals that investors react slowly to.

Global metadata teams, who received numerous reference data awards over the years, also master company identifiers and naming that is critical to news metadata.

Our partners in Reuters News enjoy the benefits of a rich taxonomy of topic tagging – as well as state-of-the-art NLP tools that auto-suggest tags to journalists but yield final authority.

Third-party news offerings include global corporate press release wires and exchange wires, and all are normalised with metadata and time-stamping alongside Reuters.

Refinitiv’s textual analytics products incorporate high-quality and proprietary technology, honed to financial content.

The architects of these products deliver high-quality standalone analytics, without benchmarking themselves against a short-lived market dynamic or trading result.

This unique combination of high-quality content and NLP fidelity has produced an ever-growing base of academic and professional research that attests to the efficacy of these datasets.

Discover more about Refinitiv’s Machine Readable News

The search for alpha starts here