Making Big Data Smarter and More Valuable

This article was contributed by Thomson Reuters Exchange, a robust forum for dialogue, where ideas, insights, and information, are shared across the global financial ecosystem.

New technologies for leveraging Big Data are springing up, but can market participants profit just by processing more data more quickly?

Big Data is in the air these days. Key decision makers such as chief technology and chief operating officers, marketing heads and research gurus are hearing a lot about it, often in conjunction with words like volume, velocity, variety and veracity that only seem to increase its importance – and the urgency to know more about it.

New technologies for leveraging Big Data are springing up, but can market participants profit just by processing more data more quickly? How can the flood of information be transformed into actionable insight?

No need to lose sleep. Thomson Reuters brings an unparalleled knowledge of financial markets along with tools that combine data from multiple perspectives – while accounting for accuracy and importance – allowing traders and investors to determine what’s critical and generate an appropriate strategy.

Dealing With Information Overload

What do decision makers need to know? The first step in dealing with conflicting market signals is to combine them to make meaningful comparisons. Thomson Reuters desktop and feeds solutions handle this, bringing together a wide variety of market perspectives and content in one place, enabling intelligent search, and allowing users to manage a portfolio against the broadest set of exposures. Alerts about new developments ensure that money managers don’t miss any crucial information.

No single company can cover all available content. Thomson Reuters solves this problem by building open platforms that blend proprietary content, the best third-party content and compelling new content as it becomes available.

Once it’s all combined, we blend the best of two worlds: human judgment and computer algorithms. This allows clients to take advantage of the knowledge and expertise generated by our research team, subject-matter experts, online messaging communities, and collaborative workflow tools, made stronger by cutting-edge data-crunching. We build intelligent analytics, such as our StarMine models, to boil down entire data sets into a single number, thereby processing millions of records to buy/sell/ignore signals that provide automated alerts about critical new information.

Linking entities and data sets to enable analytics

Thomson Reuters links data sets with a single identifier for each company, security or person – including accurate point-in-time mapping across an ever-changing landscape of entities to enable more reliable analytics. Clients are assured that new information sources are automatically placed into the data stream.

Mining past data can elucidate hidden relationships. Our Atlas technology examines massive quantities of text to assess how different companies are related, connecting previously unrelated data. Text mining techniques can be used for intelligent search, for short-term sentiment analysis, and for long-term credit risk identification. Some customers like to add their own “secret sauce” by linking up their own specialized data to find situations where they have an information advantage.

Intelligent analytics make Big Data smaller and smarter

The Thomson Reuters StarMine team has a successful track record of building robust predictive analytics for equity investors, and we’ve branched out into new asset classes and content sets. We’ve seen two clear trends with predictive analytics upon which core principles can be based: first, historical accuracy matters; and second, the highest probability strategies succeed when multiple factors properly align.

True forecasting is a repeatable skill. Thomson Reuters studies of sell-side equity analysts, economists, independent research providers and quant algorithms show forecasting ability can be consistent over time. Tracking historical accuracy and putting more weight on better performing forecasters provides a sustainable edge.

Profiting in the face of uncertainty requires careful choice among competing investment ideas – bypassing those with a lower chance of success in order to concentrate on those with higher expected payoffs. Uncorrelated signals derived from multiple independent content sets – the more the better – provide a good way to assess the odds of success on any given trade, which are boosted when various factors agree rather than disagree.

Bringing together the widest variety of perspectives, linking the independent data sources with a robust entity classification system and knowing who and what to believe in different situations, all greatly increase the odds of making the right calls in an ever-changing marketplace. And that helps our clients sleep just a little better at night.

Dr. Stephen Malinak

Dr. Stephen Malinak, Global Head of Analytics, shepherds a team that builds innovative analytics, quant factors, and models. Prior to its acquisition by Thomson Reuters, Stephen was Director of Quantitative Research for StarMine. Stephen has a Ph.D. and M.S. from Stanford in Engineering-Economic Systems. He also has a B.S. in Electrical Engineering and Computer Science from MIT, where he studied acoustics under Dr. Bose.

Enjoy this article?
Get your FREE full-copy of Thomson Reuters Exchange Magazine for iPad and Android tablet devices on the Apple App Store, Google Play, or the Amazon Appstore. Or, get your free single article by visiting the Thomson Reuters Exchange Website.