SharpeMind & IBM Watson

The first finance app on Watson

In December 2013, the Modulus SharpeMind project launched as the first financial application developed for IBM Watson, built to deliver real-time analysis of unstructured financial data straight to mobile devices. Millions of unstructured text documents, including analyst reports, news feeds, governmental reports, and social media posts, were continuously processed to give traders an instant edge.

The result gave traders on-demand trading signals and consensus-based buy and sell recommendations. Along the way, Modulus invented and patented a method that let consensus systems such as Watson process time-series data, something Watson was never designed to do. Modulus is no longer engaged with IBM Watson, and has since developed its own natural-language-processing systems for financial markets.

The Project

In December 2013, Modulus SharpeMind became the first financial application developed for IBM Watson, designed to deliver real-time analysis of unstructured financial data to mobile devices. Rather than crunching numbers alone, SharpeMind read the written world that moves markets.

Millions of unstructured text documents were ingested from a wide range of sources, then distilled into something a trader could act on in seconds.

• Analyst reports parsed for sentiment and intent
• News feeds monitored continuously in real time
• Governmental reports incorporated into the analysis
• Social media signals folded into the consensus

How It Worked

SharpeMind was built to give traders a fast, reliable way to access trading signals and consensus-based buy and sell recommendations, delivered to the device in their hand.

While building it, Modulus invented and patented a method that allowed consensus systems such as Watson to process time-series data. Watson was not designed for time-series analysis, yet our approach let it create time-series forecasts based on pre-processed text data alone, an unusual and valuable capability for financial use.

About Watson

On May 11, 1997, an IBM computer called Deep Blue beat world chess champion Garry Kasparov after a six-game match, two wins for IBM, one for the champion, and three draws. The contest drew worldwide coverage, and behind it sat serious computer science pushing forward how machines handle complex calculations.

In February 2011, IBM's Watson competed on Jeopardy! against the show's two greatest all-time champions. Watson ran software called DeepQA, developed by IBM Research. Winning Jeopardy! was the grand challenge, but the broader goal was a new generation of technology that could find answers in unstructured data more effectively than standard search, exactly the terrain SharpeMind was built to explore for financial markets.

Progress

Modulus is no longer engaged with IBM Watson. In the years since, we have developed our own natural-language-processing systems for financial markets and beyond, carrying the lessons of SharpeMind into technology we own and control end to end.

That work lives on today in the Modulus Sentiment Analysis Engine, a deep-learning system that extracts meaning and emotion from text, and in Modulus AI, our broader suite of AI capabilities for finance.

Technology behind SharpeMind

What made SharpeMind different

A first-of-its-kind financial application on IBM Watson, combining large-scale unstructured-text analysis with a patented technique for producing time-series forecasts.