Mark Robertson, senior portfolio manager multi-asset at NN Investment Partners, said sentiment analysis can sometimes outweigh fundamentals in the Dutch fund manager’s investment scorecard and allow it to spot opportunities.

Valentijn van Nieuwenhuijzen, chief investment officer (pending regulatory approval) at NN Investment Partners said at a media briefing that the asset manager started using sentiment analysis four years ago after finding the best research provider and best data to use to complement the firm’s other data sources. As a result NN IP partnered with MarketPsych, an independent consultancy that develops financial applications from behavioral economics, and found that sentiment analysis had predictive value.

“It was important to build an engine for our investment tools which reflects how the economic system works,” said van Nieuwenhuijzen. “The modern economy is a complex adaptive system in which all the components – such as the real economy, politics, financial markets – are continually influencing one another and you can only sense the direction of markets over time.”

In order to understand this behaviour, NN IP studies how sentiment drives markets. Robertson added that studying underlying fundamentals and human behaviour, especially investor behaviour, is essential to understand what is driving markets.

In 2011 the Thomson Reuters MarketPsych Indices were launched for countries, currencies, commodities and industries through analyzing news and social media in real-time.

MarketPsych built advanced statistical language processing software to quantify sentiments – such as fear, joy, uncertainty, trust, and urgency – which are known to predict risk taking. The system’s logic takes into account all the specific mentions in various news and social media posts and aggregates that data into the specific indices. In 2015 Thomson Reuters extended the analytics offered through its MarketPsych Indices to include individual companies.

NN IP uses the Thomson Reuters MarketPsych Indices sentiment data in different ways in its investment process. For example, analysis of emotions including conflict, violence and government trust are grouped together to give a measure of political risk. The asset manager then incorporates these sentiment factors into investment scorecards of about 30 independent signals.

“Next to flow and price dynamics, the analysis of observed emotions, or sentiment, of the market is very useful,” said Robertson. “Digitalisation and the emergence of social media, self-teaching algorithms and the ability to process large amounts of data, practically in real time, have created an entirely new way to measure sentiment.”

For example, before the election of President Trump NN IP saw that political sentiment was overwhelmingly negative.

However, sentiment was so low that there was the potential for a rebound once that uncertainty was removed, no matter the actual outcome of the election. Therefore, NN IP added to its risk positions and benefited from the subsequent improvement in sentiment.

“It also helped us to get an early sense in the beginning of 2017 that political risks started to fade in the first quarter of the year,” said Robertson. “The latter helped us to stay overweight on equities, as our research shows that changes in emotions can play a predictive role in changes in the direction of equity markets.”

Robertson continued that sentiment on emerging market equities has been consistently positive. As a result, NN IP has benefitted from being overweight emerging market equities this year.

“Sometimes, in the investment scorecard sentiment can have a higher weight than fundamentals,” added Robertson.

van Nieuwenhuijzen also gave the example of global bond markets where there have been warnings of a bear market for three or four years. “We saw that sentiment became more negative last summer and this protected us from going short on bonds too early,” he added.

However van Nieuwenhuijzen continued that NN IP also tries to forecast very rare erratic events which cannot be captured by data alone.

Robertson explained that that fund manager is running a project based on Superforecasting: The Art and Science of Prediction, a book published in 2015 by Wharton professor Philip Tetlock and co-author  Dan Gardner. Tetlock’s research has found that some people are far better at forecasting the near-term future than the average, and these skills can be taught and improved with practise.

“We look at the structure of our team and how they interact and share ideas,”Robertson said. “We ask questions about future events and ask them to assign probabilities, and then trace how that changes over time.”

van Nieuwenhuijzen said: “We believe in the combination of man and machine.”

NN Investment Partners, formerly known as ING Investment Management, is part of NN Group a publicly traded company listed on Euronext Amsterdam. NN IP managed nearly €194bn ($208bn ) in assets at the end of March 2017 for institutional and retail investors.