Portfolio managers are having to take into account an ever-growing deluge of information from not only analyst reports, but also numerous news sources, industry blogs and social media, to make forecasts and determine the impact on stock prices.

Working with Nomura Research Institute, NAM spent months on a proof-of-concept to see whether natural language processing utilising AI could increase the accuracy of portfolio managers’ investment decision-making.

The firms used AI technology to analyse all the information a portfolio manager would consume and score them into two groups - either positive (indicating that company performance or corporate value is likely to rise) or negative (is unlikely to rise).

First up, a natural language analysis on analyst reports was done to highlight the shifts of investment decisions; for example, a shift from neutral to overweight or from neutral to underweight. The language patterns for 'positive' and 'negative' were then identified and used as training data for AI. Finally, the AI calculated the similarities between the training data and the targeted materials, scoring whether each piece of information was 'positive' or 'negative'.

NAM says that the PoC highlighted that analysis of analyst reports using AI enabled the quantitative assessment of information which portfolio managers usually see as qualitative. In addition, even text information from news websites and blogs could be quantitatively scored and used to boost the ability of portfolio managers to make investment decisions.

The partners now plan to work with clients to create more PoCs and to explore cases where AI can help portfolio managers in their work.