UBS embraces ‘narrative alpha’ for new form of sentiment strategy

NLP engine traces how stories spread, instead of counting words

Data without context is commonly described as noise, and in systematic investing it can lead to mistakes, misunderstandings and poor decision-making.

Yet, for an army of natural language processing (NLP) robots, which scour news articles for key terms indicating positive or negative sentiment on stocks, context has largely been ignored.

This may be changing, as new models seek to measure the narratives driving behaviour across financial markets.

“It’s not about the individual words. It’s about the