Monitoring Portfolio News Using AI

Harry MamayskyAdvisor Perspectives welcomes guest contributions. The views presented here do not necessarily represent those of Advisor Perspectives.

One of the enduring challenges of portfolio management is the inability to follow all news flow relevant to portfolio positions. AI and cloud-based workflows are helping us overcome this problem.

In my years trading on Wall Street, I was always bothered by the fact that, even with a small number of trades and a team of analysts, we still were unable to follow all news flow relevant to our portfolio positions. This is despite the fact that all of us spent the majority of each working day (and weekends) reading news and analyst reports. In fact, this experience convinced me back in 2014 (or so) that the future of investing would lie in using natural language processing techniques (NLP) to interpret the voluminous amounts of text data with which we are confronted.

I’ve now spent the last 10 years working on applying NLP tools to the study of financial markets (you can check out this work here) and on teaching this material to our M.S. and Ph.D. students at Columbia Business School. More recently, I spent the last four years setting up QuantStreet, an asset allocation and wealth management firm.

Since the release of ChatGPT on November 30, 2022, the pace of technological innovation in AI has grown exponentially. Today’s AI models can ingest and intelligently reason about hundreds of pages of information without any additional training. This creates an opportunity to develop what truly are superhuman news summarization engines: Feed several hundred pages of news into an AI model, provide it with an appropriate set of questions — for example, about how the news impacts portfolio positions — and let the model generate in a few minutes output that would take dozens of man-hours, if not more.