Today (September 3, 2024) I have attended an investment conference at which the integration of artificial intelligence (AI) in the portfolio management processes was one of the topics discussed by a fund manager.
After seeing the launch of generative AI within the wealth management business of some big banks in the U.S., the review of today’s presentation might be a good reality check with regard to the integration of AI in portfolio management processes. The manager runs three portfolios with the help of artificial intelligence. While two of these portfolios are totally relying on the decisions from AI, the AI is only a support tool which pre-selects possible candidates for the portfolio for the third fund, where the fund manager makes the final decision for the portfolio holdings.
To be honest, the AI driven investment process of the first two funds looked to me like any other quant management approach, with the exception that the artificial intelligence model is able to extract data from various sources by itself and can process the data as needed. From my point of view this means that the AI is, in this case, a very advanced quantitative model. This is good, since the implementation of this kind of technology helps to develop the underlying models even further, which may lead to superior portfolios in the future. That said, one needs to bear in mind that these models are able to process more data in a systematic way than any human can. This thesis was also backed by case study from the fund manager in which he showcased single stocks selected by the AI, which would have normally been overlooked by the portfolio management team, since they are from more exotic sectors or locations. With regard to this, the portfolio manager clearly pointed out that the AI has some flaws when it comes to the analysis of balance sheets from banks, as they are very complex, as well as the analysis of balance sheets from companies based in emerging markets with different regulatory standards than developed economies.
As mentioned above, the third fund has a human interface, since the AI model is “only” analysing the data to create a pre-selection for the fund manager who then buys those stocks for which he forecasts the best returns based on additional qualitative research. As mentioned in my “Monday Morning Memo: Will AI be a Game Changer in Portfolio Management?”, I assume that this is the use case for which AI can add the most value for portfolio managers and analysts since no human being would be able to extract and process the amount of data needed at the same speed as artificial intelligence. In addition to this, AI might be able to generate new investment ideas since the model can easily expand to new markets or market segments where the respective data is available.
Within this specific case it might be interesting to see how one of the AI-driven funds will perform compared to the product with the human interface, since these two products have a similar investment objective.
More generally, we can see that artificial intelligence has become a stable within the fund management processes of a lot of asset managers. That said, most of the AI models I have seen so far are more like enhanced quant models, which have much more firepower than the quant models from the past. I do like the fact that asset managers are adopting this new technology quite fast, since this will enable fund managers and analysts to more qualitative work, which might be in the best interest for their investors. Nevertheless, I do think that AI adoption is still in its childhood, and we will see much wider use cases for this new technology (including generative AI) in the portfolio management processes in the future.
This article is for information purposes only and does not constitute any investment advice.
The views expressed are the views of the author, not necessarily those of LSEG Lipper or LSEG.
