Rather than just investing in artificial intelligence (AI) stocks, investment managers are trying out AI techniques as a way to achieve higher portfolio returns. Can it be done or is this just a fad?
Back in 2021, the Chartered Financial Analyst (CFA) Institute Institute, a global association of investment professionals, issued a report1 stating, “AI (artificial intelligence) is a must-have… With growing evidence of AI’s effectiveness in helping investment professionals in their daily work, the focus has now shifted from building conviction about AI and big data to successful adoption of the technologies.”
Indeed, over the past decade, asset managers have become more open to the potential of AI tools and techniques to lift investment performance. This is because of the three key advantages that only AI can provide:
1. Coverage
According to the World Federation of Exchanges, there are over 31,000 stocks listed in Asia Pacific (representing more than half of all listed stocks), followed by Europe, Middle East, and Africa (EMEA) and then the Americas, adding up to a total universe of over 53,600 stocks2.
It is impossible for any global or regional firm to hire enough research analysts to adequately cover all, or even a majority, of the stocks pertinent to their portfolios, especially when diversification is key. AI, on the other hand, can be used to analyse billions of data points related to a company’s fundamentals, past economic trends and decades-long market movements.
Therefore AI is not limited by the number of analysts or working hours in a day, but rather whether there is sufficient data available for analysis. The more data, the more effective the AI model. On the other hand, AI works less well for markets or sectors where there is insufficient data, for example, some frontier markets or penny stocks where financial reporting is not as onerous.
2. No biases
AI is based on machine learning, that is, a method of making predictions by ingesting vast amounts of data to detect patterns. The machine tests the data against changes in key variables called “features” to see how these affect outcomes going back in history. In the case of investment management, these features can be market prices, trading volumes, or company valuations, but their importance is not pre-determined. The features are then used to forecast a stock’s potential upside or downside, given the current investment environment.
This is different from how human analysts operate. In general, all analysts have preconceived ideas about how to achieve outperformance, including using style, geographic or technical investing strategies. This is based on their domain knowledge and experience, but as a result, analysts can overlook unusual relationships or find it hard to outperform during unfamiliar market events.
In the same way, quant strategies are often based on statistical models that reflect analyst concepts and theories. These attempt to find a set of variables that best fit past data, and project these forward into the future. This means they work in the opposite direction to AI strategies, which are designed to learn from what is happening in the real world and to adjust the key variables accordingly. The weakness with quant models is that variables that have worked in the past may not be the same ones that work in the future.
3. Scalability
AI strategies are also more easily transportable across asset classes. For example, unlike the barriers that traditionally separate equity analysts from bond analysts, the same AI methodology can be applied to both asset classes, provided that the necessary data is available.
This is because AI is driven by a bottom-up approach focused on the manipulation and transformation of empirical data. In contrast, traditional methods of bond and equity selection are based on a deep and specialist understanding of the respective market drivers.
As a result, AI strategies can be used to achieve a number of different investment outcomes, regardless of asset class. They are appropriate for security selection, but can also be used for asset allocation. In addition, AI can be used to improve risk management by identifying not only positive patterns, e.g. related securities that have the potential to outperform, but also negative patterns, e.g. related securities that have the potential to increase a portfolio’s concentration risk.
How to successfully adopt AI?
While it is clear that AI can bring many unique advantages to the investment management process, it is also clear that the enhancement of returns cannot be left to AI alone. Human analysts continue to play a crucial role because there will always be good companies that do not fit into a pattern or trend.
There is also no guarantee that the required data is readily available, and this data is, by definition, a record of what has happened rather than what could unfold in the future. Finally, risk controls will continue to be set by human managers because decisions about the acceptable amount of volatility tends to be taken at a mandate or firm level.
So in effect, the adoption of AI can be more precisely described as the integration of AI techniques into existing investment management processes. This must be done in a way that leverages existing manager expertise and yet allows for AI insights to be accommodated. As the CFA Institute report suggests, this integration can be even harder than the actual development of financial AI models!
Are you comfortable investing in funds managed using AI? Tell us how you feel about AI-driven investing by completing this short survey. |
What is AI-Augmentation@UOBAM?
It is with this understanding that UOBAM started its AI journey in 2018. The firm has long been committed to leveraging new technologies that can be shown to deliver superior outcomes to investors. Recognising the potential for AI to enhance investment performance, UOBAM began trials of its AI models for a select number of funds.
However, considerable effort was spent on merging the investments data science discipline with the firm’s local and regional investment capabilities to ensure a streamlined three-stage stock selection process.
Stage 1
Ideas Generation
To kick start the process, data relating to the relevant investment universe is ingested from a variety of data providers to generate a short list of stocks idea. Most asset managers stop at this stage of AI application.
Stage 2
Security Selection
On receipt of this shortlist, analysts can contribute to the shortlist and will rate all stocks accordingly. This rating will influence the final rated shortlist of stocks.
Stage 3
Portfolio Optimisation
In the final stage, analysts will further input any relevant risk controls such as exposure limits. The model then assigns weights to each stock in the portfolio which is assessed again by the manager before being implemented.
This highly structured approach has paid off with a few of UOBAM’s AI-augmented funds demonstrating solid performance. To date, the United Asia Fund is rated 4-stars by Morningstar3 and the 5 star-rated United Greater China fund recently received a Refinitiv Lipper award for the Best Performing Fund in the Greater China, 3-year category4.
It would appear therefore that AI can make a discernable difference to fund performance. However, we would caution that this difference comes not just from high quality AI models, but importantly, the way the models are incorporated into the wider investment management process.
1CFA Institute: T-shaped Teams: Organising to adopt AI and big data at investment firms, 2021
2World Federation of Exchanges, May 2022
3Morningstar, as of September 2023
4Refinitiv Lipper Fund Awards 2023 Winner Singapore, United Greater China Fund A SGD Acc | Please refer to uobam.com.sg/awards for the latest list of UOBAM awards
If you are interested in investment opportunities related to the theme covered in this article, here are two UOB Asset Management Funds to consider:
|
This document is for general information only. It does not constitute an offer or solicitation to deal in units in the Fund (“Units”) or investment advice or recommendation and was prepared without regard to the specific objectives, financial situation or needs of any particular person who may receive it. The information is based on certain assumptions, information, and conditions available as at the date of this document and may be subject to change at any time without notice. No representation or promise as to the performance of the Fund or the return on your investment is made. Past performance of the Fund or UOB Asset Management Ltd (“UOBAM”) and any past performance, prediction, projection or forecast of the economic trends or securities market are not necessarily indicative of the future or likely performance of the Fund or UOBAM. The value of Units and the income from them, if any, may fall as well as rise, and is likely to have high volatility due to the investment policies and/or portfolio management techniques employed by the Fund. Investments in Units involve risks, including the possible loss of the principal amount invested, and are not obligations of, deposits in, or guaranteed or insured by United Overseas Bank Limited (“UOB”), UOBAM, or any of their subsidiary, associate, or affiliate (“UOB Group”) or distributors of the Fund. The Fund may use or invest in financial derivative instruments, and you should be aware of the risks associated with investments in financial derivative instruments which are described in the Fund’s prospectus. The UOB Group may have interests in the Units and may also perform or seek to perform brokering and other investment or securities-related services for the Fund. Investors should read the Fund’s prospectus, which is available and may be obtained from UOBAM or any of its appointed agents or distributors, before investing. You may wish to seek advice from a financial adviser before making a commitment to invest in any Units, and in the event that you choose not to do so, you should consider carefully whether the Fund is suitable for you. Applications for Units must be made on the application forms accompanying the Fund’s prospectus.
This advertisement has not been reviewed by the Monetary Authority of Singapore.
UOB Asset Management Ltd. Company Reg. No. 198600120Z