The AI Pioneers in Investment Management report from CFA Institute explores global best practices in the application of artificial intelligence (AI) and big data technology in the investment process.
Since its launch last year, the report has inspired various compelling inquiries from readers and event participants that are worth addressing. Below are some of the frequently asked questions (FAQs) along with my responses. Please continue to send us your queries and comments by email or in the comments section below, and I will be sure to share and answer those that could benefit the wider audience.
Thio Boon Kiat, CEO of UOB Asset Management, Singapore, asked:
1. How can an investment firm transform itself into a technology-driven organization and achieve full buy-in from investment professionals?
We believe an organization’s competencies in investments and technology are complementary rather than competing.
At a high level, we believe the future of finance will involve collaboration of finance and technology. In one of our first explorations into fintech in the summer of 2016 (“FinTech and the Future of Financial Services” — first published in Hong Kong Economic Daily in May 2016 and later included in FinTech 2017: China, Asia and Beyond — we hypothesized that powerful fintech will be the result of collaboration between powerful fin(ancial institutions) and powerful tech(nology companies). We believe the old model where technology plays an auxiliary role to finance has failed and the successful models of the future will have equal contributions from both sides.
More specifically, in the context of applying AI and big data technologies, we believe the successful model of collaboration in investments, an area long dominated by investment professionals, i.e., human intelligence (HI), will be AI + HI. The concept was first brought to us by a guest speaker at our AI and the Future of Financial Services Forum, an event we organized in Beijing in December 2017, and is very consistent with our general philosophy of Fin + Tech.
Instead of worrying that AI will take over the jobs of investment managers, we believe the most effective approach is to embrace technology as AI and HI have different strengths and weaknesses. This is a theme discussed repeatedly in our FinTech 2018: The Asia Pacific Edition report and later elaborated on further in the Investment Professional of the Future report, where we first discussed the T-shaped teams concept.
T-shaped teams is how the above theme exemplifies itself from an operational and organizational angle. We discussed the concept more thoroughly in AI Pioneers in Investment Management, with the key being that future investment teams will have an embedded technology function in addition to the investment function that we have always had. More importantly, we suggested adding a small T to the T-shaped teams to help the two main functions collaborate better. We called it the innovation function.
2. How can we measure the contribution of AI and big data techniques?
This is an important question for decision-makers although there is no easy answer. A key challenge is that we are looking at something very new where few teams have a long enough track record. Another is isolating the impact of AI and big data techniques when they are part of an investment process.
At the current stage, AI and big data applications tend to help more in many steps along the entire process, as illustrated in the case studies in our report, rather than as a complete solution. We picked the cases included in our report based on the criteria that the AI and big data applications discussed are all actively used in the investment process, or “live in production,” as our friends in technology would like to say, and the processes are responsible for managing a significant sum of assets. We trust that managers will pull an investment tool from the process if it fails to add value and we have seen such cases taking place at firms we spoke with.
That said, we are more than happy to speak with any team who can demonstrate the precise impact of AI and big data applications in their process. Please feel free to reach out to us.
CJO Verzijl, quantitative strategist, ABN Amro, Amsterdam, asked:
3. Are machine learning (ML) techniques augmenting structural models — the things we already know about the world — or supplanting them through purely data-driven approaches?
This is similar in essence to the question we get asked a lot by fundamental managers and analysts in the context of a particular product: Do AI and big data add alpha?
More broadly and maybe more interestingly, one may also be curious from the industry overall perspective: Are investors as a whole getting better return now than before AI and big data techniques were introduced?
The ultimate question, of course, will go beyond the investment industry: Do AI and big data create wealth, or are they merely replacing other creators of wealth?
These questions are so important that we would like to set up a framework to think about it. The framework goes:
- Total wealth creation is driven by labor and technology/capital input.
- Total investment (market) return is driven by investment demand and supply.
- Each fund’s excess return (alpha) is driven by its competitive advantage in assessing and analyzing public information.
We’ll start from the one most important to investment managers: Do AI and big data add alpha? Judging from the case studies in the report, our answer is absolutely yes. AI and big data techniques have given these investment teams an advantage in obtaining and processing data while not taking away any of their existing tools.
So to the extent that these techniques are effective, which we hope the case studies have demonstrated, then they would add to the product’s alpha.
The next question could be important to end investors and investment industry regulators who look after the end investor’s interest: Do AI and big data techniques increase (net) market return overall? Using the framework mentioned above, it seems obvious that the answer is no. As a matter of fact, no investment technique to date is known to increase total market return, so the seemingly pointed question is not actually properly framed.
The last question is probably what end investors and investment industry regulators really have in mind: Do AI and big data techniques add wealth? Using the framework above, the answer is yes, if AI and big data techniques improve productivity more than they replace labor input.
This may need to be assessed on a case-by-case basis. Judging from the case studies included in the report, AI and big data techniques will at most replace some junior analysts and traders but could significantly improve overall productivity. So we stand by our answer.
Are there cases where AI and big data may replace so many people that total wealth creation may decrease as a result? That is certainly something for business and political decision makers to carefully consider but clearly outside of the scope of our report.
Lutz Morjan, senior client portfolio manager, EMEA, Franklin Templeton Multi-Asset Solutions, Frankfurt, asked:
4. How do managers that use AI and big data techniques explain the value add to their clients?
Given that AI and big data techniques tend to be used in support of an existing investment process rather than to replace it, the explanation can be structured similarly. That is, you can present your overall process exactly as before but add in explanations about where and how AI and big data are adding value.
Specific explanations will, of course, also depend on the sophistication of the investors you speak to. For institutional and sophisticated retail investors, we think you can simply structure the (more elaborate) explanations in the format of our case studies: discuss the enhancement to the investment process, specific AI and big data techniques used to make it happen, and organizational support/additional skillsets that you obtained in making it happen.
For those unsure about communicating how machine learning works, help is on the way. Many AI scientists appreciate your pain and have started working on developing ML solutions with more transparency built in from the get-go. Before then, investors will hopefully be happy with the following: It is an approach scientists use to generate output from a set of chosen input, not unlike statistics but without the restriction of being linear and without having to specifically spell out an equation or estimate all the parameters.
How do you think your investors will like the changes? Let us know by leaving a note in the comments section.
If you liked this post, don’t forget to subscribe to the Enterprising Investor.
All posts are the opinion of the author. As such, they should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute or the author’s employer.
Image credit: ©Getty Images/nevarpp
Professional Learning for CFA Institute Members
CFA Institute members are empowered to self-determine and self-report professional learning (PL) credits earned, including content on Enterprising Investor. Members can record credits easily using their online PL tracker.