Here is a scenario: You are the portfolio manager of an active equity fund. You wake up one morning to the news that a labor strike is delaying flights at an airline in your portfolio. Your job is to assess the impact of this event on the firm’s fundamental value and make a portfolio decision. You can hedge, sell, buy, or do nothing.
Evaluating events like these, at the very least, requires you to (1) gauge the severity of the loss, (2) assess its permanence, and (3) compare the updated fundamental value of shares to the current market price.
Apart from an intimate understanding of the airline’s business, you need a robust working model of the company and some “processing” time to properly recalculate the firm’s new valuation.
While market prices change instantaneously in reaction to the news, the search for the fair price continues well after the newsmaking event, as the market participants scramble to place their bets . . .
Now imagine the same exact strike scenario but with a slight twist.
Early Warning: Several hours before the news hits the market, you receive an alert of a possible labor strike. This alert pops up on your computer’s investment dashboard, which houses an array of analytical tools to help you make informed, fundamentals-based investment decisions on the go.
Intelligent Dashboard: Your dashboard is powered by artificial intelligence (AI) — a set of self-learning, self-correcting algorithms. They scour their data feeds, which range from satellite imagery to social media networks and economic, political, and financial databases, to bring you relevant information about your portfolio. They distinguish signal from noise and update your portfolio’s intrinsic value and risk metrics in real time as new information pours in.
From your dashboard, you can see the airline’s routes affected by the strike, the anticipated daily dollar loss stemming from it, and the likely effect on share price should the strike continue one more day, three more days, or another week. You can drill down and see where the data is coming from and how it’s analyzed. At any time, you can change the model’s inputs to reflect your own views or select a different method of analysis.
Action Menu: At the bottom of your dashboard, you have a menu of your available investment actions, each supplemented with its potential impact on your portfolio return and risk.
Your job? To understand what’s at stake for those involved in the strike and to recalibrate its probability to last past a certain date — say one week. You have a few questions to answer: What are the labor union’s demands? What are the chances they will be met? What can the union realistically settle for?
The dashboard described above does not exist. At least not to my knowledge, but we may not be too far from some version of it becoming reality. This realization is not easy for me. I am a finance professional with a decade of fundamental investing experience. However, after studying data science for some time, I am convinced that it will bring profound change to our profession.
While the tenets of rational, fundamental investing are likely to remain relevant, the future of investment research will be shaped by data science and by the dynamic models that don’t yet exist. I see this as a natural step in the evolution of finance. A step that will require solid understanding of data science in addition to investing skills.
Why will data science become a permanent feature of the investment landscape? Because it outperforms humans in at least three areas:
1. Unbiased Analytical Thinking: Using machines to make investment decisions minimizes human error and cognitive biases. Investment professionals may use a number of techniques to recognize and minimize them, but we can’t eliminate them. Many of these are “hardwired” into our brains as established neural pathways.
In contrast to humans, AI-based algorithms have no egos. They are agile, they can quickly absorb new information and make course corrections. Any data can be used to generate insight. AI can learn and evolve from changes in its environment. Unlike static quantitative models with limited shelf lives, AI-based systems are “alive.”
2. Processing Power: When it comes to information processing, humans are no match for machines. They can out-analyze us. Think of IBM’s Deep Blue supercomputer defeating grandmaster Garry Kasparov at chess in 1997, or Google’s AlphaGo AI outplaying the world’s top-ranked Go player in 2017.
And this edge goes beyond analytical thinking. Machines also have us beat in the more subtle associative thinking, a skill long thought to be exclusive to humans. In 2011, IBM’s Watson defeated the top human Jeopardy! champions by a wide margin. For me, this was the moment that redefined my view of analytical thought, artificial or not.
In their current form, machines like Siri and Alexa already understand human speech and can learn, process, and analyze the entire history of a human-produced domain knowledge. If this trend continues, machines will become capable of intelligent investment and resource-allocation decisions with minimal human input.
3. Software Economics: From a purely economic point of view, the value of an employee is a function of his/her contribution to the bottom line. Software that can replicate an employee costs a fraction of what firms may spend on their new hires. This threat is especially pronounced for college graduates whose starting jobs consists of collecting, organizing, and analyzing analytical data.
There are five steps in the investment decision-making process: data collection, data processing, investment analysis, investment decision making, and performance evaluation. Of these, three can be performed by pieces of code. In fact, given a data source, any task that can be broken down into its logical steps can be turned into code and automated. Hedge funds like Bridgewater already started to use AI to augment their investment decision-making processes.
The third step — investment analysis — still requires human input to evaluate such factors as strategic imperatives, the competitive landscape, government policy, and board independence. Generally, any key investment data that can’t be collected and aggregated into databases due to, say, legal/logistical restrictions, will require human input. Examples include in-person interviews with the portfolio company management where non-verbal clues come into play.
Artificial Intelligence (AI)-Based Investment Process
Source: Oqulent LLC
In all likelihood, future portfolio managers will have to be fluent in methods of extensive data collection and data analysis. They will also need to know how to transform their investment ideas into machine-readable code. To the degree that portfolio company’s decision making involves people, portfolio management will also require a good understanding of human behavior.
A Brave New World?
Instead of reading financial statements for nuggets of financial insights, future investment professionals will derive their alpha from analyzing and predicting the impact of human behavior as an overlay to an already-established array of highly analytical, flexible, AI-based investment frameworks.
In that context, I can’t rule out the black-box approach to investing. These are cases where AI-based systems could make investment recommendations the rationales for which are beyond our ability to understand. While we will have some fundamentals-based safeguards in place against extreme swings, our investment decisions would be driven by the AI’s predictive power rather than our grasp of its decision-making path. In this world, investment professionals will act more as guardians of investor interests, defining investment goals, optimizing decision-making algorithms, and training AI to do most of the analytical heavy lifting.
Judging by the pace of technological evolution, this world may come sooner than later. The good news is that data science will need finance pros with domain experience to write the next chapter in the history of the investment profession.
The question is: Are you prepared?
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/spainter_vfx
Umed Saidov, CFA, is the founder of Oqulent, LLC. He closely follows the developments in artificial intelligence and crypto-assets to help investors understand the emerging risks/opportunities from these areas. Prior to launching Oqulent, Saidov spent 10-plus years with International Finance Corporation (IFC) and EBRD, where he led and executed a number of high-profile infrastructure and renewable energy investments around the globe. He holds an MBA from INSEAD and has a bachelor’s degree in general management from French-Russian Institute of Business Administration.