Artificial Intelligence (AI) is a hot topic. The technology is now capable of assisting people with a rapidly growing number of tasks, both professionally and personally. This ranges from voice assistants on our smartphones and robotic lawnmowers to algorithms that create personalized advertisements for online shopping. AI is also becoming increasingly significant in the field of investment portfolios.
This article, originally written by Stefan Klauser, was first published as a guest piece in HZ Banking in German.
AI has been a scientific discipline for around sixty years. It encompasses a variety of theories and techniques, including mathematical logic, statistics, probabilities, computational neurobiology, and computer science. The fundamental goal of AI is to mimic human cognitive abilities. AI can either be based on predefined algorithms or developed through machine learning. In the latter case, an algorithm learns from data and repetition to independently master a task. Unlike traditional algorithms, where a specific solution path is predefined, AI algorithms learn to identify structures within data on their own.
Computer-based investment strategies have been used for some time, for example, by digital asset managers known as robo-advisors. These manage customer funds using rule-based models, which can be seen as a precursor to AI in the portfolio sector. Professional investors, such as hedge funds, are increasingly relying on AI to make investment decisions. Powerful computers analyze financial data to identify patterns and trends in the markets. They sift through large data sets, from corporate numbers and stock price trends to management statements, news, and general market sentiment, to identify promising investment opportunities. AI considers not only historical information but also key terms that provide insights into a company's current situation or the sentiment of its leadership.
AI has the potential to revolutionize traditional portfolio management. It can analyze large amounts of data, identify patterns, and assist with diversification, risk assessment, and investment selection. The greatest strength of AI lies in its ability to analyze companies or economic trends faster, more deeply, and more comprehensively, enabling decisions to be made on a logical basis. Another advantage is AI's ability to exclude human emotions, which helps to act rationally and avoid mistakes, especially during crises. A possible outcome of AI applications in portfolio management could be a list that rebalances the weights of securities in an existing portfolio based on customer risk appetite, ESG factors, or other aspects. The adjustments to the portfolio are then implemented through corresponding purchases and sales, known as “rebalancing” a portfolio in professional circles.
To illustrate how AI manages a portfolio, we will examine a simple portfolio that represents key segments of the Swiss market and stands for innovation. We will call this portfolio "Swiss Grand Eight." It includes the stocks of ABB, Georg Fischer, Sika, Roche, Lonza, Nestlé, Novartis, and UBS.
Without AI
First, we will look at how this portfolio would have performed over a certain period without AI. To include the effects of the pandemic and other global events, we choose a period of five years, from January 1, 2019, to the end of December 2023, for back-testing the portfolio. For simplicity, we assume the portfolio consists of equal shares of the mentioned stocks, i.e., 12.5% per stock of the total value. Such a portfolio (blue curve) compared to the Swiss Market Index (SMI - orange curve) looks as follows:
During the selected period, the SMI achieved an annualized performance of 5.73% with a volatility of 15.49%. Volatility measures the extent of fluctuations in the value of investments over a certain period. The "Swiss Grand Eight" portfolio generated a much better performance of 13.28%, though its volatility was also higher at 17.97%.
Now, let's see how AI would have managed the exact same portfolio over the same period. We use the AI platform developed by Aisot Technologies, which is currently used by institutional and professional investors. The AI platform is given the ability to rebalance the portfolio's weights weekly, based on predictions for the individual stocks, ranging from zero to one hundred percent. Additionally, depending on market conditions, the AI can incorporate up to 35% cash into the portfolio. Based on historical market data, macroeconomic indicators, and alternative data sources like social media, aisot's AI platform aims to optimize the portfolio's weights weekly to minimize risks, such as the impact of price drops in specific stocks, while optimizing performance.
This results in the following picture:
From January 1, 2019, to December 31, 2023, the AI-optimized portfolio achieved an annualized performance of 15.35%, which is an increase of about 15% compared to the non-optimized version. Almost more important than performance is the comparison of volatility. The AI-optimized portfolio has an annualized volatility of 15.40%, which is more or less the same as the SMI (15.49%) but significantly lower than the non-optimized portfolio (17.97%).
When looking at the historical distribution and weighting of the portfolio's investments, we see how the AI attempts to counteract times of high volatility by increasing the cash share in the portfolio, favoring stocks that were exposed to greater risks at those times. This is clearly visible at the beginning of the pandemic in 2020 and in 2022, marked by significant market fluctuations.
While the example of Swiss Grand Eight might suggest that the role of humans in portfolio management will become less important in the future, it is essential to emphasize that humans will continue to play the most crucial role. This applies both to programming the AI and to the final decision-making and implementation. Humans and machines can complement each other optimally in allocation. AI already significantly relieves humans by assisting especially with data evaluation and preparation. Given the multitude of factors that can influence the value of stocks and other investments, AI is ideally suited to support portfolio managers in the process of portfolio composition, validation, and personalization.