The investment industry is no stranger to advanced computing; quantitative trading programs and algorithms have long been used to maximize profits. However, the recent breakthroughs in AI technology pave the way to entirely new possibilities. Advances in big data, deep learning algorithms, and GPU processing add up to a significant leap in AI development and its capability to improve portfolio management. Now, financial firms have started to utilize these cutting-edge solutions.
Promising Preliminary Results
The nature of AI allows it to take into account millions of data points to create complex computations in mere seconds. This calculated approach lends itself perfectly to the financial sector, as emotional and biased decisions can be detrimental. A preliminary paper from two Korean academics, Hyungjin Ko and Jawwook Lee, has shown that even the general ChatGPT model outperformed a random selection of stocks on a risk-adjusted basis and diversification. ChatGPT’s picked portfolio of four assets showed an expected return of -13.3% compared to -23.3% for the random selection.
The experiment was run in 2022, during which stocks and bonds plummeted, so the negative results should not be surprising. While the researchers concluded that ChatGPT is not yet to be hailed as a new digital prophet, even in its current stage, it makes for a useful co-pilot in investing. Established financial firms such as Morgan Stanley are already embracing the new wave of AI chatbots by integrating their functionality into their daily activities. So what exactly can AI help with when it comes to objective asset management?
The Capabilities of AI in Portfolio Management
Fundamental Analysis: AI can be utilized to perform textual analysis of economic reports, annual reports, and other relevant information. Additional uses also include the identification of hidden correlations between various asset classes. Stock selections can then be picked on whether they will outperform or underperform regarding those correlations.
Portfolio Optimization: AI can predict the expected returns as well as variances and covariances to determine optimal asset weights. Specifically, genetic algorithms are capable of computing complex optimization tasks with pre-set constraints, like a minimum holding threshold, or a restricted number of assets.
Risk Management: AI is capable of forecasting risk variables like bankruptcy probability, incorporating qualitative data, such as social media or news reports, and validating the existing risk models to ensure that they fall within the accepted tolerance levels of a client.
Trading Activities: AI has been used by algorithmic traders for its utility in technical analysis predictions for years. Additionally, AI can be employed to assist in calculating transaction costs and making large trades in one swoop instead of having to break them into smaller ones. This helps to achieve the lowest transaction costs.
As AI’s capabilities heavily rely on the quality and amount of data it is trained on, a specialized model intended for asset management will likely perform much differently than shown in the ChatGPT experiment. Last year, over $91 billion was invested in AI companies, signifying the gold rush of AI startups is still in progress. One can only ponder what leaps of progress are being made behind closed doors. It makes sense that financial firms would not publicly share access to new AI software that performs excellently to maintain an edge over the competition.
The Challenge of Human Bias
When it comes to making investment decisions, level-headedness is crucial. Human beings are susceptible to emotions and cognitive biases, which can cloud judgment and lead to suboptimal outcomes. To some degree, confirmation bias operates in each person. Individuals tend to seek out information that is in line with their preconceived notions while disregarding evidence that points to the contrary. The BeFi Barometer 2021 study, conducted by Cerulli Associates, found that confirmation and recency biases were the top ones affecting the investment decisions of clients.
Portfolio managers may be trained to minimize their preconceived notions and emotional sway, but they are not immune to it. A bullish outlook on a specific stock could lead to selective market data and news interpretation. Fearful decisions in a market downturn can lead to panic selling at undervalued prices. Greed may influence irrational exuberance. On the other hand, AI’s algorithms are built around cold-hearted, rational precision. Therefore, having real-time, in-depth reports accessible can help portfolio managers maintain a streamlined way of operating to make objective investment choices with minimal bias.
Improved Efficiency and Widened Oversight
Over the past few years, computing power has advanced exponentially. In the field of neural learning, it is currently dominated by Nvidia. The company’s A100 chip has become the workhorse of the AI industry and is ideally suited to train artificial intelligence models, generate data, and make predictions. By being able to hastily process terabytes of data, it enables analysis far beyond previous capabilities. Combined with the vast availability of data sets, AI-powered algorithms have proven to achieve feats that make them invaluable tools for investors.
Consider the process of analyzing financial statements. Sifting through numerous reports, extracting data, and making calculations involve meticulous labour. When done manually, it is also prone to errors. Automated AI software can perform these tasks efficiently, include complex calculations, and provide consistent results with astounding accuracy. This significantly enhances operational speed and reduces the chances of error in the management of portfolios.
The Mitigation of Human Error
The prevention of errors extends beyond just calculations and data extraction. AI-powered portfolio management software can oversee investments, detect anomalies, and identify any deviations from a defined investment strategy. Real-time signals ensure that adjustments can be made promptly and that risk is kept to a minimum. Furthermore, AI can incorporate real-time market data, social media sentiment, and the latest news to present early warning signs of emerging trends.
AI-powered software can even monitor satellite data, weather reports, foot traffic in stores, and calculate whether these factors have an impact on stock prices to make ever-more accurate predictions. In industries where such data is applicable, it can give portfolio managers phenomenal oversight. This can be done for every individual asset, creating reports and insights that leap beyond what legacy systems could achieve.
AI as a Complement to Human Expertise
Taking all this into consideration, it is important to keep in mind that AI systems are not meant to outright replace human portfolio managers. Currently, these systems can be used to supplement the decision-making process and optimize portfolio performance. The objective and precise nature of AI, combined with the expertise of portfolio managers, gives investors the best of both worlds, thereby helping them achieve better-informed investment decisions with reduced risks.
Conclusion
Overall, utilizing AI in portfolio management brings significant advantages to investing. It can reduce human bias, emotional decisions, and the likelihood of errors. The objectivity provided by algorithms from processing millions of relevant data points, as well as providing continuous monitoring capabilities is a huge step forward for the financial industry. Trading firms, retail investors, and businesses will benefit from embracing AI technologies as they can vastly enhance decision-making processes, unlock new opportunities, and optimize existing portfolios. Considering the rapid progress in AI technology, the potential to further transform portfolio management is highly exciting and worth keeping an eye on.
