Artificial Intelligence in Investment Management

Introduction

Citadel LLC, the American multinational hedge fund and financial services company, generated USD 8 billion on its trades in commodities in 2022, at a time when global equity and bond markets were largely struggling. One of the likely contributing factors to these outsized gains is the use of supercomputers by data scientists and analysts to forecast weather patterns in advance, something which enabled traders to predict gas demand with a high degree of precision.[1]

A survey carried out by the London-based market intelligence specialist Market Makers revealed that 9 out of 10 hedge fund traders will use Artificial Intelligence (AI) to achieve portfolio returns in 2023. As per AI engineer Matt Forbes, one of the biggest value propositions of AI is its ability of pattern recognition which can potentially eliminate hours of research before deciding which stocks or currencies to trade. AI can also reduce human biases in the investment management process.[2]

The 21^st century has witnessed not only an exponential growth of data in finance but also the evolution of technologies that facilitate data processing at a scale far beyond the limits of human capabilities. Traditional finance as we know is already undergoing disruptions across the entire suite of wealth management services. Across wealth management, AI has taken numerous forms from chatbots for client engagement to sophisticated tools for portfolio management. In this article, an attempt is made to discuss the growing role of artificial intelligence in investment decision-making.

Select AI Technologies in Portfolio Management

The origins of AI in stock markets can be traced to the Protrader expert system, the AI-driven program that successfully predicted the 87-point drop in the Dow Jones Industrial Average (DJIA) index in 1986.[3] Artificial intelligence in portfolio management involves the use of computer programs and algorithms to make investment decisions. AI capabilities range from generating insights from voluminous and complex data sets to taking investment decisions with zero human intervention. Some of the most used AI capabilities are as follows:

• Machine Learning (ML): This technology makes extensive use of algorithms that analyze extensive data sets, identify patterns, and predict future outcomes by training and testing the ML model on historical data. ML assists in functions such as portfolio optimization and strategy generation in an evolving data-heavy investment landscape. A highlight of ML is the ability of the technology to automatically improve its predictive capabilities as more and more information gets incorporated as part of its input.
• Natural Language Processing (NLP): This technology enables the generation of insights from text information such as news articles, research reports, tweets (and now Threads), and user interactions on online platforms. This makes it possible to gauge market trends, investor sentiment, and company-specific information, thus allowing investment professionals to go beyond quantitative data for making decisions.
• Algorithmic Trading: This technology involves the use of computer algorithms to automate the execution of trades based on predefined rules with a focus on optimizing trading strategies, analyzing market conditions, identifying trading opportunities, achieving best execution, and minimizing risk. Algorithmic trading encompasses a broader range of trading styles and timeframes. As per Select USA, roughly 65-70% of the overall trading volume in the United States and some developed markets is generated through algorithmic trading. We can expect this share to only grow in the coming years. In emerging economies including India, this figure is estimated to be 40%. [4]
• High-Frequency Trading (HFT): HFT is a subset of algorithmic trading that emphasizes ultra-fast trade executions. The objective of HFT strategies is to exploit price discrepancies and market inefficiencies in time frames that span milliseconds (one-thousandth of a second) to microseconds (one-millionth of a second). This type of trading utilizes advanced technology to execute large trades in durations within which it is impossible for any human trader to manually execute a trade. HFT is best suited for arbitrage trades such as index arbitrage, volatility arbitrage, and statistical arbitrage as well as for market-making activities.

Potential Risks of AI in Portfolio Management

On May 6, 2010, the Dow Jones Industrial Average plunged over 1000 points or 9% in less than two minutes. The volatility that followed resulted in USD 1 trillion getting wiped off in market value in less than 30 minutes, though most of the losses got recovered in less than an hour. This ‘Flash Crash of 2010’ is considered as the largest such crash in history and was a result of algorithmic trades which exacerbated an already negative sentiment prevailing in the market on that trading day. Investigations revealed large buy and sell contracts executed by HFTs, with the combined sale triggering the nosedive in the DJIA.[5] Several instances of flash crashes have been attributed to HFTs and other sophisticated technologies used by traders.

Risks also emerge when harnessing the functionalities of AI in constructing long-term portfolios. AI tools largely rely on historical data and patterns for training and testing predictive models. Especially in cases where the human element in deciding portfolio allocation is eliminated, we cannot be completely sure of the potential of such AI-based models to adapt to changing market dynamics, unforeseen situations, or black swan events. Additionally, the black-box nature of these tools means that it is challenging to understand and explain the decision-making process of these tools, raising concerns about transparency and accountability.

The CFA Institute has released a framework, namely, ‘Ethics and Artificial Intelligence in Investment Management’ to address ethical considerations amid the challenges and opportunities posed by AI. The Ethical Decision Framework for AI addresses obtaining data, building-training-testing the model, and deploying and monitoring the model on the considerations of Data Integrity, Accuracy, Transparency, and Accountability.[6]

Conclusion

Ken Griffin, the CEO of Citadel LLC, has said that his companies are looking at implementing ChatGPT on a business-wide scale, acknowledging the real impact which this branch of technology has on business. The potential Citadel sees in generative AI spans includes helping developers to write better code, translating software between languages, and analyzing various types of information.[7] Thus, with the deeper integration of Industry 4.0 considerations in financial services, investment professionals have an opportunity to partner with emerging technologies and integrate them into their fund management process. In conclusion, by acquiring the appropriate skill sets to navigate the evolving landscape of wealth management, investment professionals can ensure that artificial intelligence complements their expertise rather than replaces it, thus, safeguarding the enduring value of their professional role.

Additional Reading Material

The full text of ‘Ethics and Artificial Intelligence in Investment Management: A Framework for Professionals’ is made available by the CFA Society under the Research and Analysis section. You can download the material by accessing this link:

https://www.cfainstitute.org/-/media/documents/article/industry-research/Ethics-and-Artificial-Intelligence-in-Investment-Management_Online.pdf

References

[1] Sor, Jennifer. (Updated 2023, March 06). A huge team of scientists helped Citadel generate $8 billion from commodities bets in 2022. Markets Insider.
https://markets.businessinsider.com/news/commodities/citadel-ken-griffin-gas-prices-commodities-market-bets-russia-europe-2023-3

[2](Updated 2023, January 18). Nine out of ten hedge funds to use AI in 2023, says survey. Hedge Week.
https://www.hedgeweek.com/2023/01/18/319130/nine-out-ten-hedge-funds-use-ai-2023-says-survey

[3] Jarvis Invest. (Updated 2022, May 04). AI In The Stock Market – When Did It All Start? LinkedIn article.
https://www.linkedin.com/pulse/ai-stock-market-when-did-all-start-jarvisinvest/

[4] Samuelsson. (Updated 2023, April 20). What Percentage Of Trading Is Algorithmic? (Algo Trading Volume). The Robust Trader.
https://therobusttrader.com/what-percentage-of-trading-is-algorithmic/

[5] (Updated 2021, August 21). The Flash Crash of 2010. AWA Trade. https://www.avatrade.com/blog/trading-history/the-flash-crash-of-2010

[6] Preece, Rhodri G. (Updated 2022, October 14). Ethics and Artificial Intelligence in Investment Management: A Framework for Professionals. CFA Institute.
https://www.cfainstitute.org/en/research/industry-research/ethics-and-artificial-intelligence-in-investment-management-a-framework-for-professionals

[7] Tayeb, Zahra. (Updated 2023, March 08). The hedge fund that just posted the best return in history is negotiating a company-wide ChatGPT license.
Business Insider. https://markets.businessinsider.com/news/stocks/ken-griffin-citadel-chatgpt-license-ai-hedge-fund-artificial-intelligence-2023-3

About the Author

Mihir Shirgaonkar, CFA heads the Alternative Investments division at Phillip Ventures IFSC Pvt. Ltd. and has been a part of the firm since 2021. He is a Chartered Accountant and has completed MBA-PGPX from the Indian Institute of Management Ahmedabad. He has over 8 years of asset management experience and has worked with DSP Investment Managers (erstwhile DSP BlackRock) and HDFC AMC in his previous roles. In his career so far, he has handled multiple areas which span portfolio management, market making, valuations, fund administration, and project management. His other interests include philosophy, reading, photography, trekking, coding, cooking, and travelling.