Exploring the applications and challenges of AI in finance.
By Mirella Deng, Director of Finance (McGill)
From Siri to self-driving cars, the innovation landscape for artificial intelligence (AI) is rapidly expanding before our eyes. Rooted in pioneering advancements, AI is taking over various industries by storm. In the realm of finance, it is one of the transformational technologies that can reinvent and enhance critical business functions. Its usages range from robo-advisory services to natural language processing in investment analysis. Though there are certain trade-offs and limitations that we must consider, AI ultimately serves as a competitive advantage for many firms in today’s complex and interconnected global investment environment.
AI in a nutshell
Originating during the first half of the 20th century, artificial intelligence is the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings. It is often used in developing systems that possess certain human characteristics, including the ability to learn and improve from past experiences, reason, and draw conclusions. While AI has become as competent as humans and is an integrated element in our everyday life, it remains far from being able to replace the creativity and flexibility of the human brain.
According to Vasant Honavar, director of the Artificial Intelligence Research Laboratory at Penn State University, AI can be divided into two key components: engineering and the science of intelligence. Through very different processes, the former focuses on building tools that use intelligence, while the latter studies the ways in which a computer can be programmed to come up with a solution similar to that of a human brain. Ultimately, the goal of AI is for autonomous systems to emulate functions and processes of the human brain.
Within the finance industry, the use of AI is increasingly important. Companies that invest significant amounts of capital into this technology can differentiate themselves from their competitors in aspects of high-efficiency, heightened security, and more. Their technological edge can open new doors, leading to future growth opportunities. Below captures some of the top uses of AI in the financial markets:
Making their debut in 2008 with the first robo-advisor Betterment, these AI-enabled advisory platforms provide algorithm-driven, automated financial planning services with little to no human supervision. Compared to traditional advisory models, the customized digital advice services can be offered to clients at a reduced cost. Robo-advisory tends to start with an investor questionnaire, which may include characteristics such as age, income, risk preferences, and target investment returns. Based on an investor’s general information, risk capacity, and risk tolerance, the tool provides recommendations of the appropriate asset allocation mix through the application of algorithmic rules and historical market data. Digital advisors also offer tax-loss harvesting, digital documentation delivery, portfolio allocation rebalancing, and trade execution. These automated and customizable features would have been out of reach without the technological advances seen in AI.
In light of the 2008 Financial Crisis, the financial industry has undertaken major steps in improving risk assessment capabilities, through the analysis of vast amounts of qualitative and quantitative data. For example, fund managers may perform scenario analysis and portfolio back-testing using point-in-time data to understand the liquidation cost and portfolio consequences under different market conditions. As back-testing simulations are often computationally intense, traditional risk analysis can no longer process the ever-increasing volume of data. Advanced AI-based techniques allow managers to perform “what-if” simulations efficiently to identify market conditions and trends in advance. With the result of such predictive analysis, managers can employ hedging strategies sooner to preserve asset value.
Algorithmic Trading & High-frequency Trading
Following pre-specified rules and guidelines, algorithmic trading is the computerized buying and selling of financial instruments. Most algorithm trading consists of high-frequency trading (HFT), which makes use of vast quantities of financial data to automatically place a large number of orders at rapid speeds across multiple markets based on preprogrammed instructions. Over the day, HFT algorithms may continuously revise their execution strategies based on changing prices, volumes, and market volatility. They may also determine how to place the order (limit or market order) and the most appropriate trading venue (exchange or dark pool). In general, algorithm trading dramatically enhances the speed of execution, ensures investors’ anonymity, all while reducing transaction costs.
Natural Language Processing
Natural language processing (NLP) is a field of research at the intersection of artificial intelligence, computer science, and linguistics that focuses on developing a computer program to analyze and interpret human languages. Through analyzing annual reports, call transcripts, social media posts, and other audio-based or text-based data, NLP is used to detect subtle messages and identify trends with greater scale and accuracy than what is humanly possible. The three major NLP applications in the financial markets include intelligent document search, customer services, and investment analysis. Within investment analysis, banks used to hire armies of analysts to comb through companies’ earnings reports and financial statements in order to keep their databases and valuation models up to date. However, the emergence of the NLP systems makes it possible for banks to read hundreds of documents simultaneously, while summarizing the critical information, thus allowing equity analysts to save time and focus on the alpha generation. On top of its primary data trimming function, the algorithm can also perform sentiment analysis. To determine how a company is perceived by the markets, the tool analyzes transcripts to extract critical insights and assign sentiment ratings ranging from negative to positive. Similarly, based on nuances such as word choice and inferred tone in social media, financial news, and other alternative sources, NLP can provide insights into trends in a company’s performance. For example, using an AI-armed investment platform developed by their innovation group, UBS Wealth Management uses NLP to speed up its due diligence processes by detecting negative news when reading vast amounts of documents fetched from search engines.
Challenges & Limitations
Though the proliferation of AI has proven to give financial institutions a competitive advantage, it is important to consider the following limitations.
Cost: To procure AI is costly. On top of the sheer investment that firms must make to implement the technology in an effective manner, AI also requires regular updates to cater to the needs of an ever-changing business environment. In cases of systematic failures, the costs to repair damages caused by these smart technologies can be immense. In 2012, Knight Capital Group, an American market maker suffered a $461M loss after its electronic trading systems went haywire. After taking 17 years to become one of the top trading houses on Wall Street, all was lost in under one hour.
Integration Challenges: From the potential lack of understanding of AI systems within a firm to challenges in its usability and interoperability with other systems and platforms, the integration process of AI is tricky due to a diverse combination of needs. Historical data is needed to train the machine learning models that drive AI while there is also the need to host a complex set of technologies. In addition, the prediction power of any algorithm is highly dependent on the quality of the data that it is fed.
Wide-reaching Unemployment: It is estimated that by 2030, up to 800 million people worldwide, including a third of the workforce in the U.S., will be jobless, with up to 30% of work hours worked globally being automated. As the use of AI becomes rampant, the vast wealth inequalities among countries are revealed, leading us to consider whether the development of AI is environmentally sustainable to society.
Privacy & Ethical Issues: Building on the severe unemployment concerns and wealth inequality gap, AI also poses ethical questions such as the looming fear of an AI takeover. While humans can creatively take individual circumstances into account when making decisions, AI lacks emotion and moral values, and therefore risks containing their programmers’ biases. Intertwined into the technological complexity of AI are also privacy concerns such as client security issues and the potential lack of transparency of the technology’s uses.
Over the last few decades, technological advancements in artificial intelligence have increasingly reshaped many industry landscapes, including the financial markets. Its systems can comprehensively improve operational efficiency, reduce costs, mitigate risk, generate higher returns, and enhance user experiences. Though AI’s incorporation into finance remains at an early stage, financial institutions that fail to exploit this technology will increasingly find themselves at a competitive disadvantage. As the technology becomes more accessible and computing power is continuously improved, the future presents AI with unprecedented insights and opportunities.