Artificial intelligence, or AI, may seem like science fiction to many people, but recent breakthroughs are opening the door to new possibilities. The Rise of Big Data provided the raw materials; graphical processing units (GPUs) provided the hardware; and new algorithms expanded the application of AI to new fields, such as language processing.
AI-powered technologies have also made their way into finance with the rise of algorithmic trading and robo advisors. A new crop of fintech companies are further leveraging artificial intelligence (and machine learning) to do everything from automating savings to managing risk for both individual consumers and institutional investors.
Let’s take a closer look at artificial intelligence, the implications for portfolio management and some drawbacks to keep in mind.
Click here to explore the portfolio management channel and learn more about different concepts.
What Is Artificial Intelligence?
Artificial intelligence is a broad term used to describe any technology that makes computers behave in a way that previously required human intelligence. For example, Apple’s Siri and Amazon’s Alexa use natural language processing (NLP) algorithms to interpret language whereas AlphaGo defeated a world champion Go player in a highly complex game.
Some of the most important AI techniques include:
- Natural Language Processing
- Artificial Neural Networks
- Decision Trees
- Support Vector Machines
- Cluster Analysis
- Genetic Algorithms
Machine learning is arguably the most popular branch of artificial intelligence and is focused on developing algorithms that automatically improve through experience. For example, Pandora’s music recommendation algorithms use machine learning algorithms to compare your listening history and ratings with those of other users to make intelligent suggestions.
Use the Dividend Screener to find the security that meet your investment criteria.
Implications for Portfolio Management
Artificial intelligence could fundamentally transform asset allocation, trading processes, risk management and other areas of portfolio management. In fact, many robo advisors already use these technologies to deliver portfolios with better out-of-sample performance for investors while rebalancing and automatically managing risks with minimal transaction costs.
Some of the most popular portfolio management applications include:
- Fundamental Analysis: AI techniques can be used to conduct textual analysis of annual reports, economic reports and other meaningful information. Other techniques can identify hidden correlations between asset classes and then pinpoint stocks that could outperform or underperform based on those correlations.
- Portfolio Optimization: AI techniques can more accurately predict expected returns and variances/covariances to determine optimal asset weights. Genetic algorithms are also capable of solving complex optimization problems with constraints, such as restricting the number of assets or setting minimum holding thresholds.
- Risk Management: AI techniques can incorporate qualitative data (e.g., news reports or social media), forecast risk variables (e.g., bankruptcy probability) and validate existing risk models to minimize risk and ensure that total risk falls within acceptable risk tolerance levels for clients.
- Trading Activities: AI techniques are already widely employed by algorithmic traders. In addition to predictions with technical analysis, these techniques can assist with analyzing transaction costs and executing large trades that must often be broken up into smaller chunks to achieve the best price.
These possibilities are likely to expand over time as artificial intelligence techniques evolve and expand into new areas. Portfolio managers should keep an eye on these trends, particularly as robo advisors and other competitors are actively exploiting them.
Risks & Drawbacks To Keep in Mind
Artificial intelligence makes decisions using complex neural networks, genetic algorithms and other techniques. While these strategies tend to produce better results, their sheer complexity makes it difficult to understand what’s happening under the hood. The “black box” algorithms may include inherent biases or be unprepared to cope with “black swan” events.
There are also several data-related challenges with AI:
- Training Data: Artificial intelligence algorithms often require large amounts of training data to calibrate their models. Insufficient or incomplete data sets can make it challenging to produce high-quality AI applications.
- Data Quality: Poor data quality results in what AI experts call, “garbage in, garbage out.” While imputation helps fill in some of the blanks, the process is arduous and has its limitations in terms of yielding the highest quality results.
- Bias: The quality and availability of training data may also impact an AI algorithm’s bias. These trends have been well documented in the case of bail bonds and other areas, but they may also be apparent in financial markets.
- Black Swans: The availability of historical data could lead to inadequate experience with so-called “black swan” events. Without exposure to these rare events, AI algorithms may be unprepared to deal with them and experience adverse consequences.
Artificial intelligence also requires an extensive use of technology in general, which can increase a firm’s risk exposure. One example would be a cybersecurity breach that exposes sensitive data, but other examples might include data loss that impacts the efficacy of AI algorithms or the loss of an algorithm to a competitor that hurts a firm’s competitive edge.
The Bottom Line
Artificial intelligence is a broad term that covers many different concepts, including machine learning. With advances in big data, GPU processing and deep learning algorithms, the financial markets have started to experience a significant shift. Portfolio management in particular could benefit from many of these trends.
Be sure to check out our News section to keep track of the latest updates around income investing.
