Artificial intelligence (AI) is one of the most transformative technologies that we have experienced in the last decade and will continue to be for the next decade. IDC predicts AI components and solutions spending will increase from $40.1Bn in 2019 to $95.5 Bn in 2022. In the simplest form, AI can be defined as simulating human-like intelligence in computer systems using learning, reasoning, and self-correction.
Many industries such as retail, travel, financial services and healthcare, among others have realized significant business outcomes, performance gains and cost savings as a result of adoption of AI-based solutions such as natural language processing, computer vision, machine learning, robotics, and virtual assistants. Although building blocks for AI solutions have existed for many years, two primary drivers for recent rapid adoption of AI-based solutions can be attributed to cheaper and faster access to compute and storage infrastructure, and the unprecedented explosion of data generation. IDC estimates that data generation will increase from 33 zettabyes in 2018 to 175 zettabytes in 2025 (1 zettabyte = 1 trillion gigabytes). Manual processing of primarily unstructured data at this scale is impossible. This is where machine learning and artificial intelligence can provide an unparalleled competitive advantage to parse massive volumes of data, both internal to an organization and external market or socio-economic information, to create decision models and augment human intelligence with infinite machine processing power.
"Artificial intelligence (AI) is one of the most transformative technologies that we have experienced in the last decade and will continue to be for the next decade"
According to a recent hedge fund survey conducted by industry data firm Barclay Hedge in July 2018, AI and machine learning was used by over 50 percent respondents to generate trading ideas and optimize investment portfolios. Historically, hedge funds have used static models to build investment portfolio and risk management applications. These applications are now updated to a modern AI/ ML architecture. AI-based models can analyze massive volumes of unstructured data including images, voice, and text to identify patterns and correlations between data points, and create predictive models for trade and risk management. These models are automatically updated in real-time based on changes to the underlying data or addition of new data sources.
Interestingly, AI will enable analysts to build market-relevant and real-time models that can leverage external data sources such as news stories and social media discussions to understand customer sentiment, weather, and satellite imagery to analyze impact on economy or commodity prices, and mobile device data and geolocation to evaluate consumer behavior. This was unimaginable a few years back where portfolio and risk models were static and were primarily built based on historical stock or company performance metrics and did not reflect constantly evolving market conditions. Advancements in deep learning models for trading will enable prediction of stock prices and volatility with a high degree of accuracy and make trading decisions with little to no human assistance. Hedge funds that adopt innovative AI technologies will gain a distinctive competitive advantage over the slow adopters, similar to what has been observed in other markets such as ecommerce and media.
Despite the advancements in AI-based solutions, there are current technology limitations that are expected to get resolved over the next several years. Some of these challenges are applicable to AI solutions targeted for hedge funds and alternative assets:
• Risk of bias in data and models
• Lack of availability of training data for certain markets or previously unobserved external conditions
• Immaturity with transfer and meta learning models where the algorithm teaches itself by applying skills learned in one scenario to others
What does the future hold for hedge fund managers? Will AI and machine learning completely take over these roles in the next five years? Not really. Hedge fund manager roles will begin to evolve, similar to observations in other industries that have seen meaningful AI adoption. AI evolution is expected to go through three phases–who creates insights, makes decisions, and acts on those decisions. Today, we are at the cusp of maturity of the first phase where AI/ML can generate insights. Human-led decisions are made based on those insights and humans are acting on those decisions. Over time, machines will become more adept at data correlation and decision-making to construct investment portfolios and analyze trading decisions whereas portfolio managers will spend more time to identify relevant parameters to enhance and optimize these models, ensure trading models align with overall fund strategy and investor needs, and intervene as needed for new or corner use cases and previously unobserved but relevant market dynamics.
The use of AI in hedge funds and alternative asset trading will only continue to grow and will re-shape roles and responsibilities of hedge fund managers in a significant way. The future of AI in financial services will not create a man vs. machine situation, but will instead lead to developing machine-augmented, human-governed intelligent systems that will yield several orders of magnitude higher gains and efficiency.