The advent of ChatGPT has catapulted artificial intelligence (AI) into the mainstream conscience.
AI’s ability to replicate functionality in various industries such as accounting, retail, and logistics has left many working in these sectors wondering when, if not if, AI will affect their work. I am also wondering if I can learn how to work with technology.
In the investment context, some question whether AI can predict the stock market.
The answer is not easy. There have already been some advances in AI and machine learning (ML) that have changed the way analysts manipulate and interpret markets.
Algorithms can struggle to find relationships between drivers and stock return outcomes
These include the ability to perform functions such as data cleaning, credit scoring, and transaction optimization. The bigger question is whether AI can reliably predict the stock market.
At the moment the answer is no, but I can’t imagine this changing. Geoffrey Hinton, considered the “godfather of AI,” recently said that general-purpose AI (a system that can learn any task a human can perform) could exist within 20 years of him.
However, financial markets are very different from other areas where AI works well, such as the physical sciences and consumer internet domains. One of the big differences is the low signal-to-noise ratio of financial data. This means that there is no single variable that determines how something runs.
Variables such as corporate earnings, bank interest rates, and investor attitudes can all affect investment returns. This is different from algorithms that give streaming content her channel subscribers movie recommendations based on the types of movies the subscriber has already watched on the platform.
There already exist ways for AI to indirectly predict the market by determining the factors that influence the market.
Therefore, when it comes to complex financial markets, ML algorithms can struggle to find relationships between drivers and stock return outcomes.
Another challenge for ML is the amount of data available in financial markets compared to other domains. A key factor in an ML algorithm’s ability to accurately predict trends is the amount of data at its disposal.
The data traditionally used by quantitative investors is often only available on a quarterly or monthly basis, pales in comparison to other domains.
It can also take into account the inherent fluctuations that exist within financial markets. This non-stationary character sets it apart from fields where ML has distinguished itself, such as physical sciences, which often have static systems.
While many ML algorithms can be designed to adapt to evolving systems, the question remains of how valid and applicable the historical data used to train the algorithms is.
There are significant hurdles to overcome before AI can reliably predict financial markets, but there are already ways to indirectly predict markets by determining the factors that influence them. To do.
One example is an ML algorithm that predicts company fundamentals such as company earnings. Not only are there fewer variables that can affect company fundamentals, but they are often more stable than stock returns, making them more suitable for ML forecasting.
Financial professionals want to learn how to leverage AI to gain an edge.
As AI becomes more and more entrenched in the investment process, its presence is growing and the appetite for informed predictions is growing. And, as Hinton acknowledged, AI capabilities are being developed at a faster pace than many expected.
It may not be long before we can reliably predict stock market returns.
Larry Cao is Senior Director of Industry Research for the CFA Institute.
