Artificial Intelligence (AI) is ramping at a faster pace than expected and it will soon become well engrained into almost every task we undertake, or at least that is what all of us are being told. We’ll allow for a little hype in these broad proclamations, but there is also real substance behind this. In the past year, we have seen applications of AI become more integrated into smartphone features, customer service, search engines and virtual assistants like Alexa, airline scheduling, navigation routing, healthcare diagnosis, entertainment recommendations, and social media feeds, to name a few.
In business, AI will help companies manage their resources better and improve the products and services they provide. In the investment management/wealth management industry, we should expect AI to help us make more informed decisions and provide more complex modeling in the financial planning work we do for clients.
As a process-driven firm, we have always been open to new tools and technologies to strengthen our investment strategies. As such, we thought it would be helpful to explain how Machine Learning, a powerful segment of AI, enhances our security selection process.
At a simple level, Machine Learning allows computers to learn from data by identifying patterns and relationships among different variables. Based on this, predictions of future outcomes are made with accuracy gradually improving over time as this iterative process continues at speeds beyond human capability. Our stock selection process incorporates a Machine Learning component to analyze a massive amount of data, such as historical stock prices, company financials, and economic indicators, uncovering patterns and making predictions based on historical outcomes.
Meritage incorporates these signals as part of a screening process that combines with other key evaluation criteria – Valuation, Earnings Quality, Market Reaction, and Capital Deployment. Over multiple rolling five-year periods, the Machine Learning component has identified factors as positive signals of future price performance dealing with free cash flow at a reasonable price, strong margins as seen in the ratio of price-to-forecast sales, distinguished return of capital and stock buybacks, and favorable trends in company top line growth and stock prices.
One of our key research sources began developing these tools three years ago and has recently incorporated ChatGPT to add contextual data from management commentaries from earnings calls and presentations. This broadening of Machine Learning beyond raw data expands the program’s ability to imitate the way humans learn, only faster and more consistently.
The Machine Learning program also incorporates qualitative information about companies that are embroiled in “controversy,” such as unusual price swings, negative media attention, high trading volume, and earnings volatility, to name a few. Stocks with these characteristics often underperform the market due to investors losing confidence. The Controversy Indicator can also serve as a tool for identifying stocks that have overreacted to perceived controversy issues.
We accept that systematic processes like this are not without their own shortcomings, but we do expect the contribution of AI-related tools to play a more important role in our analytical processes, just as it will continue to do in our day-to-day living.
As always, we welcome your feedback and questions.
