Finding attractive small caps with the help of artificial intelligence

In recent years, the concentration in the MSCI World Index has significantly increased. The USA dominates the index with a 74 percent share, mainly due to the market capitalisation of large US technology companies, the so-called "Magnificent 7". The IT sector is the largest sector in the index with 26 percent, followed by financials with 16 percent.

The sustainability of this concentration depends on the continued profitability of the large tech firms and whether they can meet high earnings expectations.

Small Caps: Attractively Valued with Upside Potential

In contrast, small caps are attractively valued, partly because they have underperformed large caps in most regions since 2018, except for emerging markets. Historically, small-cap companies have generated higher returns than large-cap companies over the past 100 years.

Another advantage of small caps over large caps is their broader diversification. With around 4,000 small-cap companies, this segment offers a vast universe of investment opportunities.

While the IT sector dominates large caps, it is significantly smaller in small caps. This is because small-cap companies typically cannot finance the billion-dollar investments in data centres. Instead, industrials, financials, and consumer discretionary are the largest sectors in small caps, leading to a more balanced sector distribution.

Less Efficiency, but More Opportunities for Active Management

However, small caps are significantly less covered by analysts compared to large caps. On average, only eight analysts cover a small-cap company, whereas around 20 cover large caps. Notably, Amazon, Meta, and Google are analysed by up to 72 analysts.

This lower coverage can also present an opportunity for active management, as there are more undiscovered potentials. Historically, small caps have shown higher earnings growth and higher EPS dispersion (standard deviation of earnings per share) than mid or large caps. This means there are greater differences between companies, providing more opportunities for active management. Therefore, it is even more crucial to select the right stocks with alpha potential from the small-cap universe.

AI reveals attractive small cap

Analysing the approximately 4,000 small caps and selecting them for a fund based on various factors would require an army of analysts. This is where the machine learning algorithms developed by the asset management team of Zürcher Kantonalbank come into play. They "work" for the "Swisscanto (LU) Equity Fund Systematic AI Responsible Small Caps", launched at the end of 2024.

They comb through a proprietary stock database of colossal size daily, consisting of 7,000 stocks with more than 500 metrics collected over a period of 25 years. The sheer size of the stock database is an ideal training ground for the algorithms. They learn independently from data and uncover previously hidden correlations between financial data and future stock returns, aiming to identify the most lucrative stocks – all free from human emotions.

Features of the AI algorithms (not exhaustive):

• Pattern Recognition: AI helps uncover hidden correlations between company metrics and future returns.
• Rational: AI is free from human emotions and biases that can lead to irrational decisions.
• Rapid Decision-Making: AI can analyse large amounts of data in a short time, which would be unmanageable for humans. The results serve as a basis for stock selection.
• Dynamic: AI learns independently and adapts dynamically to changing market conditions. Accordingly, trends, opportunities, and risks can be identified early.
• Perfect Memory: The AI algorithm we designed accesses company data from the last 25 years and continuously learns from the latest events.
• Risk Management: AI monitors risks and can identify high-risk stocks. The goal: loss reduction.
• Efficient: AI can make daily forecasts for thousands of stocks without the need for research teams.
• Uncorrelated: AI-driven stock forecasts differ from human forecasts and can therefore be a valuable addition to traditional investment approaches.

Investment Approach: Dynamic and Data-Driven

Investments within our systematic investment process are made along the "Responsible" sustainability criteria implemented by the asset management team of Zürcher Kantonalbank. The subsequent portfolio construction, consisting of 100 to 200 stocks, is structured to maximise expected returns while systematically controlling risks. The process is accompanied by ongoing multi-level risk and quality management.

Risk Control and Sustainability

Unwanted risks are avoided by limiting single stock, sector, and regional bets. Managing ex-ante tracking error and limiting turnover, as well as considering stock-specific transaction costs, are further measures for risk control.

In addition to ESG integration, stewardship, "Responsible Exclusions," and pursuing an ambitious <2-degree reduction path in line with the Paris Agreement are applied. The "Swisscanto (LU) Equity Fund Systematic AI Responsible Small Caps" is classified as an SFDR 8 fund.

Conclusion: using artificial intelligence to find small caps