At K&H Alapkezelő, artificial intelligence (AI) is a practical tool rather than a futuristic experiment: the firm has been developing models for nearly ten years. Those systems evaluate more than one thousand economic and market indicators daily. A ten-person mathematics team continuously improves the platform, and the AI ingests over forty thousand information sources rated on a forty-one point scale.
Tamás Ács, head of K&H Alapkezelő, explained on the latest episode of the Portfolio Checklist podcast how AI is integrated into fund management across three levels, combining machine processing with human oversight.
The three levels
-
Level one: increasing portfolio managers’ efficiency. The AI analyzes company valuations, compares analyst opinions and highlights discrepancies among expert viewpoints, providing faster and clearer information for managers.
-
Level two: concrete investment suggestions. For some funds, AI-generated recommendations appear as standalone inputs to the investment committee.
-
Level three: AI-driven allocation. In certain cases the system directs portfolio allocation decisions. Human control remains: final approval is always carried out by people.
Practical examples
For a mixed (balanced) fund, the AI determines optimal asset allocation on a daily basis — deciding how much equity, bonds or commodities to hold and what currency exposures to take. The system aims to detect anomalies, correlations and patterns to forecast expected returns by asset class, sector and region. Performing these analyses daily on such a volume of data far outpaces what traditional methods could achieve in the same timeframe.
In equity selection, the AI uses classic corporate metrics (such as P/E and EV/EBITDA), as well as profitability, growth and volatility indicators. It also processes increasingly hard-to-quantify inputs: ESG scores, patent and trademark data, and various text sources including quarterly reports, investor conference call transcripts and industry press articles.
According to Ács, the ability to process large data sets enables quicker market responses; that does not mean the portfolio is changed every day, but the system can react much faster when needed.
Risk management and limits
Ács outlined three layers of risk control:
- Data quality: the AI is only as good as the data it receives — data quality is fundamental.
- Defined boundaries: AI-managed funds operate within strict limits. For example, a mixed fund’s equity weight may deviate from its benchmark by at most plus or minus ten percent, so the AI cannot raise the fund to 100% equity exposure. Sector and regional weights are similarly constrained in equity funds. The system selects companies within these predefined limits, but it cannot override the reference benchmark itself.
- Human validation: human experts validate and approve decisions.
One advantage Ács highlighted is that AI evaluates situations consistently using the same data and rule set, reducing subjective biases in decision-making. He illustrated this with a 2025 tariff-war example: when Donald Trump presented new tariff measures and markets fell, many investors sold, but K&H’s AI-managed mixed fund — which had already held a higher equity share — further increased exposure and benefited from the subsequent market rebound.
The human role and client relationships
K&H Alapkezelő stresses that AI does not replace portfolio managers or bank advisors. Technology shortens the time spent on data processing, allowing professionals to focus more on strategic decisions and on understanding client needs. Most clients still expect personal advice, and the AI developments are intended to support fund operation and optimization rather than substitute the advisory relationship.
Additional notes
Details about the development effort come from the asset manager’s comments. The Portfolio Checklist episode is available on Spotify, Apple Podcasts and other major podcast platforms. The article’s preparation involved an AI assistant, but the final content was edited and verified by a journalist.
This article was published with the support of K&H Alapkezelő. The cover image is illustrative.
