^{Photo: LinkedIn}

The Early Roots of AI in Finance

Long before artificial intelligence became a buzzword on Wall Street, Miro Mitev was studying neural networks in the late 1990s at the Vienna University of Economics and Business. While the internet was only beginning to reshape global commerce, Mitev was already exploring how machine learning could be applied to financial forecasting. He recognized early that algorithms could detect patterns and signals far beyond human capability if they were designed correctly.

That curiosity shaped a career spanning more than 25 years, including forecasting roles for banks and global technology companies such as Siemens. Over time, his focus narrowed to one core question: how to remove emotion from investment decisions without removing human responsibility.

A Firm Run by Algorithms, Built by People

Mitev is now the CEO of SmartWealth Asset Management, a firm where investment decisions are executed entirely by a network of AI systems. The company’s latest strategy, the IVAC fund, is targeting up to 2 billion dollars in assets under management and aims for annualized returns in the 14 to 15 percent range.

Despite the absence of human traders making day to day calls, Mitev stresses that people remain central to the process. Humans define the investment universe, select and clean training data, choose variables, set constraints, and continuously refine the models. In his view, algorithms execute strategy, but humans design intelligence.

Trusting the Model Without Interfering

One of Mitev’s core principles is disciplined trust in the system once it is deployed. He warns that intervening after a model is live can be dangerous, particularly when short term results challenge human intuition. Overruling algorithmic outputs, he argues, is one of the most common reasons AI driven strategies fail.

Many investors initially struggle to trust AI because results are not always immediately intuitive. However, retrospective analysis often shows that the model’s decision was correct when evaluated weeks or months later. Removing emotional overrides is essential to capturing the full value of systematic investing.

Markets Are Emotional, Algorithms Are Not

Financial markets are driven by deeply human forces such as optimism, fear, speculation, and herd behavior. Even central banks have acknowledged that recent AI related rallies may be fueled more by fear of missing out than by fundamentals.

SmartWealth’s approach is designed to counteract these behavioral biases. By stripping emotion from execution, the firm has reported cumulative gains of over 400 percent across a ten year period ending in November 2025. Over the same timeframe, a comparable industry benchmark delivered roughly 145 percent. While past performance does not guarantee future results, the contrast highlights the potential advantages of disciplined, data driven decision making.

What AI Can and Cannot Predict

Mitev is clear about the limits of artificial intelligence. Long term market outcomes remain inherently uncertain, but his models aim to forecast trends and probabilities up to one month ahead. Consistency, not prediction perfection, is the objective. Repeatedly making decisions based on statistically informed signals has proven more effective than relying on human judgment alone.

Continuous monitoring is essential, particularly because AI systems can generate flawed outputs. These errors often stem from overfitting, poor data quality, or incorrect model assumptions rather than from the concept of AI itself.

Avoiding Overfitting and False Signals

Overfitting occurs when an algorithm focuses too heavily on noise rather than meaningful cause and effect relationships. This can lead to impressive historical results that fail in real world conditions. To prevent this, SmartWealth emphasizes rigorous model design, extensive validation, and prolonged testing in live market environments.

This process requires patience and long term investment. According to Mitev, effective AI systems evolve over years, not months. That is why in house development remains critical for firms seeking durable differentiation rather than off the shelf solutions.

The Human Role in an Algorithmic Future

While algorithms execute every trade, humans remain responsible for governance, oversight, and strategic evolution. The success of AI driven investing depends less on replacing people and more on redefining their role. In Mitev’s framework, humans do not compete with machines. They enable them.

The result is a model where technology amplifies discipline, consistency, and scale, while human expertise ensures relevance, integrity, and long term resilience.