AI Hedge Fund Navigates Market Turbulence with Positive Performance

The AI-based hedge fund Omphalos has achieved positive performance in the early months of this year – including during the market tremors triggered by Donald Trump’s trade policies. In the interview, co-founder Pawel Skrzypek highlights the advantages of AI over discretionary management.

By Tim Habicht · April 22, 2025 | Original version in German language on Fundview.de

Donald Trump has caused the stock markets to tremble. How has artificial intelligence reacted to Trump’s thoroughly opaque tariff policy?

Pawel Skrzypek: Our “AI portfolio managers” knew no more than any other market participants about what Trump would specifically announce. But they did exactly what they were developed to do: they reacted immediately to the actual market movements. We operate with a short-term outlook and very fast adjustment cycles. Every day, our system processes thousands of data points, identifies patterns, and adjusts the allocation accordingly.

Our trading agents have now reached a prediction accuracy of over 60 percent. This enables them to make sound, data-driven decisions even in highly uncertain market phases. The announcement was indeed a surprise – but the AI interpreted the resulting signals in real time and acted accordingly. And that is one of the core strengths of our model: it’s not about predicting headlines, but about responding intelligently and consistently to data.

Many discretionary fund managers have discussed whether portfolios should now be positioned more defensively overall. How has the AI reacted and positioned the portfolio?

Skrzypek: This kind of strategic repositioning is precisely what our system performs continuously and dynamically. Our average investment horizon is around seven days – meaning the portfolio is in a constant state of adjustment. There is no need for a single manual decision for a “defensive shift” because the model reassesses the risk-return profile of each position every day.

In addition, we trade a highly diversified portfolio across different asset classes – equities, currencies, and commodities – and combine various strategies. Even if individual positions or strategies come under pressure in a volatile market environment, others compensate for this effect. The result is a portfolio that adapts in real time and remains structurally resilient – delivering positive overall returns, even when parts of the system are temporarily under stress.

Was the market turmoil and the subsequent positive performance of the strategy a real litmus test for the model?

Skrzypek: Absolutely – and it wasn’t the first. Our model has already proven itself during the COVID pandemic, at the outbreak of the war in Ukraine, and now once again under geopolitical pressure. In all these situations, the system functioned completely independently and without human intervention. And if you look at the numbers, the result is clear: since launch, the fund has achieved positive performance in 37 out of 39 months – including all three positive months so far in 2025, with a year-to-date performance of +1.5 percent.

Situations like these are often more of a test for us humans. We like to believe that our intuition is superior. But we have seen again and again: the machine – the AI – delivers. Without panic, without bias, and based on clear rules. And it is precisely this discipline that makes the difference.

The markets remain volatile, at least in the short term. How does the AI deal with this high volatility?

Skrzypek: Volatility doesn’t scare us – quite the opposite. It creates additional opportunities, provided the predictions are correct. Of course, losses are part of daily trading. But the expected return of our system remains clearly positive – especially when markets are moving and delivering stronger signals.

Our prediction accuracy of over 60 percent gives us a real statistical edge. The model doesn’t simply follow momentum or trends – it processes multidimensional information and continuously updates its assessments. It is trained on a wide range of historical patterns but does not get stuck in the past. It recognizes: “This spring feels like the last – but it’s still different.” We like to call it “same same, but different.” This form of adaptive learning allows us to operate with confidence even in volatile market phases.