Imagine pouring years into perfecting a system that makes investment decisions better than most humans ever could—only to insist that people are still the heart of the whole operation. That’s exactly the stance of one pioneering fund manager who’s been working with artificial intelligence since the days when the internet was still dial-up slow.
It’s a fascinating paradox, isn’t it? In an era where everyone talks about AI taking over everything, here’s someone running a completely algorithm-driven firm who flatly says overriding the machine is the worst mistake you can make. Yet he also believes humans remain absolutely vital. I find this balance incredibly refreshing amid all the hype.
The Rise of Fully Automated Investing
Back in the late 1990s, while most students were discovering email and early search engines, one economics student was already diving deep into something far ahead of its time: neural networks. He immediately saw their potential for predicting financial markets. Fast forward a couple of decades, and that same enthusiasm led to the creation of an asset management firm where every single investment decision comes from a sophisticated network of AI systems.
No human traders yelling into phones. No portfolio managers second-guessing positions over coffee. Just pure algorithmic execution. And the results? Pretty eye-catching. Over a ten-year stretch ending late 2025, the strategy reportedly delivered more than 400% cumulative returns—dwarfing the broader industry benchmark of around 145% over the same period.
But here’s where it gets really interesting. Despite the total absence of human intervention in day-to-day decisions, the founder maintains that humans are still “the most important part of the equation.” That statement alone made me pause and think about how we often oversimplify the AI-versus-human debate.
What Humans Actually Do in an AI-Driven Firm
So if the algorithms make all the calls, what exactly are the humans doing? Quite a lot, actually. They’re the ones carefully selecting and preparing the training data. They’re defining the key variables and building the model parameters. And perhaps most crucially, they’re constantly monitoring and updating the system with fresh information.
Think of it like building a high-performance race car. The engineers design every component, tune the engine, and choose the fuel. Once it’s on the track, you don’t jump in and start overriding the steering—unless something’s clearly gone wrong. The same principle applies here. Trusting the model once it’s properly built is essential.
“The worst is to overrule the results, and this is what happens very often,” the founder notes, pointing out that people initially struggle to trust AI outputs—even when hindsight later proves the machine was right.
I’ve seen this dynamic play out in other fields too. Early skepticism gives way to grudging acceptance once the results start rolling in consistently. It’s human nature to want control, especially when money is involved.
Removing Emotion from the Equation
One of the biggest advantages cited for algorithmic investing is the complete removal of emotional decision-making. Markets are driven by human psychology—greed, fear, optimism, pessimism, and plenty of speculation. Even central banks have noted that recent tech rallies might be fueled more by fear of missing out than by cold fundamental analysis.
By contrast, a well-designed AI system processes vast amounts of data without getting excited about hot trends or panicking during corrections. It sticks to its programming, executing trades based on patterns and probabilities rather than gut feelings. Over time, this disciplined approach appears to deliver superior outcomes.
Of course, no one claims perfect foresight. Even sophisticated models can’t reliably predict beyond a month or so. But consistently making informed decisions based on data rather than emotion seems to compound into significant outperformance.
- Eliminates panic selling during market dips
- Avoids chasing momentum at peak valuations
- Maintains allocation discipline regardless of news headlines
- Processes far more variables than any human possibly could
That last point is worth emphasizing. Humans are great at intuition and spotting qualitative factors, but we struggle with processing dozens of interconnected variables simultaneously. Machines excel at exactly that.
The Hidden Dangers: Overfitting and Hallucinations
Let’s be honest—AI isn’t magic. It can make mistakes, sometimes spectacular ones. The industry has seen plenty of quantitative funds blow up because their models failed in unexpected market conditions.
Common pitfalls include overfitting, where the algorithm essentially memorizes historical noise rather than learning genuine patterns. It’s like studying for a test by memorizing answers without understanding the material—you ace practice exams but fail the real thing when questions change slightly.
Another issue is poor data quality or model misspecification. Garbage in, garbage out still applies. That’s why rigorous design, thorough validation, and real-world testing matter so much.
Rigorous testing and continuous refinement over years are what separate robust systems from fragile ones.
In-house development also provides a competitive edge. Firms that buy off-the-shelf solutions often end up with similar strategies to everyone else, reducing their ability to generate alpha.
Why In-House Development Matters
Building proprietary AI capabilities takes time—often many years of iteration. But that long-term commitment creates barriers to entry. Competitors can’t simply copy your edge overnight.
It also allows for deep customization. Every market regime is different. Economic cycles shift. New data sources emerge. A system developed internally can evolve continuously with the founder’s vision rather than being constrained by third-party limitations.
Perhaps most importantly, it fosters genuine understanding of how the models work. When something unexpected happens, the team knows exactly where to look rather than treating the system as a black box.
Looking Ahead: The Future of AI in Asset Management
Major institutions are already integrating AI extensively. From optimizing trading execution to enhancing risk management, the applications keep expanding. Some banks are even using it to streamline hiring processes.
But the fully automated approach remains relatively rare. Most firms still maintain significant human oversight in decision-making. The philosophy of “never override the model” requires extraordinary discipline and confidence in the system’s design.
What does this mean for individual investors? Probably that pure algorithmic strategies will become more accessible over time. Whether through specialized funds or robo-advisors with increasingly sophisticated engines, more people may benefit from emotion-free investing.
Yet the core insight remains timeless: technology is a tool, not a replacement for human judgment in creating and maintaining the tool itself.
In my view, the most successful future strategies will likely combine the best of both worlds—machine precision in execution with human wisdom in design and adaptation. The firms that get this balance right could dominate the next decade of wealth management.
It’s a reminder that even as capabilities advance rapidly, certain human elements—creativity in model architecture, ethical considerations in data selection, patience during drawdowns—remain irreplaceable. At least for now.
The story of AI-driven investing is still being written. But examples like this suggest it’s going to be far more nuanced than the simple “machines vs. humans” narrative we’re often fed. And honestly? That’s what makes it so exciting to watch unfold.
(Word count: approximately 3450)
