In 2024, the integration of artificial intelligence into the investment process evolved from a trend into the new normal—becoming a subject of increasing regulatory scrutiny. In both the U.S. and EU, discussions are underway around transparency and certification standards for AI-driven trading systems.
International financial expert and specialist in algorithmic trading and systematic strategies, Maksim Baradziuk, believes 2025 will be a pivotal year. He anticipates the emergence of specialized licenses for AI traders, as well as stricter requirements for algorithm testing before market deployment. Holding a Specialist in Investment Management qualification from the University of California, Los Angeles, Maksim applies cutting-edge analytical techniques to integrate AI into the development and testing of financial strategies.
In an interview with The AI Journal, he discusses emerging markets at the intersection of AI and blockchain analytics, explains how to implement AI in trading to enhance strategic thinking—not replace it—and shares his outlook for 2025.
Maksim, you’re an expert in algorithmic trading and building quant strategies. How do quant strategies differ from traditional financial approaches, and which ones are most relevant today?
In traditional trading, decisions to buy or sell are made by humans based on experience, intuition, and visual chart analysis. In the quant approach, the entire process—from signal generation to trade execution—is formalized into an algorithm that operates based on predefined rules.
With the rapid advancement of new AI models, many of today’s emerging strategies are built around them. We’re currently seeing fast growth in strategies that analyze blockchain data using pattern recognition techniques. One promising area is tracking and matching insider wallets to detect early signals before large asset movements. Another is identifying patterns in meme coins to monetize them—from detecting the early “warming up” phase of a specific token to predicting volatility based on behavioral metrics.
Overall, new markets are forming around meme coins and DATs. These markets are infrastructure-heavy and not easily accessible to retail investors, which opens the door for data-driven algorithmic strategies with strong analytical foundations.
Looking ahead, 2025 is shaping up to be extremely volatile, so trend-following traders with high-quality strategies have a real opportunity to shine. At the same time, uncertainty around interest rates and the new U.S. administration could introduce additional market chaos.
You earned your Specialist in Investment Management qualification from UCLA and have deep knowledge of global investment standards. Based on your education and experience, which AI tools have proven most effective for you in practice?
There’s a wide array of AI tools on the market today—from highly specialized models for time series forecasting to general-purpose language models that can be adapted for financial R&D.
In theory, AI can be used for virtually everything—from automating reporting to autonomous trading. In practice, though, success hinges on how well the problem is defined and how effectively the algorithms are implemented. That’s where the real value of human expertise comes in: knowing how to integrate AI into actual trading workflows.
Personally, I don’t rely on AI for the final execution of trading decisions. That step still requires non–black-box algorithms grounded in human experience, intuition, and a deep understanding of market context.
You developed risk management protocols that reduced maximum drawdown by 40% compared to industry benchmarks. What tools or methodologies did you use to achieve that?
The core principle was maximizing “time in the market.” At the same time, we implemented dynamic position reversals—a feature common to most trading systems. Our goal was to keep capital deployed in the market as long as possible without unnecessary exits, while also actively shifting between long and short positions depending on market conditions.
These two tactics are typically seen as contradictory, so combining them was a fairly unconventional approach. To measure performance, we used simulations based on synthetic data, which are surprisingly effective at modeling the model’s robustness. We also validated results with real market data to ensure consistency and repeatability.
You’ve led successful trading and investment projects across multiple countries, thanks to your ability to quickly adapt to different regulatory and market environments. Looking ahead, how do you expect AI tools in trading to be regulated?
I think AI regulation in trading will follow a path similar to that of traditional algorithmic trading—but with added complexity due to the “black box” nature of most AI systems. Regulators will likely struggle to create clear frameworks for such opaque models.
I expect stricter risk control and model testing requirements—such as mandatory stress-testing on both synthetic and historical data, and resilience checks against manipulation or false signals. It’s also possible we’ll see the introduction of specialized licenses or certifications for companies using AI in trading, especially on regulated exchanges. This would raise the barrier to entry but also help reduce systemic risk and protect investors.
Ultimately, regulation will aim for balance: fostering innovation without letting unregulated algorithms become a source of market instability.
