AI Won’t Make You a Good Trader, But Here’s How the Pros Use It Anyway

• Charles Edwards and Julio Moreno call AI an accelerant for serious research, compressing multi-hour analysis tasks into minutes using live market data feeds.
• Michael van de Poppe built a working crypto portfolio in five minutes using a chatbot and free APIs, yet it lacked macroeconomic context and asset correlation discipline.
• Professional hedge funds treat AI as infrastructure requiring continuous human oversight, not as a standalone trading system or predictive oracle.

• 5 minutes Time to build a functional crypto portfolio dashboard using AI and free data feeds
• 4 analysts Leading independent and institutional researchers unified on AI’s limits and actual use cases
• 100s Number of metrics and data sources combined in professional quantitative trading models

Artificial intelligence is reshaping how professional traders and analysts approach crypto markets, yet not in the way retail investors might expect.

Speaking on separate industry panels, four prominent figures, Charles Edwards of Capriole Investments, Julio Moreno of CryptoQuant, Benjamin Cowen, and Michael van de Poppe, delivered a unified message: AI amplifies existing skill and accelerates routine work, but it does not substitute for market judgment or deliver returns to untrained traders.

The consensus matters because it reflects how institutional capital is actually deploying machine learning and language models across trading desks and research teams, separating practical gains from hype.

Edwards and Moreno position AI as research acceleration, not market prediction

Charles Edwards, who founded Capriole Investments as a quantitative Bitcoin hedge fund, framed AI’s value narrowly: it shifts competitive advantage toward professionals who invest time in rigorous model development. “I think AI as well is making that playing field more opportunistic for certain people,” he said, emphasizing that the tool rewards effort, not luck.

The tool sets to do that are much more powerful and it can be done more quickly today with AI.

Charles Edwards, Capriole Investments

Edwards pointed to speed gains in routine analysis as the clearest institutional benefit. Tasks that once consumed hours, scanning on-chain data, comparing market signals, or stress-testing portfolio assumptions, now run in minutes. Julio Moreno, who heads research at CryptoQuant, echoed this from the vantage of a data platform serving institutional clients.

Institutions trust AI-generated insights, he noted, but verify them continuously and monitor data relevance over time. That verification step is not optional; it is part of how professional funds operate.

The implication is that AI adoption is becoming table stakes for serious research shops, but only as a component within larger, tested frameworks.

Van de Poppe’s five-minute portfolio exposes the gap between speed and judgment

Michael van de Poppe, an independent analyst with broad institutional following, demonstrated the practical limits of AI with a tangible example. He used Claude, an AI chatbot, and free data APIs to construct a working crypto portfolio and performance dashboard in five minutes.

The exercise was genuine: the portfolio functioned, pulled live data, and scored assets on technical, fundamental, and on-chain flow metrics.

The portfolio’s flaws, however, revealed why speed does not equal skill. Van de Poppe noted immediately that the AI agent created a basket of cryptos without addressing correlation or macroeconomic positioning. “It didn’t create a basket of uncorrelated cryptos.

It doesn’t have any macros in there,” he explained. The portfolio would perform mechanically but lacked the discipline institutional managers apply to asset selection and risk construction. Human judgment, understanding macro conditions, market structure, and portfolio theory, remained essential to the work.

Van de Poppe cautioned against viewing AI tools as autonomous systems: they will not deliver “some sort of magic that creates an infinite money loop.” That warning aligns with industry skepticism toward fully automated trading bots, which few professional funds endorse as standalone strategies. The tool accelerates labor; it does not replace decision-making.

Professional funds treat AI as tested infrastructure, not as crystal ball

How do institutions actually deploy these tools? Edwards described the architecture at Capriole: hundreds of metrics and data sources fed into comprehensive models that combine on-chain technicals and macroeconomic data developed over years. That is infrastructure, not innovation.

The firm built models first, then applied AI to run them faster and test more scenarios in parallel. The AI layer automates execution of established logic, not the logic itself.

This approach mirrors how professional traders have always worked: build a thesis, backtest rigorously, then scale execution when conditions align.

Moreno’s observation about institutional trust-but-verify discipline reinforces the pattern. Large asset managers deploying CryptoQuant data or similar platforms will ingest AI-ranked signals, but they maintain independent control and continuously audit whether those signals remain predictive.

Market regimes shift, correlations break, and data quality degrades; treating any analytical output as static truth is a path to losses. AI is a worker, not a decision-maker.

The four analysts’ consistency on this point reflects a broader institutional reality: adoption of AI in crypto trading is accelerating, but within guardrails. Hedge funds and trading desks are deploying language models and machine learning to compress research timelines and broaden scenario testing. They are not replacing governance, model validation, or human oversight.

That distinction separates genuine institutional adoption from the speculative retail expectation that new technology will magically generate alpha.

What institutional investors should monitor about AI trading tools

For allocators evaluating crypto-focused funds or trading strategies, the AI question is becoming routine due diligence. A manager’s ability to explain how they validate AI-generated signals, how often they retrain models, and how they handle regime shifts will separate credible operators from those chasing novelty.

Capriole’s multi-year model development cycle and CryptoQuant’s emphasis on continuous monitoring are institutional standards; shops without such discipline are taking unmonitored risks.

The speed advantage is real but temporary. As AI tools democratize, as evidenced by van de Poppe building a portfolio in five minutes with free tools, the edge shifts from access to execution quality. That quality comes from data cleanliness, rigorous backtesting, and human judgment applied to edge cases and market breaks.

Institutions with established research infrastructure and capital to invest in model refinement will capture the gains; those hoping AI will substitute for that work will underperform.

The next test will arrive as crypto market conditions stress-test the models now entering production. Institutional funds are currently monitoring whether AI-assisted strategies retain predictive power during the next major correction or volatility spike. That real-world validation will clarify which firms understand AI as a tool within disciplined investment processes and which treated it as a shortcut. Watch for disclosures from quantitative hedge funds and trading platforms about model performance during drawdowns, that will reveal whether AI adoption is truly institutional or merely cosmetic.

Original reporting: beincrypto.com