Artificial Intelligence in Equity and Crypto Markets: Progress, Profitability Evidence, and the Limits of Automated Investing
Why AI's investment breakthroughs don't always translate to real profits
AI systems show genuine progress at predicting markets and processing financial news, but almost none consistently make money after accounting for trading costs and real-world constraints. The gap between what AI can predict and what actually turns a profit is larger and more persistent than most published studies suggest.
Investors and firms betting on AI-driven trading strategies need honest accounting of what works. The paper exposes common statistical traps—like testing strategies on old data or cherry-picking winners—that make mediocre systems look brilliant. Without addressing these flaws, money will keep flowing to strategies that underperform, while genuinely profitable AI approaches remain harder to identify.