False Findings in Finance: The Hidden Costs of Misleading Results in the Age of AI

In finance and economics, where empirical research underpins investment strategies, regulatory design, and policy decisions, false findings have material consequences. This article investigates the mechanisms that lead to misleading results in financial research, from flawed assumptions to misaligned incentives. At the heart of the issue lies a fundamental methodological error: the widespread assumption that financial time series are both stationary and a-Holder continuous. The case studies explored in this article, spanning AI-driven models and advanced quantitative techniques, rely on statistical tools built for stable environments. Yet, financial markets are inherently non-stationary, shaped by regime shifts, stochastic volatility, and structural breaks. Ignoring this reality leads not only to invalid inference but to costly misallocations in practice. Bloch argues for a foundational shift in financial research: one that embraces models tailored for non-stationary dynamics and confronts the epistemological limits of prediction in complex systems.