A Survey of Explainable Artificial Intelligence (XAI) in Financial Time Series Forecasting

Explainability and interpretability are distinct concepts that require different approaches: The survey emphasizes that these terms are often used interchangeably but should be treated separately in practice, suggesting that choosing the right XAI method depends on understanding which concept—making AI decisions understandable or making the underlying logic transparent—is most important for your specific financial application.

AI accuracy alone is insufficient for high-stakes financial decisions: While AI models have achieved impressive accuracy levels, their complexity creates a "trust gap" that slows adoption in finance. This highlights that responsible AI deployment in finance requires not just better predictions, but better explanations of how those predictions are made.

A taxonomy of XAI approaches now exists for financial forecasting: The survey provides a structured categorization of XAI methods specifically designed for financial time series prediction, offering practitioners a practical guide for selecting appropriate explainability techniques rather than applying generic solutions.

XAI is actively being applied in the finance industry: The paper documents real-world examples of XAI implementation in finance, indicating that explainability is moving beyond academic research into practical use—suggesting it's becoming a standard consideration for responsible AI deployment in this sector.