Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

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

Pierre-Daniel Arsenault, Shengrui Wang, Jean-Marc Patenande

TL;DR

This survey addresses the challenge of applying AI to financial time series forecasting by distinguishing interpretability from explainability and surveying advances in the past five years. It presents a dual taxonomy of interpretable models and explainability methods, organized around feature/time importance, decision rules, and trend analysis, and it catalogs a wide range of techniques from linear models and trees to attention, graphs, fuzzy logic, and hybrid approaches. The paper reviews practical evaluation issues, industry practices, and visual/explainability interfaces, highlighting SHAP and related perturbation methods as dominant explainability tools, while emphasizing the need for rigorous faithfulness and user-contextual explanations. Overall, it provides guidance for selecting appropriate XAI techniques in finance, discusses industry uptake, and identifies key gaps such as standardized interpretability metrics and robust faithfulness assessments. The work underscores the importance of treating interpretability and explainability as distinct but complementary components in responsible, transparent financial AI systems.

Abstract

Artificial Intelligence (AI) models have reached a very significant level of accuracy. While their superior performance offers considerable benefits, their inherent complexity often decreases human trust, which slows their application in high-risk decision-making domains, such as finance. The field of eXplainable AI (XAI) seeks to bridge this gap, aiming to make AI models more understandable. This survey, focusing on published work from the past five years, categorizes XAI approaches that predict financial time series. In this paper, explainability and interpretability are distinguished, emphasizing the need to treat these concepts separately as they are not applied the same way in practice. Through clear definitions, a rigorous taxonomy of XAI approaches, a complementary characterization, and examples of XAI's application in the finance industry, this paper provides a comprehensive view of XAI's current role in finance. It can also serve as a guide for selecting the most appropriate XAI approach for future applications.

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

TL;DR

This survey addresses the challenge of applying AI to financial time series forecasting by distinguishing interpretability from explainability and surveying advances in the past five years. It presents a dual taxonomy of interpretable models and explainability methods, organized around feature/time importance, decision rules, and trend analysis, and it catalogs a wide range of techniques from linear models and trees to attention, graphs, fuzzy logic, and hybrid approaches. The paper reviews practical evaluation issues, industry practices, and visual/explainability interfaces, highlighting SHAP and related perturbation methods as dominant explainability tools, while emphasizing the need for rigorous faithfulness and user-contextual explanations. Overall, it provides guidance for selecting appropriate XAI techniques in finance, discusses industry uptake, and identifies key gaps such as standardized interpretability metrics and robust faithfulness assessments. The work underscores the importance of treating interpretability and explainability as distinct but complementary components in responsible, transparent financial AI systems.

Abstract

Artificial Intelligence (AI) models have reached a very significant level of accuracy. While their superior performance offers considerable benefits, their inherent complexity often decreases human trust, which slows their application in high-risk decision-making domains, such as finance. The field of eXplainable AI (XAI) seeks to bridge this gap, aiming to make AI models more understandable. This survey, focusing on published work from the past five years, categorizes XAI approaches that predict financial time series. In this paper, explainability and interpretability are distinguished, emphasizing the need to treat these concepts separately as they are not applied the same way in practice. Through clear definitions, a rigorous taxonomy of XAI approaches, a complementary characterization, and examples of XAI's application in the finance industry, this paper provides a comprehensive view of XAI's current role in finance. It can also serve as a guide for selecting the most appropriate XAI approach for future applications.
Paper Structure (31 sections, 2 equations, 6 figures, 2 tables)

This paper contains 31 sections, 2 equations, 6 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Definitions
  3. Methodology
  4. Taxonomy
  5. Interpretable Models
  6. Feature and Time Importance
  7. Linear Regressions
  8. Decision Trees
  9. Attention Mechanism
  10. Fuzzy Logic
  11. Graph
  12. Mask
  13. Greedy Algorithm
  14. Hybrid Techniques
  15. Decision Rules
  16. ...and 16 more sections

Figures (6)

  • Figure 1: Classification of XAI Approaches
  • Figure 2: Architecture of AMS
  • Figure 3: Architecture of AT-LSTM
  • Figure 4: Architecture of CTV-TabNet
  • Figure 5: Architecture of the Model in yun_interpretable_2023
  • ...and 1 more figures