Paper Type

Complete

Paper Number

PACIS2026-2095

Description

Stock market Decision Support Systems often lose predictive reliability during crises characterized by structural breaks and extreme volatility. Although prior reviews examine Artificial Intelligence (AI) based forecasting, crisis regimes, and financial applications separately, an integrated synthesis explaining how multimodal data fusion supports forecasting across diverse geographies remains limited. Addressing this gap, this study conducts a design-oriented systematic literature review of 72 peer-reviewed studies to examine how fusion-driven AI can achieve resilience across the data divide in developed and emerging markets. Findings reveal a strong focus on predictive artifacts that integrate market, news, and sentiment signals through hybrid architectures. The study applies a design-oriented framework and introduces three design principles that synthesize insights across the data divide, using fusion to guide resilient forecasting systems for both developed and emerging markets. The paper concludes with a research agenda prioritizing data infrastructure innovation, explainable fusion mechanisms, and governance frameworks for resilient stock forecasting.

Comments

01-AIML

Share

COinS
 
Jul 5th, 12:00 AM

Resilience Across the Data Divide: A Design-Oriented Systematic Review of Multimodal Fusion for Developed and Emerging Stock Market Forecasting under Crisis

Stock market Decision Support Systems often lose predictive reliability during crises characterized by structural breaks and extreme volatility. Although prior reviews examine Artificial Intelligence (AI) based forecasting, crisis regimes, and financial applications separately, an integrated synthesis explaining how multimodal data fusion supports forecasting across diverse geographies remains limited. Addressing this gap, this study conducts a design-oriented systematic literature review of 72 peer-reviewed studies to examine how fusion-driven AI can achieve resilience across the data divide in developed and emerging markets. Findings reveal a strong focus on predictive artifacts that integrate market, news, and sentiment signals through hybrid architectures. The study applies a design-oriented framework and introduces three design principles that synthesize insights across the data divide, using fusion to guide resilient forecasting systems for both developed and emerging markets. The paper concludes with a research agenda prioritizing data infrastructure innovation, explainable fusion mechanisms, and governance frameworks for resilient stock forecasting.