Predicting BRICS NIFTY50 returns using XAI and S.A.F.E AI lens

Area of Science:

• Quantitative Finance
• Econometrics
• Machine Learning

Background:

• Global fund managers diversify portfolios internationally to enhance returns and manage risk.
• International investments expose investors to country-specific volatility, macroeconomic shocks, and exchange rate fluctuations, complicating return forecasting.
• The Goldman Sachs BRICs Nifty 50 Developed Markets Index (BRICS NIFTY 50) represents a complex global financial asset.

Purpose of the Study:

• To develop and present a novel forecasting framework for predicting returns of the BRICS NIFTY 50 index.
• To identify significant country-specific explanatory variables influencing global fund returns.
• To aid global fund managers and investors in making informed investment decisions.

Main Methods:

• Gradient Boosting Regression (GBR) and SHAP-based Explainable AI (XAI) were employed to identify key country-specific predictors.
• Six machine learning models (GBR, CatBoost, Light Gradient Boosting Machine (LGBM), Extreme Gradient Boosting (XGBoost), Random Forest (RF), and Extra Tree Regressor (ETR)) were applied for return forecasting.
• The S.A.F.E AI framework and a Multi-Criteria Decision-Making (MCDM) framework were utilized for evaluating model performance, predictive accuracy, sustainability, and predictor contributions.

Main Results:

• Country-specific market volatility, industrial performance, financial sector development, and exchange rate fluctuations were identified as significant drivers of global returns.
• Explanatory factors originating from India, China, and Brazil were found to be particularly influential on the BRICS NIFTY 50 returns.
• The study successfully developed a two-step forecasting framework with robust evaluation metrics.

Conclusions:

• The developed forecasting framework offers practical utility for global fund managers and investors.
• The findings provide insights for policymakers regarding factors influencing foreign direct and portfolio investments.
• The methodology highlights the importance of country-specific factors in global portfolio return prediction.