EURO 2025 Leeds
Abstract Submission

3065. A multi-target radial basis function (RBF) network with variable shape parameters for return predictions from generative South African ESG indexes

Invited abstract in session TD-23: Data Analytics for Business Resilience and Sustainability - Leveraging ML Models, stream OR for Societal Development.

Tuesday, 14:30-16:00
Room: Esther Simpson 3.01

Authors (first author is the speaker)

1. Nina Kajiji
Computer Science and Statistics, University of Rhode Island, and The NKD Group, Inc.
2. Gordon Dash
Finance and Decision Sciences, University of Rhode Island
3. Helper Zhou
School of Accounting, Economics and Finance, University of KwaZulu Natal
4. Bruno Kamdem
Department of Business Management, SUNY Farmingdale State College, School of Business

Abstract

A company’s commitment to environmental, social, and governance (ESG) principles has sparked interest in developing firm-level ESG scores, particularly in a VUCA world—characterized by Volatility, Uncertainty, Complexity, and Ambiguity. These challenges create rapid changes, unpredictable scenarios, and intricate interdependencies, influencing stock prices and complicating decision-making. This study aims to establish an innovative algorithmic framework to identify dynamic ESG factors that assess sustainability risk in securities traded on the Johannesburg Stock Exchange (JSE). The proposed methodology employs an alternative data approach, using asset returns in exploratory factor analysis to derive high-validity factor scores, forming JSE-specific sustainability factors. The performance of these factors is evaluated using a multi-target multi-objective radial basis function neural network (RBFN). A Bayesian-enhanced regularized RBFN, with a variable shape parameter strategy, is applied to analyze complex valuation signals. Utilizing explainable AI (XAI), findings highlight the newly developed S-factor as the key predictor of next-period returns, followed by the E-factor, the SA volatility index, and a precious metals factor. XAI demonstrates that these factors effectively capture the unique ESG priorities of South African firms, offering meaningful insights into the impact of sustainability metrics on stock performance.

Keywords

Status: accepted

Back to the list of papers