Start main page content

Events

From Chaos to Quantum: Modelling Commodity Markets with Deep Learning, Nonlinear Systems, and Cyber

When: Thursday, 24 April 2025 - Thursday, 24 April 2025
Where: Hybrid Event
Braamfontein Campus West
New Commerce Building (NCB) 221
Start time:13:00
Enquiries:

Mamosa Moletsane

Please join our next BBL Seminar, presented by Prof. Aristeidis Samitas, Professor of Finance at the National and Kapodistrian University of Athens/ Greece.

Join the meeting now

Abstract

This study introduces a pioneering framework for forecasting commodity prices by integrating cybersecurity-driven macroeconomic indicators with advanced Machine Learning (ML), Deep Learning (DL), and Quantum Neural Network (QNN) models. Motivated by the rising influence of digital infrastructure and cyber threats on global markets, we construct a novel multi-source dataset combining daily price series for seven key commodities, Copper, Crude Oil, Gold, Natural Gas, Steel Rebar, Live Cattle, and US Coffee, with granular cybersecurity investment and revenue metrics. Our empirical approach evaluates 27 forecasting models, ranging from traditional linear regressions and decision trees to hybrid architectures such as LSTM-Transformer models, Chaotic Neural Networks, and a newly developed Chaotic Quantum Neural Network (CQNN), applied here for the first time in financial forecasting. Results demonstrate that while classical ML models offer competitive accuracy, nonlinear models exhibit significantly greater robustness across volatile series. Notably, the Nonlinear Autoregressive model with Exogenous Inputs (NARX) outperforms all benchmarks with near-perfect R² values, while the CQNN offers compelling results with minimal parameterization, underscoring its promise as a frontier tool in quantum-enhanced finance. Multiple Linear Regression (MLR) is used not only as a baseline but also as a robustness validation framework, affirming the statistical relevance of cybersecurity variables across asset classes. This paper contributes novel empirical insights into the predictive utility of cybersecurity indicators, introduces an original dataset and modelling architecture, and reinforces the critical intersection of digital risk and commodity price dynamics. The findings hold substantial implications for investors, regulators, and policymakers navigating the complex, data-driven contours of modern economic systems.

 

 

calendar iconAdd event to calendar