Recently, Professor Robert Anderson, Chair Professor of the Institute for Advanced Study in Mathematics at Harbin Institute of Technology (HIT), in collaboration with Professor Baeho Kim from Korea University and Professor Dean Ryu from the Instituto Tecnológico Autónomo de México (ITAM), has achieved a landmark breakthrough in operations research. The team proposed the Long-History Principal Component Analysis (LH-PCA) method, which successfully breaks through the core bottleneck in the risk management of high-dimensional financial portfolios. The research findings, entitled Long-History Principal Component Analysis in a Dynamic Factor Model with Weak Loadings, have been published in Operations Research, the top-tier journal in the fields of management science and operations research.
Traditional risk models rely on short-term data analysis, which produces significant errors in weak-factor and high-dimensional scenarios, and tends to cause severe underestimation of portfolio risk. To address this problem, this study extends principal component analysis to 6-year long-cycle data, adapts to time-varying factor loadings and dynamic market structures, theoretically proves the consistency of the proposed method, and effectively alleviates the bias in factor loading estimation.
The study constructs observable second-order risk measurement metrics. Through empirical analysis based on 20 years of data from US and European stock markets and large-scale simulation verification, it fully confirms that LH-PCA can drastically reduce risk prediction errors and significantly improve the robustness of the global minimum variance portfolio compared with traditional methods. The research results not only provide an efficient tool for risk control of large-scale asset portfolios, volatility forecasting and quantitative investment strategy optimization, but also deliver a new solution for the financial risk control field, with important theoretical value and practical significance.
Harbin Institute of Technology is the first affiliated institution of the paper.
Article link:
https://pubsonline.informs.org/doi/10.1287/opre.2024.1134
