Empirical Research of Energy Sector Based on Principal Component Analysis
Download as PDF
DOI: 10.23977/mcee2020.038
Author(s)
Haoran Sun, Jie Li, Feiyu Long
Corresponding Author
Jie Li
ABSTRACT
In this paper, 76 stocks in the energy sector were selected as sample stocks and data from 2013 to 2017 were used for single factor test to select pre-selected factors, factor orthogonality was carried out through principal component analysis, return prediction was made by principal component factors, and quadratic optimization considering risks was introduced to study how the multi-factor model could guide investors' asset allocation in the energy sector. In the empirical, without introducing the future data and considering new listings, the empirical parameters were obtained by solving the model with in-sample data to realize the weight proportion of the selected stocks. It obtained an annual yield of 46% out of the sample with a high winning rate, and beat the HS300 index with the energy sector with a low valuation, highlighting the weight matching ability of the strategy and verifying the effectiveness of the model. This model uses principal component analysis to reduce the noise of factor information, and combines risk analysis to optimize the feasibility of the strategy. It provides a new perspective for multi-factor quantitative investment.
KEYWORDS
Single factor test, principal component analysis, return and risk prediction, multi-factor model