Аннотация
Direct linear regression prediction of index returns has long been recognized as a challenging task in industry time-series timing research. Constrained by the extremely low signal-to-noise ratio and pervasive nonlinear characteristics of financial data, ordinary least squares (OLS) regression suffers from poor out-of-sample performance and hardly outperforms the historical average benchmark. Although academia has proposed a nonlinear prediction framework based on sign-magnitude decomposition and Copula function dependence coupling, the selection and estimation of Copula families introduce substantial model uncertainty in practical multi-factor timing implementations. Against this backdrop, this study draws on the core framework published in the Journal of Banking and Finance, constructing a concise, logically consistent nonlinear return prediction framework by conditioning return signs (ups and downs directions) on contemporaneous magnitude (volatility states), while eliminating complex Copula dependence modeling. Based on the Conditioning Sign on Magnitude (CSM) method, this paper elaborates the model construction logic, econometric advantages, and empirical performance. Empirical tests based on 74-year monthly excess return data of the S&P 500 index verify that the CSM framework achieves superior out-of-sample statistical accuracy and economic utility with low implementation costs. It effectively overcomes the inherent limitations of linear models and the parameter instability of Copula-based methods, providing a lightweight and efficient optimization scheme for medium- and low-frequency time-series quantitative timing.