Time-Weighted Nonnegative Bridge Index-Tracking Model and Its Application

With the continuous development of fintech, there has been ongoing improvement in the methods used for compiling stock indexes. Index tracking, which involves constructing a suitable portfolio to achieve a similar return as the target index, has become a crucial skill for investors. This paper introduces a high-dimensional sparse model with nonnegative coefficient constraints. To account for the impact of time on exponential tracking, a time-weighted nonnegative bridge exponential tracking model is proposed. The model exhibits asymptotic consistency of estimation and variable selection under specific conditions. The solution to the model is obtained using the local group coordinate descent method. Empirical results demonstrate that the time-weighted nonnegative bridge index tracking model yields smaller out-of-sample tracking errors. Furthermore, the time-weighted approach outperforms the non-time-weighted approach in terms of the obtained results.