结合聚类和时间加权非负岭回归指数追踪模型

Combining Clustering and Time-Weighted Non-Negative Ridge Regression Index Tracking Models

指数追踪是一种特殊的被动投资管理形式,其目的在于从一个目标指数包含的所有成分股中选择一部分成分股来建立股票投资组合,用其来追踪目标指数,尽可能的得到一个接近股票市场的累计回报。本文将变量聚类与经典统计模型岭回归相结合,提出聚类和时间加权非负岭回归指数追踪模型,该模型首先使用变量聚类法选择部分股票构建投资组合,然后考虑到时间因素对指数追踪的影响,以及股票市场不允许卖空约束,在岭回归的基础上引入了时间加权函数和对权重的非负约束。为了验证所提出模型的性能,本文将聚类和时间加权非负岭回归指数追踪模型与现有模型进行比较。实验结果表明,使用聚类和时间加权非负岭回归指数追踪模型可以得到较小的追踪误差,所构造的投资组合具有较好的追踪效果。

Index tracking is a special form of passive investment management, which aims to select a subset of all constituents of a target index to build a stock portfolio, and use it to track the target index to obtain a cumulative return as close to the stock market as possible. This paper com-bines variable clustering with the classical statistical model ridge regression, and proposes a clustering and time-weighted non-negative ridge regression index tracking model, which first uses the variable clustering method to select some stocks to construct a portfolio, and then considers the impact of time factors on index tracking and the stock market does not allow short selling constraints, and introduces a time-weighted function and a non-negative constraint on weights on the basis of ridge regression. To verify the performance of the proposed model, this paper compares the clustering and time-weighted non-negative ridge regression exponential tracking model with the existing model. Experimental results show that the tracking error can be obtained by using clustering and time-weighted non-negative ridge regression index tracking model, and the constructed portfolio has better tracking effect.