基于LASSO变量选择方法的投资组合及实证分析

Portfolio Construction Using Variable Selection:Based on LASSO Method

由于传统的Markowitz均值方差模型具有很强的不稳定性,且容易产生较大的空头头寸,因此,资产选择成为构建投资组合的一个研究热点。针对以风险分散和指数跟踪为目的的资产组合构建提出了基于变量选择观点的LASSO选择方法,说明了LASSO方法在资产选择中的优势,并应用LASSO方法使用2010年3月到2012年2月的日收益率数据对沪深300指数的指数跟踪做了实证检验,结果表明LASSO方法在组合构建和预测中都有很好的效果。

As the instability of traditional Markowitz Mean - variance Model and its weakness of bringing big short position easily, stock selection in portfolio construction becomes a hot spot in researching. This paper focuses on LASSO method utilized in stock selection for risk diversification and index tracing in perspective view of variable selection. Four merits of LASSO in stock selection are stated and an application using real data from March 2010 to February 2012 in SHSZ300 index tracing via LASSO is made. Comparison indicates good performance of LASSO method both in portfolio construction and forecasting.