摘要
自适应Lasso回归算法是近年来统计选元的一个新兴方法,具备良好的统计性质。在当今社会资产数量众多的金融投资市场中,资产选择的传统方法Markowitz均值方差模型虽然简单,但不稳定,且容易产生空头头寸。自适应Lasso算法基于变量选择的基本概念,针对资产组合构建而提出,以指数跟踪为目的,构造复制效果良好的稀疏股票投资组合,并进一步对指数的未来趋势做出预测。文章以中国深沪300指数的指数跟踪为例进行分析,结果表明自适应Lasso算法在资产选择和预测中都有良好的效果。
Adaptive Lasso regression algorithm is a new method for statistical variable selection in recent years with good statistical properties. For a large number of investment markets today, traditional Markowitz Mean-variance Model of asset selection method, although simple, is instable and prone to bear position. Adaptive LASSO algorithm based on the basic concept of variable selection is proposed here for portfolio construction and exponential tracking purposes to construct a sparse stock portfolio with good replication effect, and the trend of the index is predicted further. The paper takes the index tracking of CSI 300 index as an example to make an analysis, which indicates good performance of adaptive LASSO method both in stock selection and forecasting.
引文
[1]Markowitz H M.Portfolio Selection[J].Journal of Finance,1952,7(1).
[2]Larsen G A,Resnick B C.Empirical Insights on Indexing[J].The Jour-nal of Portfolio Management,1998,25(1).
[3]Bamberg G,Wagner N.Equity Index Replication With Standard andRobust Regression Estimators[J].Operations Research Spectrum,2000,22(4).
[4]Zorin A,Borisov A.Traditional and Index Tracking Methods for Port-folio Construction by Means of Neural Networks[J].Information Tech-nology and Management Science,2002,3(8).
[5]李俭富,马永开.基于证券价格时间序列的协整优化指数跟踪方法研究[J].系统工程理论与实践,2006,26(11).
[6]刘睿智,杜溦.基于LASSO变量选择方法的投资组合及实证分析[J].经济问题,2012,(9).
[7]Breiman L.Better Subset Regression Using the Nonnegative Garrote[J].Technometrics,1995,37(4).
[8]Tibshirani R J.Regression Shrinkage and Selection via the Lasso[J].Journal of the Royal Statistical Society,1996,58(1).
[9]Wu L,Yang Y H,Liu H Z.Nonnegative-Lasso and Application in In-dex Tracking[J].Computational Statistics and Data Analysis,2014,70(2).
[10]Zou H.The Adaptive Lasso and Its Oracle Properties[J].Journal ofthe American Statistical Association,2006,101(476).
[11]Sha F,Lin Y.Multiplicative Updates for Nonnegative Quadratic Pro-gramming[J].Neural Computation,2007,19(8).
[12]Wu L,Yang Y H.Nonnegative Elastic Net and Application in IndexTracking[J].Applied Mathematics and Computation,2014,227(C).