基于AdaBoost集成算法和Black-Litterman模型的资产配置

Asset allocation based on AdaBoost ensemble algorithm and the Black-Litterman model

行业主题型ETF的迅速发展为投资者带来了多元化的行业资产配置机会,然而投资者应如何进行合理有效的行业资产配置,才能获取行业中优质企业长期健康的发展红利.本文以2005年1月14日至2022年7月15日的沪深300一级行业指数为研究样本,运用自适应提升算法(adaptive boosting,AdaBoost)生成投资者观点,并通过Black-Litterman(BL)模型建立行业资产的最优配置策略.算例结果表明,基于AdaBoost集成算法的BL模型(BL_AdaBoost)相较于市场组合、等权重组合、MV模型能够获得更优的投资绩效.此外,我们进一步考察了BL_AdaBoost在不同时期的绩效表现.结果显示,在极端风险事件冲击以及股市整体持续下行的时期,BL_AdaBoost在一定程度上能够抵御风险,从而减小投资组合的亏损.而在股市正常波动时期,BL_AdaBoost相较于其他投资组合能够获得更高的回报.

The rapid development of industry-themed ETFs has brought investors diversified industry asset allocation opportunities.However,the critical question is how investors can allocate rational and practical industry assets to obtain long-term healthy development dividends from high-quality enterprises in the industry.The CSI 300 primary industry index is taken as the research sample from 14 January 2005 to 15 July 2022.This paper uses the adaptive boosting(AdaBoost)algorithm to generate investor views and establishes the optimal allocation strategy of industry assets through the Black-Litterman(BL)model.The numerical example results show that the BL model based on the AdaBoost ensemble algorithm(BL_AdaBoost)can obtain better investment performance than the market portfolio,equal weight portfolio,and MV model.In addition,we further examine the performance of BL_AdaBoost in different periods.The results show that BL_AdaBoost can resist risks to a certain extent during the shock of extreme risk events and the overall continuous downturn of the stock market,thereby reducing the loss of the investment portfolio.During regular stock market volatility periods,BL_AdaBoost can achieve higher returns than other portfolios.