Research On Active Fund Investment Strategies Based On Fund Manager Evaluation Index

With the optimization of residents’ asset structure and the increasing scale of their investable assets,China’s capital management industry has a bright future.2006-2021,the total scale of investable assets held by Chinese individuals grew rapidly,from 25.6trillion yuan to 268 trillion yuan.Among all types of investors,professional institutional investors have a significant advantage in terms of profitability.Since the promulgation of the new regulations on asset management in 2018,its core regulatory ideas of netting,de-channeling,deleveraging and breaking the rigid exchange rate have allowed public funds to stand out among a host of asset management industries by virtue of their standardization and high transparency of net worth.Therefore,under the general trend of public fund explosion and institutionalization of investment,it is especially important to combine machine learning to realize fund screening and asset allocation,and to include the assessment of managers in its performance evaluation system.This thesis takes into account the market trends,considers the industry-themed nature of funds for the first time,adds the examination of fund managers’ counterpart professional background to fund manager evaluation indexes,forms a comprehensive fund performance evaluation system,and optimizes the subjective view vector of the Black-Litterman model through integrated learning,combines three other asset allocation strategies,and makes an all-round comparison with the general fund performance evaluation system.Exploring the optimal investment strategy,briefly described as follows.(1)In the evaluation system design,fund manager evaluation indicators are introduced and quantified through the perspectives of personal characteristics of fund managers and characteristics of funds under management to form a new comprehensive fund performance evaluation system,considering that fund managers with relevant professional endorsement have advantages in industry industry chain and professional knowledge,so a new counterpart professional background indicator is set up,taking Wind industry theme as the standard,and the fund manager’s studied The professions are classified and correspond to each other.(2)In fund screening,Xgboost,Catboost,and Light GBM are applied to two different evaluation systems simultaneously for training,referencing,and forecasting,and then the corresponding learning method with the highest prediction accuracy under the two systems is selected,and the high-quality industry-themed active funds in each quarter of the backtest interval are obtained under this learning method.(3)In asset allocation,the Black-Litterman model optimized by integrated learning is proved,modeled,parameterized,and conditioned,and the data is back-tested with the equal weight model,risk parity model,and minimum variance model in the2020-2022 interval,and the comparison between strategies and benchmarks is performed by annualized return,annualized volatility,and other indicators,and finally the performance is visualization.The empirical results show that the counterpart professional background is the positive indicator with the highest Rank＿IC value in the factor effectiveness analysis and has a significant positive impact on fund performance.The fund’s comprehensive performance evaluation system with the addition of fund manager evaluation indicators has good optimization of all indicators in the final strategy backtest compared to the normal performance evaluation system,with particularly significant improvement in annualized return,and the Black-Litterman model optimized by integrated learning outperforms all other strategies.In particular,the fund performance portfolio of the integrated performance evaluation system under the optimized Black-Litterman model demonstrates superior retracement control during market pullbacks.