Research On The Quantitative Investment Model Of Large Assets Based On Machine Learning

In recent years,with the arrival of China's "new regulations on asset management" and the change of investment environment under the new normal of the economy,there are few single assets that can meet various investment needs in the financial market,and asset allocation has gradually become an ideal investment strategy for institutional investors.In foreign countries,the related research of asset allocation has gone through nearly a hundred years,and obtained a good return on investment in the investment practice.However,the application of China's relevant asset allocation theory research in investment practice is still in its infancy,especially lacking the support of quantitative investment model with strong operability and stable return on investment,so it cannot obtain ideal returns in strategic trading.Based on this,this paper,based on the research of existing theories and methods,takes China's capital market as the main investment target and constructs a quantitative investment model of asset allocation with the help of machine learning algorithm.First combed the domestic and international major categories of asset allocation model,quantitative investment model and the types of asset allocation model based on machine learning research present situation,this paper introduces the basic theory of asset allocation and classic configuration model,according to the problems existing in the existing research,using the python language to build an effective quantitative categories of assets investment model based on machine learning,the model consists of two modules,based on machine learning is a part of the assets or direction prediction model,play advantage screening of assets,raise the yield model;Part of the model is based on the prediction results of fixed proportion investment backtest model,using the asset allocation method to control the overall risk of the model and check the effectiveness of the model through backtest.First,according to the data requirements of the model,the asset representative index and macroeconomic indicators required by the prediction model are selected,as well as the asset investment targets required by the back-test model.Then,six machine learning algorithms including support vector machine,random forest,XGBoost,GBDT,BP and LSTM are introduced to transform the prediction of the rise and fall direction of return on assets into a classification problem in pattern recognition.At home and abroad more than ten kinds of macroeconomic indicators monthly figures as the characteristic variable,stocks,bonds and gold three categories of six kinds of assets at the end of the closing price for the label,the future 1 month return on assets in response to changes in direction factor,learning ability,with the aid of a powerful machine learning data to build characteristic the mapping relationship between variables and labels,the output of six kinds of assets between January 2015 and October 2019 the rise and fall of predicted results.The prediction results of the prediction model are used as the trading signal,and the fixed proportion allocation model in the classic asset allocation model is used to set five capital allocation ratios for back-test transactions.Finally,according to the comparison of backtest results,the best prediction model is BP neural network model,and the optimal ratio of assets is 100/0.Based on BP neural network,the 100 / 0 ratio of large asset quantitative investment model obtains 36.47% strategic return rate and 21.4% maximum return rate in the backtesting range,which is far higher than the same period return of the 300 benchmark assets of Shanghai and Shenzhen,so the model is an effective quantitative investment model.This model has some advantages and guiding significance for the short-term investment.Compared with the existing research,the core contribution of this paper is to build a quantitative model of asset allocation based on China's financial market by using a variety of machine learning algorithms.In addition,by combining the machine learning model with the traditional asset allocation model,the paper improves and applies the big asset allocation model,which effectively broadens the design idea of quantitative investment strategy for asset allocation,and has practical reference significance for the further application of machine learning method in the field of quantitative investment.Moreover,it has important theoretical and applied value to the healthy development of China's capital market and the investment practice of investment institutions.