| Stock is an important way for people to invest and doing financial and asset management.Precise investment prediction for stocks can not only bring benefits to investors,but also greatly promote the development of the national economy.Therefore,the stock market has always been attentioned by investors and experts.As the changes in the stock market present a complex volatility trend,the stock price is greatly affected by uncertain noise signals,which in turn makes many challenges in investing in stocks and predicting price trends.Therefore,as a product of the integration of the Internet and finance,the intelligent investment adviser-Robo-adviso emerged.Robo-adviso,as an online tool,is an innovative form of wealth management.Through automated algorithm analysis applications and professional Internet trading platform analysis technology combined with big data technology,we provide professional and personalized stock investment and financial advice for customers and investors,and provide consulting service of professional asset allocation risk management.From the perspective of information fusion and MACD technical indicators,in this paper,we build a model of expert stock reviews and stock daily data on the website,which can predict stock trends,and realize investment management of stocks.Main tasks are as follows:(1)Build a framework for text classification,combine the idea of weighted fusion,and fuse at the feature layer and decision layer to achieve accurate identification of the subject of expert stock review texts.Specifically including: feature selection layer,weighting fusion of multiple feature selection methods,so that it can fully characterize stock text characteristics;Decision layer,based on SVM-score,integrates decision layers of multiple classifiers to build an enhanced classifier for final text topic discrimination.To avoid overfitting,the kernel function of the support vector machine uses a linear kernel function.(2)The weighted average thought of MACD is integrated into the singular value decomposition,and then the MACD model is modified.First,a special reconstructed attractor two-dimensional signal matrix is constructed based on the idea of weighted average;Second,use singular value decomposition to eliminate the interference of uncertain noise;Finally,based on the revised MACD,the stock buying and selling points are judged,so that it can accurately reflect the fluctuation pattern of stock prices and achieve accurate stock prediction.The innovation lies in cleverly integrating the MACD algorithm into the singular value decomposition matrix,and then using SVD to achieve the elimination or reduction of uncertain noise and complete the prediction of stocks.The experimental data is derived from the text report of "Broad Market Analysis" by Oriental Fortune.com and the daily data of the Shanghai and Shenzhen stock markets are downloaded from the Juchao website from January 1,2016 to November 3,2017,used to test the effectiveness of the above method.Experiments based on measured data show that compared with a single-mode text topic recognition method,the recognition accuracy of the multi-layer fusion algorithm proposed in the article is significantly improved.Compared with the traditional MACD technical indicators,we can achieve accurate stock prediction by MACD correction model based on reconstructed attractor singular value decomposition. |
PDF Full Text Request