Research On The Forecast Of Trading Volume Based On Portfolio Model

With the gradual deepening of information technology in people's daily life,the financial field has also begun a full range of scientific and technological reform,the internet-based online banking has been highly concerned by the financial industry.In order to improve their own competitiveness and accelerate the transformation from traditional banks,all major banks put the development of online banking on the agenda.The essence of online banking is virtual banking,which does not exist any entity,only uses the Internet as the carrier to connect customers and banks.Online banking can help customers to achieve transaction needs and obtain all-round services at any time and any place in any way.In this fast-paced society,the advantages of real-time,fast,convenient and low cost make online banking greatly meet the needs of customers.In recent years,the development of personalized products of online banking is going deeper and deeper,and the total annual transaction volume is increasing year by year.Compared with traditional business,the proportion of consumers and transaction volume of online banking business is relatively small,but there is still a large space for development.Therefore,the research on the transaction volume of online banking can provide an important basis for the product development and multi-dimensional development of online banking,which is of great significance to the development of the whole banking industry.This paper selects the transaction volume data of a bank's online banking from September 2018 to March 2019 as the research object,selects Eviews,Excel and R to process the data and establish the model analysis.Before the establishment of the model,the 180 day trading volume data from September 2018 to February 2019 is selected as the training data set,and the trading volume data from March 2019 is selected as the test data set.Through the comparative analysis of domestic and foreign research,ARIMA model,ridge regression,lasso and elastic network are selected to model the data and compare the prediction results.Empirical analysis shows that the prediction accuracy of ARIMA model is higher than that of regularized linear model.ARIMA model can make a systematic description of trading volume data,and regularized linear model can capture the impact of multiple factors on trading volume data.Therefore,the first mock exam is the first mock exam of the transaction volume data by using the combination model of ARIMA model and regularized linear model.The prediction accuracy of the combined model is higher than that of the single model after comparing the prediction results of the single model and the combined model.Finally,the prediction results of the seven models are compared in order to find the model with the best interpretation and prediction accuracy.