Font Size: a A A

Verification Of Candlestick Charts Validity Based On Convolution Neural Network

Posted on:2019-08-11Degree:MasterType:Thesis
Country:ChinaCandidate:X F WangFull Text:PDF
GTID:2429330593950916Subject:Finance
Abstract/Summary:PDF Full Text Request
Since the birth of technical analysis index,there has been controversy over whether it is effective or not.In the existing research,different conclusions have been obtained in different stock markets or different time periods.As an important research area of graphical analysis in technical analysis,candlestick charts has accumulated a series of analysis and application rules over the years through the application of historical investors.However,the candlestick charts analysis relies heavily on subjective factors such as the individual's experience of the analyst,Different people tend to get different or even opposite results on the application of uniform rules.Therefore,as a representative of the historical information,the candlestick charts,whether it really contains information that can predict future price changes has become an attractive research direction.With the extensive study of deep learning in recent years,the application of deep neural networks to the stock market prediction has also become a hot research direction.However,the existing studies all have time-series pricing as the input form of the network and seldom directly talk about the candlestick charts applied to neural network research.Convolutional neural networks have unique advantages in image classification and recognition,and the commercial applications are relatively mature.Therefore,in this dissertation,the unique advantages of convolutional neural networks in image recognition are combined with the research of Candlestick charts,and we applied to control Through the control of rigorous experimental conditions,a series of control experiments were set up to conduct a comparative study,trying to verify whether the candlestick charts really contains information that can predict the price changes in the future.Finally,the research in this dissertation finds that it is feasible to study the Candlestick charts directly as the input form of convolutional neural network.And initially got,Candlestick charts does contain the information that can predict future price changes,in other words,the technical analysis is effective from the perspective of this thesis.
Keywords/Search Tags:Technical analysis, Convolutional neural networks, Candlestick charts, Verification
PDF Full Text Request
Related items