A Research On Stock Price Fluctuation Mode Based On CSI300

While stocks constitute complex networks in the stock market,correlation of price fluctuation and the rotation of stock sector are reflection of external information out of stock market.In the current research of the stock market,the static network analysis is more mature than the dynamic network analysis which is still in the initial stage.Its method is that splitting the stock network according to different timeline and then comparing the results.The networks of different time slices are still relatively independent.Therefore,this paper uses the method named alignment analysis of networks,which is often used in biological networks,to deeply study the spread mode of stocks' price fluctuation from the angle of alignment analysis of networks.In order to achieve our purpose,the research steps are divided into three parts,and we have analyzed the stock price of CSI300 as a sample.In the correlation analysis and clustering analysis of stocks,this study uses Spearman rank correlation coefficient as the measure of stocks' correlation,which can theoretically reflect the stock price fluctuation correlation more accurately.After our calculation,the mean correlation coefficient of CSI300 stocks is 0.392.Then we use the hierarchical clustering method to divide the stocks into 25 clusters and analyze with the fundamental information.To understand the stock blocks' propagation characteristics better,the stock blocks are used to construct the alignment model of networks as units.For the construction of the stock blocks' index,this paper proposes a set of calculation method for the general stock price sequence,which uses the stock value as the weight.After calculating the index of the stock blocks,the results show a good representation of the stock blocks' index trend.Through the static network analysis,the stock blocks,which have less consistency of industry in a cluster,have much more different kinds of influence to the stock price so that they have more similarity to other blocks in their index trend and have a central position in the stock blocks network.Finally,in the analysis of networks alignment,this paper uses the Isorank algorithm whose idea is that only if the adjacent nodes of the node pairs are similar will the node pairs have the possibility of similarity.Combined with the power method and greedy algorithm,we align the 319 block networks from CSI300 and get two conclusions: First of all,the stock blocks in the network structure tends to be stable so that self-alignment is the main result in network alignment.Secondly,when the stock blocks network structure changes,each block has a stable mode named "block circle",which results in that the alignment result of stock blocks networks has some high frequency conditions.General speaking,this paper provides a new research perspective for stock network analysis.