Dynamic Interaction Network to Model Co-Integration in IDX Leading Stocks
Most studies in the field of dynamic system analysis and modeling have focused more on the task of predicting movement of contributing variables only, whilst the extraction of co-integration patterns that govern movement of these variables has received little attention. This study is concerned with advances in the analysis and modeling of co-integration between leading stocks in the Indonesian Stock Exchange using a method capable of extracting multiple relationships between variables named the Dynamic Interaction Network. Being able to accomplish such task is expected to lead to a better understanding of co-integration state between leading stocks of the observed market, which in the end would lead toward a better means in predicting their price movement. Results from conducted experiment in this study suggest that: (1) the Dynamic Interaction Network is capable of modeling dynamic pattern of interactions between variables and (2) that the idea of including the states of co-integration between an assembly of time dependent variables into a prediction model appears to be beneficial to tackle the problem of time series prediction.
Artikel ini telah dipublikasikan pada: