MODELING JUMPS IN INTRADAILY FINANCIAL DATA: A REVIEW OF METHODS AND APPLICATIONS

Abstract

Jumps in intradaily financial data are a key feature of market dynamics and accurate modeling of these jumps for developing effective risk management and investment strategies could help identify potential risks and opportunities in financial markets. This article reviews methods and applications of time series modeling of intradaily financial data with emphasis on jumps modeling. The study identified challenges associated with intradaily data analysis, among which is the fact that sophisticated methods may be required to capture and account for complex seasonal patterns and anomalies inherent in intradaily data; this sophistication has repercussions for estimation (which may prove difficult, particularly in the presence of limited data), interpretation, and computational efficiency. Hence, efficient algorithms are required for modeling. The need to focus on development of more robust and flexible models capable of capturing complex dynamics of intradaily data’ s jump behavior in future research efforts was recognized. Such efforts could also include evaluation of robustness and sensitivity of existing methods. Efforts could be geared towards creation of hybrid models combining statistical and machine learning models to create robust hybrid models. This should complement efforts so far in that direction. Not much has been done on application of Hybrid ARMA and GAS models on intradaily financial data for instance. There is no large-scale comparison of hybrid models and stand-alone models on intradaily asset prices yet. Such a comparison should be a welcome development. The article concluded that through improved understanding of jump behavior (gained from availability of better techniques), operators and researchers will have a deeper understanding of financial asset markets and their dynamics, ultimately helping investment decisions. Efforts towards improvements of existing methods are therefore, a worthy endeavor.