High frequency trading is steadily taking overthe equity trading world. High frequency trading involves veryhigh speed systems placing trades at sub millisecond speedsacross multiple stock exchanges. HFT is a good example forBig Data analytics - especially the velocity aspect of big data. For HFT strategies to be profitable, real time processing of bigdata is essential. In this paper we discuss the challenges facedby HFT systems and the opportunity for big data processingwith low latency in the field. Most HFT systems are designedusing real time stream processing, which have certaindrawbacks. We present a theoretical framework for buildinghigh frequency trading systems using the complex eventprocessing paradigm which could overcome the drawbacks ofstream processing. Complex event processing enables detectingpatterns of events from disparate events streams and respondsto the detected pattern. The applicability of the framework forHFT applications is discussed.