Concurrent Data-Structures Applied to Financial Data-Stream Processing Applying Concurrent Lock-Free Data-Structures to the design and development of a Financial Options Pricing Stream Processor

Abstract

This Thesis focuses on the efficient utilization of lock-free concurrent data structures in the scope of financial data-stream processing to achieve low latency and high throughput parallel solutions responding to the continuously increasing high throughput and low latency demand to process financial streams of data [17, 14, 30]. The two main problems address in the scope of this Thesis are options pricing and risk assessment based on volatility aggregation. A proof-of-concept financial stream processing engine has been designed and developed consuming a stream of data representing the real-time behavior of the underlying stock exchange market, and a stream of data representing the specifications of the option contracts to be priced to produce an output stream of priced option contracts. The throughput and latency results obtained when evaluating the different proposed solutions suggest that the ScaleGate data-structure, [7, 22], when efficiently used expediting its behavior with a heartbeat mechanism, satisfactorily responds to the aforementioned high throughput and low latency demand in addition to guaranteeing the correct ordering of the resulting output stream in non-decreasing timestamp order.