Algorithmic trading systems generally rely on pattern methods in order to detect and exploit market inefficiencies for speculative purpose. Efficient computational algorithms are required to process massive amounts of data collected from many co-evolving data streams. Recent progress in network technology allows the delivery of data streams almost in real time and at a very high frequency. DataCatalyst is a set of GPU accelerated pattern recognition algorithm which can be used in trading systems to perform online adjustment of their decision making strategy, or to evaluate the performance of a trading strategy in a back-testing.