Reliable profit is only possible with sound trading strategies that are well tested. The removal of the human component from trading requires not only powerful indicators, but also a supporting structure that allows to build algorithms and strategies that can evolve with time. It's therefore important to have a strong back-end pipeline that deals with data mining and that enables accurate simulation and testing.
Our continuous-build infrastructure periodically gathers market data and utilizes such data for regression testing. Regression testing automatically keeps track of the performance of existing algorithms and allows us to decide if some models should be revised.
We utilize both analytical and machine learning approaches to automatically select the most suited algorithms and their relative parameters for any given market.
The run-time is built in modern C++ from the ground up. C++ allows us to leverage maximum performance where necessary, expanding the range of possible techniques to be used, including the most demanding number crunching approaches, as well as reducing latency in connectivity with the exchanges. All of our calculation-heavy code is multi-threaded and it automatically adapts to utilize of available CPU cores of the hardware it runs on.