A Multithreaded FX Algo Backtesting System

Yoshiharu Sato

Warsaw School of Economics & University of Warsaw

Figure 1: Our backtesting system in action

Abstract

Over the past decade, algorithmic trading has been increasingly growing its trade volume shares within the foreign exchange market. Thanks to the introduction and subsequent popularization of the MetaTrader 4 electronic trading platform, even individual retail traders are now able to use their own algorithms to trade currencies in the market. In order to devise a profitable FX trading algorithm, it is vital for developers to backtest their algorithms using different sets of parameters, different time frames, different time periods, and different currency pairs. Backtesting therefore is inherently a time-consuming process, thus it is critical for developers to reduce the amount of time taken up by backtesting so as to increase productivity. For this reason, we designed and built a backtesting system for FX trading algorithms completely from scratch in order to maximize the speed, flexibility, and accuracy of backtesting. Using our framework, developers are able to implement their own FX trading algorithms in C++, backtest them in an efficient and accurate manner, and consequently assess the historical profitability and risk of their algorithms.

Keywords: Algorithmic Trading, Backtesting, Foreign Exchange, Quantitative Finance, Monte Carlo Simulation, Concurrent Programming, Asynchronous Parallelism

Video