DMForex: a data mining application to predict currency exchange rates and trends

The foreign exchange (FX) market is one of the most liquid markets in the world today. In this market currency pairs are sold and bought based on the exchange rate. It is important to market participants to accurately predict future values and trends. In this project, DMForex has been designed and implemented to predict currency exchange rates and trends using data mining techniques and algorithms. DMForex is a composite WPF application designed and built using the PRISM 4 guidance. It uses the Model-View-ViewModel (MVVM) UI (user interface) pattern to separate the presentation and business logic of the application. To implement the forecasting of the exchange rates and trends, DMForex uses SQL Server 2008 with Analysis Services. Daily historical price data for some of major world currency pairs for the last 12 years is used. Microsoft SQL Server Analysis Services' Analysis Management Objects (AMO) are used to programmatically implement the predictions of exchange rates and trends by creating the analysis database, mining structure, and mining models. These mining models can then be queried through ADOMD.Net using Data Mining Extensions (DMX) commands. To predict the exchange rates, the program creates a mining model that uses the Microsoft's Time Series Algorithm using the standard dataset. For the prediction of trends, the user is able to change the dataset by adding the Simple Moving Average (SMA) and Relative Strength Index (RSI) columns. These new columns are calculated using the TA-lib library of technical indicators. The user defines a trend. This column is defined in terms of days and PIPs. The program then predicts the next trend using Microsoft's Decision Trees Algorithm. A testing module allows the user to test the accuracy of the rates and trends predictions.