The ability to forecast markets has become a major factor contributingto the success of any enterprise in today’s business world. Byemploying a profitable prediction technique, a company can stay ahead ofthe game and thus gain a competitive edge on the market. The aim of thiswork is to investigate the forecasting capabilities of artificial neural networksin general, as well as provide some concrete guidelines that can be followedto effectively utilize their predictive potential. To approach this challengingobjective, we have designed and implemented a potent artificial neuralnetwork class library, whose ambition is to provide all necessary functionalityrequired in the process of choosing the ‘best’ forecasting model from all‘candidate’ models. On top of this library, a lightweight console applicationhas been built to automate the process of trying out and benchmarking thethese models. We have centered our experimentation around a typical economictime series containing trend and seasonal variations.