Predicting Trading Signal Of The Foreign Exchange With Neural Networks

Nowadays, the environments of economy are trending to open increasingly, governments, banks and enterprises even individuals have to accept this fact, they are affected by exchange rate initiatively or passively. When they possess hard currency, the rising of exchange rate will bring them profit, when they possess soft currency, the falling of exchange rate will bring them loss. In order to overcome the weakness of human nature and erase negative effect of foreign exchange trading, quite a number of bargainers, including governments, enterprise and individuals, research and develop all kinds of machine trading systems. They look forward to obtain scheduled profit through these systems.Machine trading system includes three subsystems, they are Trading Signal Decision-making Subsystem, Risk Management Subsystem and Money Management Subsystem. The first one is the foundation of the two others, they make the risks management and investment management according to the output of deal signals by trading signal decision-making subsystem.Trading signal, the key part of machine trading system, is deeply researched in this thesis. In short-term FX trading system, prediction model system is designed using current prevalent NN theory(RRL,BP,SOM). By comparing with the experiments results of given intervals, writer will find out which theory is the best one to obtain the signal of foreign exchange.