Learning algorithms such as neural networks are certainly suitable for forex forecasts, since there are thousands of learning samples available at low costs, and there are no very clear rules how these markets work.
Forex is one of the most interesting fields for optimization of algorithms:
- if you make everything perfectly, you might get very rich.
- competition is very high, so only perfect solutions might be good enough.
- it is not obvious how learning samples should be extracted from past market data:
what should be the desired output of the system, at any given moment ?
- it is not obvious how the goal function(s) should be designed.
- last but not least, it is not even clear whether there is a solution or not. It is not possible to give a mathematical proof that an algorithm will be profitable, in the future.
robot optimization