Neural Networks are powerful tools. But you need experience to model them.
Echo State Network is a powerful concept that gives good price predictions in forex trading.
Feed Forward Neural Networks are not good when it comes to predicting high frequency financial time series data.
This is due to the fact that a Feed Forward Neural Network doesn't cater for the past history.
We all know price is serially correlated with its past values.
This is the basis of Technical Analysis. This is what we know as Chartists.
By looking at the past prices we can predict the market in the future.
Recurrent Neural Networks were developed to cater for the past values serially correlated with the present value.
But training Recurrent Neural Networks are very time consuming.
Enter Echo State Networks that are very fast to train.
Echo State Networks use the concept of Reservoir Computing by building a large reservoir in between the input and output layers.
I have written this blog post that explains in detail what are Echo State Networks and how to use them in daily forex trading.
I provide the python code and then show you how to improve the price predictions with your Echo State Network Model.