Building a neural network. Easy to build. Training it though is another matter altogether. A friend of mine is basing his system on a neural net.
Using FFT's. I've coded FFT in the past for digital signal processing. This is ok for simple datasets.
Both FFT and Neural Nets require a lot of computational horsepower if the dataset is large and relatively noisy as is the case with price data.
I'm going to stick with what I've started based upon TA and other more exotic things I'll try and derive.
I know nothing about statistical trading. WRT fundamental's, I've decided to go down the TA route rather than the fundamental route. I get a sense that market behaviour has little to do with the company and rather more about the faddy nature of human behaviour.
Hello everyone, I am new here
I am also interested in algorithmic trading - by profession I am a senior DSP researcher with PhD and extensive experience in automated signal interpretetion (in pro audio, medical applications, seismic signal, and industry).
Now I would like to play a bit with forex signals.
---
To clarify above:
:idea:Neural Network - training and verification is crucial. This is not a magical tool, but just a non-linear type of classifier. The more straightforward parameters describing the situation to be predicted, the prediction (with NN) will be more accurate. In other words, if you end up with a set of parameters allowing for very good prediction of the trend behaviour, the network will work very well.
But if you know these, you do not need the neural net...
:idea:FFT - works well for periodic signals (e.g. sinusoidal signal, square waveform, or other quasi-periodic waveforms). If there is periodicity in the trend, than FFT is useful for prediction (as well as linear prediction, and other periodicity DSP algorithms).
--
Both are computationally inexpensive - considering available today's CPU power of a personal computers.
==
I have one question: Where can I obtain historical data for forex (with > 0.3 Hz sampling rate = updated at least every 3 seconds, or more frequently
) -> for the last 10 years?
Is it possible to obtain it for free?
I am also interested in purchasing the data sets - please suggest where.
Thanx, 3d