Hello,
I am seeking an expert in Forex trading to help me finalize a prediction method for up/down trends in Forex data. Mainly, i am interested in what features are most important. Technical Analysis libraries for C# make many functions available, which ones are best??
My name is Daniel and i am a software developer with an interest in Machine Learning. Recently, in an attempt to improve my portfolio i am attempting to implement various Machine Learning methods.
Recently, i found a research paper outlining how to apply Support Vector Machines using Technical Analysis to make reasonable predictions on Forex data.
This is my method.....
Loop so SVM is always up to date:
1) Collect historical data from TrueFX, Historical/Live.
2) Parse ticks into database.
3) Take 30 days of data per sample.
4) Sample at time 't', calculate up or down trend for time 't+1'. Uptrend definition: At time t, Ask(t+1)>Bid(t)
5) Train SVM using Samples.
6) Predict, with confidence value.
7) If confidence>~70%, accept as up trend.
Using most up to date SVM, trained on largest sample set possible, make a prediction for t+1.
Issues:
1) Historical data has no volume values specified.
2) I am using EUR/USD data series.
3) I am using tick data to create daily OCHL values.
And one last question, if someone can clear this up for me. The Open/Close/High/Low values, are these calculated for the Bid or Ask price? I am assuming it's Ask price as that is the value to track.
Any assistance would be greatly appreciated. The goal is to demonstrate the application of machine learning on one of the hardest signals to predict/understand. Eventually, i want to create a demo video outlining how this method works and it's results for my YouTube channel.
Kind regards,
Daniel
I am seeking an expert in Forex trading to help me finalize a prediction method for up/down trends in Forex data. Mainly, i am interested in what features are most important. Technical Analysis libraries for C# make many functions available, which ones are best??
My name is Daniel and i am a software developer with an interest in Machine Learning. Recently, in an attempt to improve my portfolio i am attempting to implement various Machine Learning methods.
Recently, i found a research paper outlining how to apply Support Vector Machines using Technical Analysis to make reasonable predictions on Forex data.
This is my method.....
Loop so SVM is always up to date:
1) Collect historical data from TrueFX, Historical/Live.
2) Parse ticks into database.
3) Take 30 days of data per sample.
4) Sample at time 't', calculate up or down trend for time 't+1'. Uptrend definition: At time t, Ask(t+1)>Bid(t)
5) Train SVM using Samples.
6) Predict, with confidence value.
7) If confidence>~70%, accept as up trend.
Using most up to date SVM, trained on largest sample set possible, make a prediction for t+1.
Issues:
1) Historical data has no volume values specified.
2) I am using EUR/USD data series.
3) I am using tick data to create daily OCHL values.
And one last question, if someone can clear this up for me. The Open/Close/High/Low values, are these calculated for the Bid or Ask price? I am assuming it's Ask price as that is the value to track.
Any assistance would be greatly appreciated. The goal is to demonstrate the application of machine learning on one of the hardest signals to predict/understand. Eventually, i want to create a demo video outlining how this method works and it's results for my YouTube channel.
Kind regards,
Daniel
Aucun commentaire:
Enregistrer un commentaire