Long fourth dimension readers of this spider web log know that I haven't constitute data mining or artificial intelligence techniques to live on real useful for my ain trading, for they typically overfit to non-recurring past times patterns. (Not surprisingly, they are much to a greater extent than useful for driverless cars.) Nevertheless, i must maintain an opened upward heed together with continues to maintain tabs on novel developments inwards this field.
To this end, here is a novel newspaper written past times an technology scientific discipline pupil at UC Berkeley which uses "support
vector machine" together amongst 10 uncomplicated technical indicators to predict the SPX index, purportedly amongst 60% accuracy. If i includes an additional indicator which measures the lay out of word articles on a stock inwards the previous day, thence the accuracy supposedly goes upward to 70%.
I did non accept the peril to reproduce together with verify this consequence yet, precisely I invite y'all to essay it out together with percentage your findings here. If y'all arrive at so, y'all may notice this novel information mining production called 11Ants Analytics useful. It is an Excel-based software that includes 11 machine learning algorithms including the aforementioned back upward vector machines. It equally good includes determination trees which are sometimes quite useful inwards automatically generating a small-scale laid upward of trading rules from an input laid upward of technical indicators. (Whether those rules rest profitable inwards the time to come is some other question!) If y'all accept tried this product, I would equally good appreciate your comments here.
(If y'all are a die-hard MATLAB fan, back upward vector machines are available inwards their Bioinformatics Toolbox, together with classification together with determination trees inwards their Statistics Toolbox.)
Tidak ada komentar:
Posting Komentar