Forecasting Exchange Rates Better than the Random Walk Thanks to Machine Learning Techniques

Christophe Amat, Tomasz Michalski, Gilles Stoltz

Abstract

Using methods from machine learning – adaptive sequential ridge regression with dis-count factors – that prevent overfitting in-sample for better and more stable forecasting performance out-of-sample we show that fundamentals from the PPP, UIRP and monetary models consistently improve the accuracy of exchange rate forecasts for major currencies over the floating period era 1973–2013 and are able to beat the random walk prediction giving up to 5 % improvements in terms of the RMSE at a 1 month forecast. “Classic ” fun-damentals hence contain useful information about exchange rates even for short forecasting horizons.
BibTeX
@misc{amat2014forecasting,
  author = {Christophe Amat and Tomasz Michalski and Gilles Stoltz},
  title = {{Forecasting Exchange Rates Better than the Random Walk Thanks to Machine Learning Techniques}},
  year = {2014},
  doi = {10.2139/ssrn.2448655},
  howpublished = {SSRN},
}

Loading related content...

Figure