Exchange Rate Forecasting in Economies with High Levels of Inflation

---

Author: Doctoral Student PhD Boyan Lomev (Sofia University "St Kliment Ohridski")

Keywords: Exchange rate forecasting, FARIMA (p, d, q), Attractor, Back Propagation Neural Networks, GARCH.

The classical theories studying exchange rates do not provide base for accurate forecasts especially concerning currency crisis. This research tests the principal possibility for currency exchange rate forecasting by: -Linear methods FARIMA (p, d, q) with long term relations and heavy tails; -Methods for nonlinear prediction of chaotic time series. Null forecast and average are used as main comparison measure. As additional comparison measures are used the results from the methods: Back Propagation Neural Networks, Generalized Auto Regression with Conditional Heteroscedastisity, several smoothing methods.