Forecasting of chaotic changes of exchange rates usually is based on historical data and depends on the choice of time intervals. This study seeks to develop new forecasting method based on data of different time zones. This paper demonstrates how the using of London and New York divisions of the trading day allows getting additional information from predicting exchange rates. This was modelled with the help of ensemble of EVOLINO for obtaining of predictions of the distribution of expected values. The obtained results show that double forecasts evaluation reveals a possible trend in the exchange market and enriches the choice of real-time trading strategies.

DOI: https://doi.org/10.3846/bm.2016.29