In this project supervised by SEB, the analysts Markus Lexander, Paulina Ibek, and Philip Mattisson investigate factors that predict the exchange rate of the Norwegian Krone, EURNOK. By employing ML techniques such as feature importance and random forest in conjunction, the investigation focused on the predictive power of variables such as the price of the Brent oil, EURSEK, the DAX, among others, on the future price of NOK on various time horizons.
The results indicate that different combinations of the selected features have the best predictive power on different time scales. The Martingale model theory of price movements was found to hold better on short (minutes) timescales, while Brent-price driven trends dominated on longer (days) time horizons.