Search

Market price fluctuations follow a non-stationary process and accurately modelling them is absolutely impossible, however attempts have been made and any results, even the unsuccessful ones, contribute in a better understanding of the fluctuations dynamics. Through the use of a multivariate regression model analysing 237 distinct variables, consist mainly upon macroeconomic announcements, which portray the state of the economy, this study investigats their correlation of actual with predictive patterns. Adding further through the asset market approach, comparing the performance of substitute investment options did not prove any significant results. On the last group of variables, the refinance interest rates as set by central banks, showed an even less impact in a sense proving that monetary policy strategies on a free floating currency has fairly low. Even though multivariabe regression model could not accurately predict the price level, there was a positive by-product of trend predictability, among all 42 currecny pairs have predictive probability 15 pairs (represent 36% of total sample) lie in the range from 58-59%, an indication of relative high predtictive probability among all pairs considered for analysis. In addition, 62% of the pairs (26-pairs) documented the predictive probability between 55-57%, that is the dominant prediction and only a smaller pair percentage lies below 55. Although the prediction power of the market price movements is not much high, therefore based upon the notion of simplicity and accessibility to data, it has laid down the foundation stone for future research, thus it is expected that it will inspire others to further investigate the phenomena from various angeles with relatively more complex methodology.