Although exchange rate determination and forecasting is a popular topic among the researchers, the dicsuccion on exchange rate disconnect puzzle still exists, meaning that fundamental models are unable to provide better short term forecasting results than the random walk. Therefore, the following objective for the dissertation is stated: After performing analysis of the exchange rate forecasting theories, to define information flows in the foreign exchange market and empirically test the information flow model in determining the movements of exchange rates. Scientific problem of the research: What information flows are in the foreign exchange market and how information flows determine the movements of exchange rates. The classification of exchange rate determination and forecasting models was clarified by distinguishing fundamental, historical price indicators’, alternative and efficient market hypothesis models. The reasons for possible poor performance of the models were determined. Based on the analysis of previous research in the field, the concept of information flows in the field of exchange rate determination was clarified and research model was composed. Information flows are composed of three sources of information affecting the movements of exchange rates, i.e. fundamental information, historical price indicators’ information, and information of market participants’s positioning data. The empirical part proves that market participants’ positioning data should be analysed as a separate information flow among fundamental and historical price indicators’ information flows. It was determined that the information flow model is able to outperform random walk forecasts in various volatility periods and among various groups of currency pairs. However, the determination coefficients in the regressions were among the ones already researched in the previous researches.