Moreover, the researchers do not agree on which
fundamental models are the most suitable for predicting the
future exchange rate. Hsing (2010a) in his research on
determination of USD/AUD exchange rate makes a
conclusion, that uncovered interest rate parity model reflects
exchange rate movements the best, followed by purchasing
power parity, flexible price monetary and Mundell Fleming
models. On the other hand, when the same author makes a
research on RON/USD exchange rate, he concludes that the
fluctuations of this exchange rate are mostly reflected by the
same uncovered interest rate parity model, but the second best
model is the flexible price monetary, followed by Mundell
Fleming and purchasing power parity models [24]. Rasekhi
and Rostamzadeh (2011) made a genetic algorithm based on
various fundamental models for predicting EUR/USD
exchange rate and noticed, that after the calculations by the
algorithm were made, the largest part of the algorithm was
taken by portfolio balance model, which means that this model
is better for determining the exchange rate fluctuations than
the other models, while relative purchasing power parity
model comprises the smallest part of the algorithm. On the
other hand, other researchers in the field say, that conventional
fundamental models are not appropriate for predicting the
exchange rate [40, 13, 3, 29]. The discussion is resumed by
[12, 33], who say that these models show poor results because
of the inappropriate judgment of the forecasts which are
generated by the models. Engel et. al. (2007) draws the
attention to alternative microstructure approach which was
introduced by Lyons in the first part of 2000s