Abstract:

In the study, the TCMB’s Gross Currency Reserve was predicted by the Techniques of the Artificial Sinery Network (YSA) and the data from 2013-2021. After the study was assessed whether the gross currency reserves of Turkey did not match the optimal reserve size of the IMF, the YSA model was established. In the selection of variables that have influenced the gross currency reserves, macroeconomic factors have been accepted in the literature and have been used in particular empirically applied verified variables for Turkey. The YSA structure, predicted by the single layer and 20 neuron model, is trained with the Levenberg-Marquardt back spread algorithm. The model has been trained with 1000 interaction trials, solving the two-problem, and demonstrated the model success according to the values of RMSE, MAE, MAPE and R2. After the successful model was determined, the gross currency reserves for July 2021 and June 2022 were forecasted. It is estimated that in the next 12 months the gross currency reserves are expected to move horizontally.

Keywords: