Research on Market Risk Measurement of Foreign Exchange in the International Trade Based on Deep Learning
Yuan Zhang · 2022
Aiming at the problem of poor market risk prediction of traditional exchange rate risk prediction model, a exchange rate risk prediction model based on deep learning was constructed. Firstly, the risk prediction model combining recurrent neural network (LSTM) and deep belief network (DBN) was used to predict exchange rate risk of the foreign exchange market. Then, the prediction results of the model were processed in an integrated way. Finally, this model was compared with the benchmark model ARMA model. Experimental results show that compared with traditional ARMA model, the proposed LSTM+DBN risk prediction model can accurately predict exchange rate of the foreign exchange market, and has a better performance in the data test of the mainstream exchange rate market, which proves that the prediction model has advantages in the exchange rate test.