Forecasting Foreign Exchange Volatility Using Deep Learning Autoencoder-LSTM Techniques

Table 4

Five error measures for each model in the BPVIX.

	MSE	RMSE	MAE	MPE (%)	MAPE (%)
Period 1
LSTM	0.1550	0.3937	0.2656	−0.4904	3.0105
Autoencoder-LSTM	0.1846	0.4296	0.2839	−0.2830	3.2280
Period 2
LSTM	2.6645	1.6323	1.3887	9.5562	19.0619
Autoencoder-LSTM	1.4706	1.2127	0.8508	−3.0532	10.1669
Period 3
LSTM	2.2017	1.4838	1.0805	−6.1679	11.1552
Autoencoder-LSTM	1.6874	1.2990	0.9039	−0.5791	9.1082