Research Article

Multivariate CNN-LSTM Model for Multiple Parallel Financial Time-Series Prediction

Table 4

Comparison of descriptive statistics between actual and predicted time-series data produced by the proposed multivariate CNN-LSTM model in testing period, i.e., last 72 trading days of 2020. Note: Act = actual and Pred = predictions.

Descriptive statisticsHang Seng (HSI)Japan (N225)Singapore (STI)Indonesia (JSX)
ActPredActPredActPredActPred

Mean1.19171.18021.09971.08531.02601.02211.03151.0421
Median1.18721.17831.09561.08471.03361.04011.00931.0102
Standard deviation0.03200.02960.06770.06430.02770.02910.07870.0766
Sample variance0.00100.00120.00460.00430.00080.00710.00620.0058
Kurtosis0.32950.31871.77251.75230.81610.80211.33991.3187
Skewness0.33760.31520.12840.11791.19101.18760.37930.3685
Range0.14970.15020.20220.20130.11320.10450.25200.2471
Minimum1.12881.12561.01211.00970.95700.96130.92260.9197
Maximum1.27861.26911.21431.20951.07021.02731.17461.1637
Confidence level (95.0%)0.00750.00860.01590.01370.00650.00610.01850.0157