Deep Learning for Price Movement Prediction Using Convolutional Neural Network and Long Short-Term Memory

Table 6

The average performance of the experiments with different .

Metrics		CNN3D	CNN3D-DR	LSTM-D	CNN3D + LSTM	CNN3D-D + LSTM	CNN3D-DR + LSTM
Accuracy	60/40	0.5082	0.4957	0.5202	0.4993	0.5693	0.5833
	65/35	0.5095	0.5009	0.5348	0.5130	0.5605	0.5728
	70/30	0.4798	0.4952	0.5339	0.5079	0.5991	0.6160
	75/25	0.4856	0.5200	0.5229	0.5150	0.5968	0.6316
	80/20	0.4878	0.5180	0.5175	0.5010	0.5653	0.5916
	Average	0.4942	0.5060	0.5258	0.5072	0.5782	0.5991
Precision	60/40	0.5019	0.4954	0.5124	0.4991	0.5671	0.5818
	65/35	0.5071	0.4985	0.5317	0.5029	0.5529	0.5650
	70/30	0.4904	0.4927	0.5150	0.5004	0.5929	0.6107
	75/25	0.5045	0.5165	0.5006	0.5222	0.5921	0.6281
	80/20	0.4949	0.5183	0.5089	0.5009	0.5620	0.5894
	Average	0.4998	0.5043	0.5137	0.5051	0.5734	0.5950
Recall	60/40	0.5017	0.4964	0.5107	0.4992	0.5635	0.5779
	65/35	0.5062	0.4975	0.5276	0.5027	0.5513	0.5621
	70/30	0.4938	0.4937	0.5130	0.5002	0.5910	0.6094
	75/25	0.5034	0.5155	0.5003	0.5218	0.5897	0.6249
	80/20	0.4951	0.5184	0.5087	0.5008	0.5596	0.5860
	Average	0.5001	0.5043	0.5121	0.5049	0.5710	0.5921
F-measure	60/40	0.4954	0.4951	0.4962	0.4986	0.5602	0.5754
	65/35	0.4989	0.4980	0.5154	0.5000	0.5507	0.5609
	70/30	0.4632	0.4923	0.5043	0.4989	0.5912	0.6096
	75/25	0.4791	0.5157	0.4911	0.5141	0.5894	0.6250
	80/20	0.4875	0.5163	0.5072	0.4995	0.5577	0.5845
	Average	0.4848	0.5035	0.5028	0.5022	0.5698	0.5911