A Static-Dynamic Hypergraph Neural Network Framework Based on Residual Learning for Stock Recommendation

Table 2

Ranking performance of different methods on the NASDAQ and NYSE datasets when considering different numbers of N.

Model	NASDAQ						NYSE
	IC (↑)	Rank IC (↑)	Precision@N (↑)				IC (↑)	Rank IC (↑)	Precision@N (↑)
			Top3	Top5	Top10	Top30			Top3	Top5	Top10	Top30
MLP	0.023	0.015	49.45	48.43	46.32	44.16	0.021	0.013	49.04	47.72	47.93	45.28
SFM	0.027	0.022	51.19	50.42	50.21	49.92	0.024	0.024	51.19	51.05	50.22	49.57
GRU	0.031	0.029	52.74	51.86	50.43	50.13	0.031	0.028	52.17	51.47	51.44	50.93
LSTM	0.030	0.027	51.65	51.22	50.73	50.18	0.030	0.027	51.75	51.52	50.89	50.48
ALSTM	0.036	0.034	53.79	52.03	51.98	51.39	0.036	0.035	53.03	52.33	50.71	50.30
Transformer	0.029	0.026	51.58	51.19	50.30	50.08	0.030	0.027	52.02	51.17	50.63	50.19
ALSTM + TRA	0.040	0.040	54.02	53.49	53.41	52.86	0.039	0.038	53.95	53.52	52.96	52.24
GATs	0.038	0.037	53.87	53.21	52.48	52.09	0.038	0.038	53.87	53.62	52.89	51.72
HIST	0.042	0.041	54.71	54.15	53.43	53.02	0.047	0.045	55.14	55.08	54.81	53.97
SD-RL	0.044	0.044	55.28	55.34	54.85	54.43	0.047	0.046	55.68	55.41	54.84	53.83

The best result in terms of each metric is indicated in bold.