Financial Innovation

Table 6 Gaps between actual and predicted returns under the quadratic function

From: Predicting the returns of the US real estate investment trust market: evidence from the group method of data handling neural network

Day	Quadratic-300/30	Quadratic-600/30
1	− 2.85	− 1.36
2	− 4.42	− 0.86
3	− 3.15	− 0.16
4	− 3.25	− 0.39
5	− 2.88	− 0.34
6	− 4.32	− 1.23
7	− 1.99	0.03
8	− 2.57	− 0.83
9	− 0.74	− 0.10
10	− 1.01	− 0.39
11	0.07	0.58
12	− 0.92	− 0.34
13	− 2.18	− 1.82
14	− 1.14	− 0.57
15	0.15	1.18
16	0.20	1.12
17	− 1.22	− 0.49
18	− 0.97	0.64
19	− 0.87	0.58
20	0.50	0.25
21	− 0.14	− 0.69
22	0.63	0.87
23	− 0.72	0.50
24	− 0.28	− 0.56
25	1.01	0.41
26	− 1.92	− 1.88
27	− 0.10	0.26
28	− 0.86	− 0.12
29	− 0.61	0.41
30	− 2.52	− 0.95
31	− 1.31	− 0.02
32	− 5.18	0.34
MSE	4.35	0.62

Note: The gaps between the actual returns and the predicted returns of quadratic-30/30, quadratic-100/30, and quadratic-200/30 are not calculated because the results of these three cases are not acceptable

Back to article page