Research Article

A Stock Selection Model of Image Classification Method Based on Convolutional Neural Network

Table 4

Results of CNN under different parameters.

Training sizeSample widthTypeBenchmarkAnnual returnSharp ratioMax drawdown

363145285643510.01460.2176
363245285612480.01960.2420
363345285201710.01340.2188
24312986412915−0.00110.2437
243229864750600.02430.2225
24332986415195−0.01350.2548
3661452856464−0.00030.2171
366245285277370.01590.2060
366345285588980.01940.2041
246129864−6565−0.00380.2429
246229864−24867−0.02100.2394
24632986412791−0.00130.2159