Sixteen target NNs with different learning algorithms (Algs) and network structures.

The first column lists the notation of each NN. The network structures (MSp, MRp, CSp, and CRp) used in the previous experiments are highlighted. The column Layers lists the number of fully connected layers of MLP and the number of convolutional layers for CNN. The column Units/Filters lists the number of units in the hidden layers of MLP and the number of filters in the convolutional layers of CNN. The columns Ours and Baseline list the cumulative assets obtained by the target NNs over the entire test period.