Research Article
[Retracted] Beat Wash-Sale Tax with Multigraph Convolutional Neural Networks Based Trading Strategy
Table 11
Summary for 30% expected return weight portfolio trading without limitation.
| 30% Expected return weight | Model | MDD | Return (%) | Counts | SR | Win rate (%) | Loss (106) | Profit (106) | Net volume (106) |
| Baseline | 0.33 | 103.96 | 1 | 2.36 | | | 2.08 | 1.69 | Random Forest | 0.14 | 17.83 | 2297 | 0.87 | 0.53 | 2.24 | 0.36 | −0.57 | XGBoost | 0.17 | 10.48 | 2290 | 0.53 | 0.51 | 2.46 | 0.21 | −0.74 | GRU | 0.21 | −2.68 | 1927 | −0.17 | 0.51 | 3.21 | −0.05 | −1.19 | LSTM | 0.13 | 121.96 | 1155 | 4.48 | 0.51 | 3.16 | 2.44 | 0.40 | Multi-GCN-GRU | 0.05 | 327.14 | 1718 | 8.89 | 0.59 | 2.33 | 6.54 | 3.30 | Multi-GCN-LSTM | 0.06 | 282.88 | 1601 | 9.33 | 0.62 | 2.94 | 5.66 | 2.51 |
|
|