Closed ToughStyle closed 6 months ago
Hi,
Thank you for your feedback and for experimenting with our code. We acknowledge the issues you’ve encountered and appreciate your patience as we work through them.
During our initial development, the random seed was not set, and we have noticed some discrepancies due to different GPUs. We are currently conducting extensive experiments to adjust our model and enhance its precision. Moreover, the manuscript is under revision to address peer review comments and has not yet been published due to some details we are actively improving.
We have also taken into consideration the reviewers' suggestions regarding the effectiveness of traditional metrics like MAE, RMSE, MAPE, and accuracy in quantitative trading. We are in the process of adopting more relevant quantitative metrics such as Information Ratio (IR) and Information Coefficient (ICIR) for a more meaningful comparison of models. We plan to update our paper and code accordingly.
Please stay tuned for the updated version. We are committed to ensuring that our methodology is robust and our findings are reliable.
Best regards, Eric
The directional accuracy claimed in paper is 62.39%, but here I got only less than 52% with your code. All settings are original, any suggestions?