search

SJTU-Quant / MASTER

This is the official code and supplementary materials for our AAAI-2024 paper: MASTER: Market-Guided Stock Transformer for Stock Price Forecasting. MASTER is a stock transformer for stock price forecasting, which models the momentary and cross-time stock correlation and guide feature selection with market information.
88 stars 18 forks source link

AR and IR Evaluations framework. #2

Closed SUFEHeisenberg closed 2 weeks ago

SUFEHeisenberg commented 1 month ago

Thanks for the remarkable job!

And I have a small question about the calculation framework about AR (Annualized Return ) and IR (Information Ratio) are not mentioned in the current framework. I think it might run in the qlib?

Could you please add this part for the comparison at same level please?

Looking forward to your reply!

Best Regards

LITONG99 commented 1 month ago

Thank you for your attention to this work.

Yes, to report the same portfolio-based metrics in the paper, the Qlib framework is enough. Please use the following configuration: {"class": "TopkDropoutStrategy", "module path": "qlib.contrib.strategy", "kwargs": {"signal":"<PRED>","topk": 30,"n_drop": 30}}

We regard your suggestion as of great value. Instead of rebuilding the entire backtest framework in this repo, we are collaborating with other researchers to integrate MASTER into the open-sourced Qlib framework, allowing more convenient backtest for Qlib users. Once it is ready, I will send a notification.

LITONG99 commented 2 weeks ago

Dear all, please note that the Qlib version is ready and we have updated the Readme.md. Thanks.