Awesome

Readme

This is the official code and supplementary materials for our AAAI-2024 paper: MASTER: Market-Guided Stock Transformer for Stock Price Forecasting. [Paper] [ArXiv preprint]

MASTER is a stock transformer for stock price forecasting, which models the momentary and cross-time stock correlation and guides feature selection with market information.

Our original experiments were conducted in a complex business codebase developed based on Qlib. The original code is confidential and exhaustive. In order to enable anyone to quickly use MASTER and reproduce the paper's results, here we publish our well-processed data and core code.

Usage

1. Install dependencies.

• pandas == 1.5.3
• torch == 1.11.0

2. Install Qlib. We have minimized the reliance on Qlib, and you can simply install it by

• <code>pip install pyqlib </code> (Pip installation only supports python 3.7 and 3.8, please refer to its Readme.md.)
• pylib == 0.9.1.99

3. Download data from one of the following links (the data files are the same) and unpack it into <code> data/ </code>

• OneDrive link
• MEGA link
• Baidu link.

5. Run main.py.

6. We provide two trained models: <code> model/csi300master_0.pkl, model/csi800master_0.pkl</code>

Dataset

Form

The downloaded data is split into training, validation, and test sets, with two stock universes. Note the csi300 data is a subset of the csi800 data. You can use the following code to investigate the datetime, instrument, and feature formulation.

with open(f'data/csi300/csi300_dl_train.pkl', 'rb') as f:
    dl_train = pickle.load(f)
    dl_train.data # a Pandas dataframe

In our code, the data will be gathered chronically and then grouped by prediction dates. the <code> data </code> iterated by the data loader is of shape (N, T, F), where:

• N - number of stocks. For CSI300, N is around 300 on each prediction date; For CSI800, N is around 800 on each prediction date.
• T - length of lookback_window, T=8.
• F - 222 in total, including 158 factors, 63 market information, and 1 label.

Market information

For convenient reference, we extract and organize market information from the published data into <code>data/csi_market_information.csv</code>. You can check the datetime and feature formulation in the file. Note that m is shared by all stocks. The market data is generated by the following pseudo-code.

m = []
for S in csi300, csi500, csi800:
  m += [market_index(S,-1)]
  for d in [5, 10, 20, 30, 60]:
    m += [historical_market_index_mean(S, d), historical_market_index_std(S, d)]
    m += [historical_amount_mean(S, d), historical_amount_std(S, d)]

Preprocessing

:fire:[Update Detailed Description] The published data went through the following necessary preprocessing.

1. For features, we first perform RobustZScoreNorm, which computes median and MAD for each feature of all stocks in the training timespan for normalization. It then clips outliers as -3 and 3. When processing the test data, the median and MAD for each feature are estimated by (or borrowed from) the training data, so that we have no data leakage. We then use Fillna to fill the NA features as default value 0.

2. For labels, we first drop NA labels and 5% of the most extreme labels. Then, we perform CSZscoreNorm on labels. CSZcoreNorm is a common practice in Qlib to standardize the labels for stock price forecasting. Here 'CS' stands for Cross-Sectional, which means we group the labels on each date and compute mean/std across stocks for normalization. To mitigate the difference between a normal distribution and groundtruth distribution, we filtered out 5% of most extreme labels in training. Note that the reported RankIC compares the output ranking with the groundtruth, whose value is not affected by the label normalization.

An Alternative Qlib implementation

We are happy to hear that MASTER has been integrated into the open-sourced Qlib framework at this repo. We thank LIU, Qiaoan and ZHAO, Lifan for their contributions and please also give credits to the new repo if you use it.

As a brief introduction to the new version, with the Qlib framework, you can

• report AR, IR, and more portfolio-based metrics,
• modify experiment configuration with .yaml files,
• compare with various models from the Qlib examples collection,
• benefit from other merits of Qlib.

In the meantime, please note that

• The new version utilizes a different data source published by Qlib, which covers a different timespan. The new data source is considered logically equal to our published data but may differ in values.
• The new version uses stock universe CSI300 & CSI500, because qlib does not include a CSI800 dataset. Correspondingly, the representative indices to construct market information are different, it uses CSI100, CSI300, and CSI500, which is different from CSI300, CSI500, and CSI800 as in this repo.
• The new version does not include the 'DropExtremeLabel' operation in data preprocessing but also reports decent performance.

Cite

If you use the data or the code, please cite our work! :smile:

@inproceedings{li2024master,
  title={Master: Market-guided stock transformer for stock price forecasting},
  author={Li, Tong and Liu, Zhaoyang and Shen, Yanyan and Wang, Xue and Chen, Haokun and Huang, Sen},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  volume={38},
  number={1},
  pages={162--170},
  year={2024}
}