FinReport: Explainable Stock Earnings Forecasting via News Factor Analyzing Model

PyTorch implementation for [WWW'2024] "FinReport: Explainable Stock Earnings Forecasting via News Factor Analyzing Model"

​Important Notice​

This repository contains a ​demonstration version​ of the code and data, designed to help readers understand the model architecture and workflow. Due to commercial agreements and technical confidentiality constraints after the publication of the paper, we are unable to open-source the full implementation or the complete dataset used in the experiments. The provided code and data are for illustrative purposes only and ​do not represent the actual implementation or dataset used in the paper.

Example Reports

Obtaining Factors

There are two ways to get the factors:

First Way

The first way relies on the stockstats library, which is a wrapper for pandas dataframes.

pip install stockstats

Commonly you can input a pandas dataframe with the following columns: ['date', 'open', 'close', 'high', 'low', 'volume']to use the wrap function to get the factors.

import pandas as pd
from stockstats import wrap

data = pd.read_csv('stock.csv')
df = wrap(data)

Check the documentation for stockstats for more details. More details and examples can also be found in src/Tech_Indicators.py.

Second Way

The second way to obtain the factors is to API from platforms like IFind, which provides a variety of factors. We recommend using this method to obtain more factors more esaily.

Codes are available in src/IFind_Indicators.py. You can run it after filling the access_token in Line 10, customizing necessary parameters of form_data in Line 14.