-
Project summary: Using the stock market data and data science methodologies (including NLP, predictive modeling and classification model ) to generate strategy. Major steps are:
- Select the major investing horizon as the days befor and after earning release.
- Using a classification model to predict whether it would be a wise choice to long or short.
- Using NLP to use the related earning release date information, building up the advanced model on how to trade.
-
Executive Summary: We combined core value of 8 published articles related to Earnings Release and improve it with machine learning strategy. Finally, we got two models:
- long-before earning release
- short-after earning release
-
The model is based on one assumption:
- Before earning release, stocks with strong momentum will attract people to buy since people believe that the Earnings Report exceed expectations, the price will increase in a short period. And after earning release, some people will cash in and the price will drop in a short period. In summary, long before Earning release, short after it.
-
Our model could tell you, if you long the stock before earning announcement, whether you could a positive return. If you short the stock after earning announcement, whether you could get a positive return.
-
Initially, we use 1000 stocks data in 10 years, got 963 K observation and use forward stepwise to select the most important two factors which are momentum. This step is to find a pattern of stocks which are sensitive to Earning announcement.
-
Based on the pattern we found in step 1, we selected fewer observations and use machine learning model to do classification and got 0.88 accurate rate with Random Forest
Contribution statement: (default) All team members contributed equally in all stages of this project. All team members approve our work presented in this GitHub repository including this contributions statement.
Following suggestions by RICH FITZJOHN (@richfitz). This folder is orgarnized as follows.
proj/
├── lib/
├── data/
├── doc/
├── figs/
└── output/
Please see each subfolder for a README file.