The objective of this project is to create a prescriptive model with the aim of informing the user about the best timing to buy or sell stocks based on parameters set by the user.
The product is based on a Forecasting model that will predict the stock price (user's choice) for the next 10 business days. Based on these values it will make recommendations for buying or selling, based on strategies (simple crossing of moving averages).
This project is composed by the main features:
- Mutable list of ticker symbols. You can add how many ticker symbols you want. Just make sure they follow the Yahoo Finance naming standard.
- Forecasting of each ticker symbol using XGBoost (adding more models is WIP). The features are:
- Time known features: day, month, year, quarter, week
- Moving average features. 3 days moving averages default, but you can add how many you want
- Lag features. Lag 1 default, but you can add how many you want.
- Model testing in the last 10 business days.
- Simple trading algorithm based on crossing of moving averages.
- Front-end page where you can change the timeframe, colors and moving averages of the trading algorithm.
Use the following code to install the project locally:
git clone https://github.com/micheldearaujo/forecasting_stocks.git
cd forecasting_stocks
python3 -m venv forecasting_stocks
source forecasting_stocks/bin/activate
make installAfter installing the project, all one need to do is to run the data pipeline:
# Data pipeline
python3 src/data/make_dataset.py
python3 src/features/build_features.pyThen, just run the modeling pipeline
python3 src/models/train_model.py
python3 src/models/test_model.py
python3 src/models/predict_model.py
streamlit run src/models/app.pyLast but not least, launch the Streamlit interface:
streamlit run src/app/app.py