This repo contains an actual version of the master disertation beating_stock_market.
All codes are documented in spanish following numpy style guide.
You can see what to expect from this project reading in this medium post Machine Learning, Fintech & Trading | by: Esteban Sánchez | Medium or its spanish version Machine Learning, Fintech & Trading | ESP | by: Esteban Sánchez | Medium.
First of all you will need a free or premium API key from Alpha Vantage that you have to write on line 7 of constants.py
You have to decide how many tickers or over which tickers you want to use the app, you can change that on lines 59 and 60 of constants.py
. There is a notebook called get_symbols_list.ipynb
where you can find all the ticker list available in the API Finnhub to use as SYMBOLS.
Last thing you have to do before to start using it is to unzip the zip files inside ./data/model
typing unzip file_2_unzip.zip
To run the app you have to activate the environment and then type:
python index.py
Numbers in red is to navegate between pages.
This view is to update close price data clicking on the left button or technical indicators clicking on the right button.
This will update tickers in SYMBOLS variable of constants.py
- Choose a ticker to analyse.
- Data range to see previous data.
- Choose technical indicators to analyse.
- Candlestick chart.
- Technical indicators chart.
- Days to deploy the pct_change.
- Pct_change chart.
- Box-plot of the distribution of the pct_change
- Choose a model.
- Choose the category to predict.
- Choose a ticker.
- Choose days to predict.
- Table with past data description on each category.
- Radar-plot to compare the actual ticker with the past values. White line is the actual prediction.
- Prediction of the actual ticker for the chosen date.
- Cosine similarity for the actual prediction with the past data.
- Table with the 50 highest tickers in relation to cosine similarity