mrauer / siklik

Identify cyclic continuous positive variations over a large quantity of stocks.

Sīklik

Sīklik is a small software designed to identify cyclic continuous positive variations over a large quantity of stock.

Current Version 1.0

Objective

The objective is to identify the subset of stocks that have had a large number of cyclic continuous positive variations over different steps, and % growth to guarantee diversification. A $1,000 target per asset can be build as portfolio. It will eventually be possible to build a model using these attributes in the future. The number of cycles and the growth rate will be a way to process the data in a higher dimension, and adding a layer of machine learning on top this summarized data should be a way to pseudo-predict the risk or at least weigh it.

Outline

In a nutshell, the sofware will perform the following tasks:

• Call the api.iextrading.com against a static list of stocks.
• Store date and close in memory.
• For each stock, and for an increasing value step, compute the number of continuous positive variations cycles, and store the average increase as well changeOverCycle. (ex: for a step 30, if the number of cycles is 8, and the changeOverCycle value is 4.34%, then the stock has had a monthly continuous growth of 4.34% for 8 months)

How to run it?

// Install the virtual environment
python -m virtualenv env
./env/bin/pip install -r requirements.txt    

// Run Sīklik
./env/bin/python main.py '{"action":"update"}'
./env/bin/python main.py '{"action":"run"}'

// Run the tests
./env/bin/pytest

API endpoint sample

https://api.iextrading.com/1.0/stock/aapl/batch?types=quote,news,chart&range=1m&last=10

Versions

0.1

• Virtualenv
• API key/initial call
• Module structure

1.0

• Scale to 1K+ stocks
• Workable output
• Testing platform

1.1

• Confidence score
• Automated R

Further development

• Predicting model
• Automation
• Python 3
• 3D clustering