This project is work in progress. It's purpose for me is to learn some new things in python, c++ and devops.

marksim is a package that helps to detect anomalies in unbalanced panel data using markov chain simulations. More about what is a panel data here.

The inspiration mainly comes from the following paper:

• Henderson, A.D., Raynor, M.E. and Ahmed, M.: 2012, How long must a firm be great to rule out chance? benchmarking sustained superior performance without being fooled by randomness, Strategic Management Journal 33, 387-406 link

The first version of this code was used in this thesis. This repository contains a cleared and documented version of what I had initially.

Let's say you have an unbalanced panel containing the growth rates of 1 mln. firms in a country in the last N years (could be any discrete periodic variable). Let's also suppose that you observe 10 firms that grow that fast that they are in the top 1% of growers each year. In other words, these firms grow very persistently. Think of google or amazon or any other successful company being on news.

marksim helps to answer the following questions:

• Is this number 10 random?
• Given the dynamics of firm's growth rates in the data, how many companies on average can we expect being so persistent?

here is my plan:

• python implementation of the calculation of transition probability matrix
• python implementation of the simulation module
• python implementation of the analytics module
• release package on PyPI
• docker image
• django app with basic user interface
• c++ implementation of simulations

Package is a work in progress, so for now please install with

pip install .

For more detailed documentation, please see here

All dependencies are specified in requirements.txt file.

pip install -r requirements.txt

pip install -e.

To run tests from the root folder do

python -m unittest discover -s src/tests

I use SemVer for versioning. For the versions available, see the tags on this repository.

• Vladimir Korzinov - Initial work - vvkorz

This project is licensed under the MIT License - see the LICENSE.md file for details