A financial forecasting tool for Canadians planning retirement.
The project is designed as a library for use by client code.
It provides a Forecaster object which generates Forecast
objects based on some set of strategies (e.g. for determining
living expenses, debt repayment, contributions and withdrawals,
etc.), assumptions about future market returns, and some
Person and Account objects representing the plannees and
their assets and debts. The end result is that you can track
the plannees' financial picture from year-to-year at low- or
high-resolution (e.g. net worth vs. specific account balances).
The top-level modules are all generic and designed to be extensible and country-agnostic. For example, Canadian functionality is carved out into a subpackage to show how to extend the code via subclasses. Other countries can be added by client code in the same way.
This is a let's-learn-Python project. It targets Python 3.7. For the most part, it adheres to Google's Python style guide and uses pylint to keep it that way. (The big exception is its use of certain "power features", such metaclasses, which are too fun not to use.) Along with learning Python, one of the goals is learning best practices, so it also aims to provide extensive comments and tests, provide informative commit messages, and just generally maintain a tidy little package of nicely-documented code.
This project is grew out of an Excel spreadsheet. Most of the relevant financial concepts are taken from there (e.g. elgibility ages, benefit rates, tax rates, etc.), but the design of the Python project is all-new. The Excel sheet was terribly inefficient for Monte Carlo simulations, so one of the goals of the software is to implement these at least somewhat efficiently.
As noted above, enabling efficient Monte Carlo simulations (across many simulated futures and/or backtesting to historic data) is an eventual goal. Providing a basic UI (e.g. a web UI) is also on the roadmap - although that might wind up being an entirely new package!
In the meantime, though, the goal is to provide a minimally-functional working implementation and then expand it version-after-version until it reaches feature parity with the original Excel spreadsheet. ("Parity" might be the wrong term - the package already does lots of things that the spreadsheet didn't do, but the goal is to make this package's functionality a strict superset of the spreadsheet's.)
The GitHub repo has a list of milestones that provides a high-level overview of each version's functionality. There's also a ReadTheDocs instance with documentation (generated with sphinx-apidocs).