This web site has been retired. Please follow my activities at

the personal site of patrick paul


This page showcases a few of my recent undertakings.

I'm the CEO and lead software developer at Gridpar LLC

Gridpar LLC #

Est. August 2013 — Detroit, Michigan

Gridpar LLC is a technology startup in the @Bizdom accelerator developing software and web applications for the solar industry.

Our first product, Gridpar.js™, is an embeddable solar calculator widget that solar installers and publishers can use to attract sales leads from homeowners that want solar.

Guake Tabs Indicator for GNOME Shell 3.8

Guake Tabs Indicator #

Version: 3

Guake Tabs Indicator is an extension for GNOME Shell 3.8 which displays terminal tabs on the GNOME top panel.

The project source is available on Github and the extension is available at the GNOME Shell Extensions site. Twin Pints Cooperative

Twin Pints Cooperative #

Status: Fundraising

Twin Pints Cooperative is a start-up intended to operate a brewer incubator and brewpub in an entertainment district in Lansing, Michigan, which will engage craft beer drinkers, homebrewers, and artisan brewers through a variety of services and experiences. These include a homebrewing tool library, a cooperatively-owned microbrewery line, and a tap room open to the public.


Librewary #

Status: Under development

Librewary is an online equipment lending management system written in the Python django framework.

Originally conceived as the solution for Twin Pints Cooperative’s homebrewing tool library information management needs, I developed the application with a view to making it fully extensible to similar use cases involving lending equipment.

Licensed as open source dually under the MIT license and GPL.

Midas Backtester

Midas Backtester #

Status: Frozen

Midas Backtester is a historical stock market backtester written in Python with Zope ZODB, wxPython GUI, and matplotlib.

It can do cool things like screen stocks by fundamental-analysis metrics, simulate a portfolio using the aforementioned stock screens to test profitability, and even algorithmically optimize an idealized portfolio strategy.

Of course, the data sets I work with are limited, my algorithms are flawed, and the features and configurability are lacking – but it is a valid proof-of-concept and is similar to work done every day every second by hedge funds.

I am not currently working on this project, and am happy with the mere working prototype, but the next feature I would like to implement (i.e. expose to the genetic algorithm) is position sizing, as well as some sort of macroeconomic binary on-off switch for having a given portfolio agent invested in stocks or waiting in cash. I also plan to move the data and computing to the cloud using boto to speed things up.

Pretty screenshots and other records of the project can be found at my old website.