Introduction

Orca is a Python library specifically designed for backtesting statistical arbitrage strategies(“alpha” henceforth) on China A-Shares market.

This library is integrated with a delibrately structured MongoDB as its data source. It depends heavily on the high-performance Python library Pandas, thus it is recommended to get familiar with its basic data structures(Series, DataFrame, Panel).

To experiment with Orca, it is advised that one creates a virtual Python environment with virtualenv:

$virtualenv /your/local/directory

and install the following dependency libraries properly:

• Pandas
• PyMongo
• PyTables
• TA-Lib

Orca has 9 main components:

• MongoDB interface
• Data cache
• Alphas
• Operations
• Performance
• Universes
• Utilities
• Updaters
• Alpha DB interface

For backtesting purpose, one only have to focus on the first 7 components which are explained in detail in this documentation.