Algorithmic Trading Research PlatformΒΆ
The platform is built around the Zipline framework enabling to connect multiple data sources with multiple analytical approaches for conducting diverse studies. The figure provides a conceptual outline of its architecture.
Within this framework, the Pipeline is a custom collection of cross-sectional trailing-window tasks (Factors, Classifiers, and Filters) propagated by Pipeline Engine through the backtest interval. The Zipline data layer is organized as a collection of named data bundles associated with different data sets. The Flounder-Sharadar extension provides consolidated interface for ingesting and accessing bundles produced from the Sharadar data sets with daily prices, fundamentals, and metadata of US assets.