Rebalance Ep 15: Principle Component Analysis

Rebalance is a weekly flash briefing of new features and updates for you, our QC community. In our 15th episode we’re happy to share:

  • An increase in backtest speed when using tradebar type data! We’ve implemented a new way of parsing raw data into data points, avoiding in-between sub-strings and parsing. This improves performance and reduces resource consumption. Look for the feature expanding to other data formats soon!
  • Updated docs! We released more documentation on methods to create, update, track and cancel trading orders. Use it to better understand how your trades interact with brokerages via order tickets in your algorithm. Check out the new documentation here.
  • The latest Morningstar industry classification codes! You can use our updated helper classes to access the latest field definitions. If you’re running a live strategy we recommend redeploying your algorithms to pick up the latest updates.
  • A principal component analysis-based template! In this week’s “From Research to Production” post, Jack walks us through a strategy using principal component analysis to identify the dimensions of data, or number of components, that contribute most to data variance. We can use PCA transformed data to build regression models, reduce the number of factors in our model to avoid overfitting, and make predictions. Find the template and clone it here.
Sherry Yang

By: Sherry Yang

21.11.2019