improved

November 2022 update

A small update this month, with the main highlights being a new BigQuery integration in Import Jobs, making data import easier for a wider range of data sets and partners; and a new DSL function to retrieve Visible Alpha revisions, and more generally point-in-time revisions for a given raw data signal.

New features

  • Import jobs: Added Google BigQuery as a new integration for data import.
  • DSL: Added a new va_revisions() function to retrieve Visible Alpha consensus revisions.
  • DSL: Added a new revisions_for_date parameter to graph_signal(), to retrieve point-in-time revisions for a given date.

Improvements

  • Signal Explorer: Added a "New chart" action to reset to blank state and a new share link action to share plotted signals and entities. (It does not share chart settings)
  • Starred items: Starred prediction models and portfolio strategies now appear in the navigation menus.
  • Library picker: Starred items are now easily available in the picker.
  • Import jobs: Allow setting the "to/from entity type" in relationship stages and "entity type" in time series stages, instead of requiring these to be column headers in the SQL output. This change is backwards-compatible and does not require update to existing stages.
  • Import jobs: Improved logging by reducing unnecessary log statements.
  • Import jobs: Added support for loading time series on the global entity by specifying entity type 'global'.
  • Import jobs: Imports of all resource types are now performed in batches. Large import jobs will no longer run out of memory.
  • DSL: Allow doing .group_by_entity(..) on a graph_signal(..) without setting the leaf_entity_as_label flag to False.

Bug fixes

  • Dashboards: fix bug where downloading a dashboard table with many columns failed.
  • Library: fix selection bar hiding items at bottom of list, when the Library is opened in a modal to select signals.
  • Fix incorrect company/entity count being displayed in the user interface, when selecting tags as well as individual companies/entities that were also present in the tags. This was previously only impacting the user interface and did not affect the underlying analytics.