Today, we shipped an update that improves the quality of our FactSet data integrations prior to 2012. We recently learned of a bug that was causing a significant number of records to get dropped in our data integrations in the period from 2002-2012. The missing records would look like NaN or None in a pipeline simulation. Today’s update makes our system pick up the vast majority of the missing records. This was a systemic issue in our data processing, so we are filling in the missing data retroactively (as opposed to us keeping a point-in-time record of the missing values). Since we are backfilling the missing data, results of backtests or pipelines that use one of our FactSet data integrations and begin prior to 2012 may change if you re-run them. The changes in results will be due to your algorithm receiving more (and more correct) data than it was previously receiving.
If you have any questions or concerns, please let us know.
Jamie