NEW
Data Table Versioning
Data tables are now automatically versioned, as needed. This ensures that projections always point to the data tables with the original data they were using at run time. These older tables can then be viewed in read-only mode in the editor by clicking on the table name from the projection. Note that this does not yet apply to tables that are nested within other tables.
Tables will be automatically updated to the latest version on a projection when duplicating or re-running the projection. Older versions of tables are also now available for selection both on templates and projections, to enable users to replicate older results.
IRR Error Handling + Defaults
We’ve added the ability to warn or even ignore IRR failures (instead of causing an error), with a fallback formula to be used in the case of non-convergence.
Projection Setting – Output All Results
A new setting, “Output All Results”, has been added to projections and templates. This setting will likely be much more common than other options going forward when debugging/troubleshooting. This setting enables all variables that were calculated to be output to the reporting module, including results that were calculated before/after the projection period, allowing for easier troubleshooting, particularly for liabilities that project back to issue date or forward past the end of the projection. Note that this setting is very performance-intensive, and should only be used for troubleshooting purposes.
Projection Error Limit
An error limit has been added to projections and templates within the run settings dialog. This error limit allows the projection to quickly fail in the case in which there are a large number of errors, to save run time. The default value has been set to 100 for all existing projections and can be set between 0 and 1000.
IMPROVEMENTS
- Projection names and tags can now be edited in any status, to allow for better organization of projection results.
- Files can now be renamed in the file manager.
- ‘Select All’ functionality has been improved in the Model Execution and Model Results module so that if using a tag filter, only those projections are selected. A selection count has also been added to make this behavior more clear.
- Duplicate error messages for the same variable chain have been eliminated in the run logs to make it easier to see all relevant errors.
- The column headers on the data and mortality table editors are now fixed.
- Improved error messaging around invalid values when calculating scenario tables and attempting to perform the modulus operator on a zero denominator.
- Improved validation messaging around product output groupings, and automatically delete an output grouping upon deletion of the last variable in the grouping.
- Validation messages around various Display Precision fields has been improved to make it easier to understand what needs to be fixed.
- The projection/template setting that was previously called “Output All Variables” has been renamed “Calculate All Variables” to clarify its intent. No changes have been made to how this behaves.
- Added instructions to the table editing landing page to make it more clear how to proceed.
- Minor improvements around navigation for ease-of-use.
FIXES
- Fixed behavior when copying and pasting percentage values from excel into the data table editor.
- Tooltips now appear again on icons in the formula builder headers to show what each icon means.
- Empty values in table index results now report a proper error instead of an unknown error in the run logs.
- Fixed some issues around the array value system variable, array variables referenced across levels, using numbers as array values, and caching of array variables during runs.
- The Force Recalculation option now works appropriately when a product is run in static mode.
- Fixed an issue that sometimes caused products to not be copyable if they contained array variables.
- Improved behavior when changing variables between time-indexed and not time-indexed.
- Fixed the ability to save a model point reference with no model point specified (this will result in a validation message).
OPERATIONS
- We’ve upgraded some backend libraries and made some changes to improve stability and performance in various areas of the application.