When Things Get Weird with Custom Calendars in Tabular Models

Calendar-based Time Intelligence in Power BI and Fabric Tabular models can produce unexpected results and pitfalls. Issues arise when comparing periods of unequal length, such as a leap year causing previous-year calculations with DATEADD to be shifted by a day. A potential workaround for this is to calculate the previous year by moving back 12 months instead of one year. The content also details strange inconsistencies with weekly calculations that may require rebuilding the data model and discusses the challenges of mixing weekly and monthly logic in the same calendar.