Rolling

Overview

Rolling window measures aggregate data over a sliding time period. Use rolling_window to define moving averages, cumulative totals, and time-based comparisons.

Example

measures:
  - name: rolling_7_day_orders
    type: count
    rolling_window:
      trailing: 7 day
      offset: start

  - name: rolling_30_day_revenue
    type: sum
    sql: amount
    rolling_window:
      trailing: 30 day

Syntax

rolling_window Properties

Property	Description
`trailing`	Time period to look back (e.g., `7 day`, `1 month`)
`leading`	Time period to look forward (rare)
`offset`	`start` or `end` — which end of the window to align

Trailing Window

- name: rolling_7_day_avg
  type: avg
  sql: amount
  rolling_window:
    trailing: 7 day

Cumulative (Running Total)

Use unbounded for cumulative calculations:

- name: cumulative_revenue
  type: sum
  sql: amount
  rolling_window:
    trailing: unbounded

Offset Options

# Window ends at current row (default)
rolling_window:
  trailing: 7 day
  offset: end

# Window starts at current row
rolling_window:
  trailing: 7 day
  offset: start

Common Patterns

7-Day Moving Average

- name: daily_orders
  type: count

- name: rolling_7_day_avg_orders
  type: avg
  sql: "\{daily_orders\}"
  rolling_window:
    trailing: 7 day

Month-to-Date

- name: mtd_revenue
  type: sum
  sql: amount
  rolling_window:
    trailing: 1 month
    offset: start

Year-to-Date

- name: ytd_revenue
  type: sum
  sql: amount
  rolling_window:
    trailing: 1 year
    offset: start

Requirements

Rolling window measures require a time dimension in the query
Works best with pre-aggregations for performance