What Are Rolling Returns?
Rolling returns compute the return for every possible holding period of a given length within a time window. If you pick "1Y rolling over full history", we calculate the 1-year return starting from every single trading day in the fund's history.
This gives you the full distribution - the best case, worst case, average, and median - instead of just one snapshot return that depends on your start date.
How It Works
Rolling return = (NAV on d) / (NAV on d − window) − 1
For windows ≥ 1Y, this is annualized to CAGR.
We use an ASOF JOIN in DuckDB to find the nearest prior NAV when the exact lookback date falls on a holiday.
Two Time Concepts
Rolling window = the holding period length (3M, 6M, 1Y, 3Y, 5Y).
Evaluation period = the time range over which we compute rolling returns (Full History, Last 1Y, Last 3Y, Last 5Y, or Custom).
The evaluation period must be longer than the rolling window. You can't compute 5Y rolling returns over a 3Y evaluation window - there wouldn't be enough data.
Worked Example
1Y Rolling Returns, Last 3Y Evaluation
We look at every trading day in the last 3 years. For each day, we compute the 1-year return ending on that day. This gives us ~500+ return observations.
Statistics: Avg 14.2%, Median 13.8%, Min −8.1% (Mar 2023), Max 42.5% (Apr 2024).
The Min tells you the worst 1-year stretch, the Max the best, and % Negative tells you how often you'd have lost money holding for exactly 1 year.
Statistics We Show
- Avg: Mean of all rolling return observations
- Median: Middle value - less affected by extreme returns
- Min / Max: Worst and best rolling return, with dates
- Std Dev: Volatility of rolling returns
- % Negative: Percentage of rolling periods with negative returns
- Beat %: Percentage of rolling periods where the fund outperformed the benchmark (Nifty 50). See the dedicated Beat Rate tab for details.
- Obs: Number of rolling return observations computed
Edge Cases
- Short-window returns (3M, 6M): Shown as absolute return, not annualized, because annualizing a 3-month return can exaggerate seasonal patterns.
- Invalid combos: When evaluation period ≤ rolling window (e.g., 5Y rolling over Last 3Y), the button is disabled.
Related metrics
More Returns methodology from the MFPRO analytics tool: