Rolling returns
Returns computed over every possible overlapping window of a fixed length, removing start-date and end-date bias from performance assessment.
Trailing returns (e.g. "3-year return as of today") depend entirely on your chosen start date. If you happen to measure from a market trough, the number looks great. Rolling returns remove this cherry-picking problem by computing returns for every possible start date over a fixed window length, then summarising the distribution.
What it measures
For a 3-year rolling return analysis, MintByte computes the 3-year CAGR starting on every NAV date in the fund's history, producing hundreds of overlapping windows. The result is not a single number but a distribution — with a median, 25th percentile (worst-quartile), and 75th percentile (best-quartile). This gives a far more honest picture of how the fund actually performs across different market entry points.
How it is computed
For a rolling window of W years and a daily NAV series:
For each date t from t_start to (t_end − W_years):
CAGR(t) = (NAV(t + W) / NAV(t)) ^ (1/W) − 1
This produces a series of N CAGR values. MintByte then computes:
- Median rolling CAGR: 50th percentile of the series
- P25 (worst quartile): 25th percentile — "what the bottom 25% of investors experienced"
- P75 (best quartile): 75th percentile
- % of windows positive: fraction where CAGR > 0
Example: Mirae Asset Large Cap Fund, 3-year rolling over 8 years of history (≈1,460 windows):
- Median: 14.2% CAGR
- P25: 9.8% CAGR
- P75: 19.1% CAGR
- 100% of 3-year windows were positive
This shows dramatically more information than "3-year trailing return = 16.3%."
How to interpret
- High median + narrow P25–P75 band: Consistent fund — good returns regardless of entry point.
- High median but wide P25–P75 band: Volatile — good if you time entry well, painful if you don't.
- Low P25 (e.g. < 5%) or negative P25: Fund has delivered poor returns in at least 25% of all historical 3-year windows; a warning for investors with a fixed 3-year goal.
- % positive windows < 90% for a 3-year horizon: Insufficient return consistency for medium-term goals.
For SIP investors, rolling return analysis is especially relevant because SIP entries are spread across time — the distribution of rolling returns approximates the range of investor experiences better than any single trailing number.
Limitations + caveats
Rolling returns require a long history; a fund with only 3 years of NAV data has exactly one non-overlapping 3-year window and zero rolling spread. The analysis is also look-back limited — it cannot capture future regime changes. Funds that changed strategy or manager mid-history may show a distribution that blends two different regimes.
Related metrics
- Returns — point-to-point and trailing CAGR; rolling returns extend the same concept across all windows.
- Quartile Rank — where the fund sits in its peer group on a rolling basis.
- Volatility — the dispersion measure that explains why the rolling distribution is wide or narrow.
Sources
Daily NAV history: AdvisorKhoj API. MintByte maintains full NAV history for all AMFI-registered schemes. Rolling CAGR series recomputed monthly; minimum 5 years of history required to publish 3-year rolling statistics.