Rolling Returns vs Point to Point Returns in Mutual Funds
A point to point return measures performance between two fixed dates, such as the 5 year CAGR from 1 April 2019 to 1 April 2024. A rolling return repeats that calculation across every possible start date in a period, giving you hundreds of overlapping observations instead of one.
The difference matters because point to point returns depend heavily on the start date you pick. Shift the start by three months into a market bottom and a mediocre fund suddenly shows a spectacular five year number.
Rolling returns take that choice away from whoever is selling you the fund. They show the range of outcomes an investor could have had, not the one outcome that happened to look best.
How Point to Point Returns Are Calculated and Cherry Picked
Point to point CAGR is simple: take the ending NAV divided by the starting NAV, raise it to the power of one over the number of years, subtract one. SEBI’s standard performance format requires schemes to disclose 1, 3, 5 and 10 year and since inception returns this way, against the benchmark TRI, and those are the numbers on every factsheet.
The weakness is the anchor. A fund measured from a March 2020 style crash low will show a much higher CAGR than the same fund measured from a pre-crash peak, even though the manager did nothing different.
Watch for two tells in marketing material. One is an oddly specific period, such as “returns since 15 June 2020”. The other is a period that conveniently starts just after the fund’s worst stretch.
How Rolling Returns Are Computed
Pick a window length, say 3 years. Pick an evaluation period, say the last 10 years. Now compute the 3 year CAGR starting on day one, then starting on day two, then day three, and so on until the last date that still leaves a full 3 year window.
Over 10 years of data with daily steps, that produces roughly 1,700 separate 3 year observations. Each one overlaps its neighbours by almost the entire window, which is why rolling returns are described as overlapping.
Reading the distribution
- Median: the middle outcome, a better central estimate than any single point to point figure.
- Minimum and maximum: the worst and best three year stretch an investor could have caught.
- Percentage of observations below zero: how often a three year holding period actually lost money.
- Percentage above a threshold: for example how often the fund cleared 12% a year, which is often used as an equity expectation.
- Spread: the distance between the 10th and 90th percentile tells you how consistent the fund has been.
A fund with a median 3 year rolling return of 14% and a minimum of 2% is a very different proposition from one with the same median and a minimum of minus 9%, even if both show identical 5 year point to point CAGR.
| Feature | Point to point return | Rolling return |
|---|---|---|
| Observations | One per period | Hundreds, overlapping |
| Start date sensitivity | Very high | Removed by design |
| Shows consistency | No | Yes |
| Easy to cherry pick | Yes | Much harder |
| Where you find it | Factsheets, SEBI format | Research tools, you compute it |
The Limits of Rolling Returns
Overlapping windows are not independent samples. Two adjacent 3 year observations share almost all their data, so you cannot treat 1,700 observations as 1,700 independent trials or build confidence intervals from them. Use the distribution descriptively.
Rolling returns also inherit whatever period you feed them. A 10 year evaluation window on Indian equity that contains one long bull run will look flattering no matter how you slice it. Longer histories that include 2008, 2013, 2018 and 2020 style drawdowns are more informative.
And they say nothing about why a fund performed as it did. A fund manager change, a mandate change after SEBI recategorisation, or a fivefold growth in AUM can all break the link between past distribution and future behaviour.
Frequently Asked Questions
Which rolling window length should I use?
Match it to your intended holding period. Use 3 year rolling returns for a medium term equity view, 5 or 7 year for goals like a child’s education, and 1 year rolling only to study volatility rather than expected outcome. Shorter windows always show wider extremes.
Do rolling returns work for SIP investors?
They help, but the closer match is rolling SIP XIRR, which computes the internal rate of return for a monthly instalment series starting on every possible date. That captures the averaging effect of instalments, which lump sum rolling CAGR does not. Many research tools offer both.
Why do factsheets not show rolling returns?
SEBI prescribes a standard point to point format so that every scheme’s disclosure is comparable and cannot be dressed up with a chosen period. Rolling returns are an analytical layer on top, produced by research platforms or by you from NAV history. AMC websites publish daily NAV history you can download.
Can a fund have good rolling returns and still be wrong for me?
Yes. A small cap fund can show a strong rolling return distribution and still be unsuitable if you might need the money in two years, because its worst case window is deeply negative. Read the minimum and the below-zero share, not just the median.
How many observations do I need for the picture to be meaningful?
You need a scheme history comfortably longer than the window, ideally at least twice as long, so a 3 year rolling study wants 7 to 10 years of NAV data. A fund launched three years ago simply cannot produce a useful 3 year rolling distribution.
Key Takeaways
- Point to point returns describe one start date and are easy to cherry pick.
- Rolling returns repeat the calculation across every start date in overlapping windows.
- Read the median, the minimum, the share of negative windows and the percentile spread.
- Overlapping observations are not independent, so use them descriptively, not statistically.
- Match the rolling window to your own holding period before drawing conclusions.
