Rolling Returns vs Trailing Returns: How to Compare Mutual Funds Fairly

Two funds show a five-year return of 12% a year. It is tempting to treat them as equivalent. Yet that number describes only the period ending on the displayed date. Investors who started a year earlier, or needed to exit a year sooner, may have experienced very different outcomes.

This article uses an invented NAV series you can reproduce in a spreadsheet. It makes no claim about the performance of a named fund. The purpose is to help you read research tables, ask better questions and avoid choosing an investment because one favourable date range makes it look exceptional.

What are trailing returns?

A trailing return looks backward from a specific end date. A three-year trailing return shown as of a particular month-end measures the change from three years before that date to that month-end. For a multi-year, single-investment comparison without intervening distributions, it is usually expressed as compound annual growth rate, or CAGR.

For exactly three years, CAGR = (ending value ÷ beginning value) raised to the power of 1/3, minus 1. Multiply the result by 100 to express it as a percentage. For irregular dates, use a stated day-count convention rather than pretending every observation is exactly three years apart.

Trailing returns are useful. They provide a quick comparison and can show how a fund behaved over recent one-, three- or five-year periods. The limitation is that each displayed result depends on one starting point and one endpoint. A strong rebound from an unusually depressed starting value can produce an impressive headline without proving consistent superiority.

Check whether a short-period return is absolute or annualised before comparing it with a longer one. A six-month gain of 8% and a three-year CAGR of 8% are different measurements. The period label and annualisation basis are part of the result, not optional footnotes.

What are rolling returns?

Choose a holding period, such as three years, and move the starting date forward repeatedly while keeping that period fixed. Calculate the return for each window. The output is a series; the mean, median, minimum and maximum are summaries of that series, rather than separate definitions of rolling returns.

DSP's rolling-return guide explains this repeated-window approach and its use alongside point-to-point returns. Before reading a chart, identify three inputs: the holding period, the full observation span and how often the window moves.

A claim such as five-year rolling return of 13% is incomplete unless it identifies which summary statistic is being shown and over what history. Is 13% the mean, median, latest observation or best window? Each can be correct while communicating a different story.

Worked example: calculate four three-year rolling returns

Assume the following Growth-option NAV observations occur at exact one-year intervals, with no distributions. Year 0 through Year 6 are mathematical labels, not a claim about actual market dates. Annual spacing keeps the example readable; real research commonly uses more frequent observations.

The first complete three-year window runs from Year 0 to Year 3. Move it one year forward for each subsequent calculation. Seven annual observations spanning six years produce four complete three-year windows, including both endpoint windows.

At Year 6, the latest three-year trailing return is 11.46%. The rolling analysis adds three other outcomes and shows a range from 7.72% to 15.87%. The mean of the unrounded observations is approximately 11.05%, and the median is 10.30%. Use unrounded values for the summaries, then round the displayed results.

Notice the strongest window begins at the low Year 2 NAV of ₹90. The full history includes a fall from ₹120 to ₹90 between Year 1 and Year 2, a 25% decline. Positive three-year outcomes in the table do not mean investors avoided uncomfortable losses during those holding periods.

What to read beyond the average

In the invented example, none of the four windows is negative. Reporting that as a 100% chance of positive returns would turn a tiny historical sample into a false forecast. The minimum is simply the smallest value observed under this dataset and method.

A benchmark win rate also needs context. If a fund wins many windows by 0.1 percentage point but loses a few by 5 percentage points, the win rate alone hides the magnitude of underperformance. Compare the distribution of excess returns and the investment risks, rather than treating every win as equally valuable.

For a goal requiring an assumed return, you can calculate the share of historical windows that cleared that assumption. Label it as a historical threshold count. It is useful for challenging an optimistic plan, but it is not a calibrated probability of achieving the goal.

Overlapping windows are not independent experiments

A five-year window ending today and another ending tomorrow share almost their entire history. Daily rolling calculations create many observations, but they do not create thousands of independent five-year market experiences. The outcomes can be strongly related because most of the underlying days are the same.

This matters when a chart says a fund beat its benchmark in 92% of windows. That can describe the chosen sample accurately, yet it should not be translated into a 92% chance of future outperformance. A long market regime may affect a large number of overlapping observations simultaneously.

Try reading daily or monthly results alongside non-overlapping periods and distinct market phases. Each method has limitations: non-overlapping samples are smaller and still depend on the chosen starting point. The objective is to see whether a conclusion survives reasonable changes in the measurement setup, rather than search for the version that gives your preferred fund the best number.

Compare the same plan, option, dates and benchmark

1. Use an identical holding period and rolling frequency for every fund.
2. Restrict the comparison to a common observation span when histories differ. Show longer individual histories separately.
3. Keep Direct and Regular plan comparisons consistent, because their costs and NAVs differ.
4. Use Growth NAVs or a properly distribution-adjusted return series. Raw IDCW NAV changes omit distributed cash.
5. Choose a benchmark appropriate to the fund mandate and match the exact window dates.
6. Disclose gaps, scheme mergers, mandate changes and other breaks that affect interpretation of the history.

A fund's ten-year record compared with another fund's three-year record can tell you more about different market periods than about management quality. Similarly, a small-cap fund and a large-cap index fund take different exposures; the higher historical return alone does not settle which belongs in a particular investor's portfolio.

For equity benchmark comparisons, understand the distinction between a price index and a total return index. NSE Indices explains that a TRI includes dividend receipts as well as price changes. Comparing a fund that retains portfolio income with a price-only index can flatter the apparent outperformance.

An index series is also not a frictionless product you can buy directly. An actual index fund has expenses and tracking differences. Use the benchmark to assess the strategy, and compare investable alternatives separately when deciding how to implement an allocation.

Rolling returns, CAGR and SIP XIRR answer different questions

Rolling returns and CAGR are not rival formulas. A multi-year rolling analysis commonly calculates CAGR repeatedly. The difference from a trailing CAGR is the number of windows, not a new source of investment gains.

A SIP involves multiple contributions at different dates. Its investor cash-flow return is usually measured with XIRR. A fund's five-year rolling lump-sum return does not equal the return on a five-year SIP, because the later SIP contributions were invested for less time.

You can analyse rolling SIP outcomes, but that requires defining a SIP schedule for each window and computing its cash flows. State the contribution amount, frequency, transaction-date convention, final valuation and whether costs or taxes are included. Calling an ordinary rolling NAV CAGR a SIP return would be a category error.

Use the CAGR calculator to check a single multi-year window. Use the XIRR calculator for your dated cash flows. Our historical SIP tool uses actual NAV history for a selected scenario; a single backtest is still one scenario, not an analysis of every possible entry date.

What a rolling-return table cannot tell you

A return distribution does not show everything that happened inside each window. Two investments can finish with the same CAGR even if one suffered a much deeper interim drawdown. If you might need the money before the intended endpoint, that path can matter more than the end-to-end figure.

• Liquidity and redemption restrictions: a return table does not establish when cash can actually be accessed.
• Credit, concentration and valuation risks: a favourable history does not remove present exposures.
• Survivorship bias: a list containing only funds that exist today can omit failed, merged or closed alternatives.
• Strategy continuity: historical results may span different managers, mandates or operating conditions.
• Investor behaviour: withdrawals, missed contributions and switching decisions can make personal results differ from a model.
• Future regimes: no minimum observed historical return sets a floor on future losses.

A younger fund may simply lack enough completed windows to assess a long holding period. Mark those results unavailable. Substituting the parent brand's history or another scheme's returns creates an appearance of evidence that the fund itself does not have.

A practical research workflow before choosing a fund

1. Define the goal, expected holding period and ability to tolerate losses before screening returns.
2. Shortlist comparable mandates and read the current factsheets, risk disclosures and costs.
3. Use trailing results to understand the recent period, then inspect rolling outcomes across a sufficiently broad common history.
4. Compare the median, weak windows, benchmark-relative results and drawdowns, with the methodology stated.
5. Inspect portfolio overlap and concentration so the chosen fund contributes something useful to your existing holdings.
6. Model the goal using several return assumptions and a contribution plan you can maintain. Keep historical research separate from a future promise.

The fund comparison tool is a starting point for the measures it displays; do not relabel its point-to-point return figures as rolling returns. Use an AMC's disclosed rolling analysis or a transparent calculation from verified data when you need that specific measure. AMFI's NAV portal is an official starting point for checking published NAV information.

Follow with the overlap checker and goal-planning calculator. If you want help interpreting a fund comparison, contact Mahesh Jain MFD with the actual table, observation dates and the goal it is meant to support. A useful discussion begins with the evidence behind the return, rather than the largest percentage on the page.