Open any fund-rating site. The first column is a number. Three months. Six months. Twelve months. It is the most-clicked column on the page and the least informative. Across 30 years of Indian equity history, the 1-year rolling return on the Nifty 50 itself has ranged from −57% to +104% — a 161% spread with no fund manager involved. The number you call a track record is mostly a draw from this distribution. The skill, if any, is buried in the rounding. This paper walks through the empirical proof, the academic literature, and the horizon at which a track record actually starts saying something.
A retail investor opens a fund-screener. The default sort is 1-year return. The first fund is up 38%. The investor invests. Six months later, the fund is up 4%. The investor is unhappy. The fund manager has not changed. The mandate has not changed. The team has not changed. Only the market has changed.
This is the central confusion of recency-based selection. The investor believed they were buying a manager. They were buying a market regime. When the regime turned, the "manager" appeared to fail.
The literature has been clear on this point for thirty years. Carhart (1997) showed that what looks like persistent fund skill is mostly momentum-factor exposure. Fama & French (2010) showed that after costs, almost no active managers have positive alpha distinguishable from luck. S&P's annual SPIVA scorecards have, for two decades and across geographies, found that ~85% of active funds underperform over a decade — and that survival itself is a selection effect.
Across 30.5 years of daily Nifty 50 closes (Nov 1995 → May 2026), there are 7,334 overlapping 1-year windows. The chart below is their distribution — without any fund manager attached. This is what the index alone delivered.
Computed from 7,586 daily Nifty 50 closes, 06-Nov-1995 to 07-May-2026. 1-year window = 252 trading days; rolling daily. Bins are 10 percentage points wide. Source: NLE backtest engine, NSE archive + yfinance.
Two things on this chart matter more than any fund return.
First, the tails are long. A 1-year window can deliver +50%, +70%, even +100% — and can also deliver −30%, −40%, −57%. None of this requires manager skill. None of it predicts anything about the next 1-year window.
Second, the centre is wide. The interquartile range alone is roughly −1% to +27% — a 28-percentage-point band that contains half the windows. Any fund return inside this band has, by definition, told you almost nothing.
Performance attribution literature (Fama-French, Carhart, Brinson) decomposes any fund return into four blocks. Their relative magnitudes, in any 12-month window, look approximately like this:
The cleanest test of "is past performance a signal" is persistence: does a top-quartile fund stay top-quartile?
The empirical answer, across two decades of US and Indian data, is almost no.
| Top-Quartile Funds in Year T | Stay Top-Quartile in T+1 | Stay Top-Quartile Through T+5 |
|---|---|---|
| If returns were pure skill | 100% | 100% |
| If returns were pure luck | 25% | ~0.4% |
| S&P SPIVA (US, 2002–2024 average) | ~28% | ~3% |
| S&P SPIVA India (2018–2024 average) | ~26–32% | ~2–5% |
Source: S&P Dow Jones Indices — SPIVA (Standard & Poor's Indices Versus Active) Persistence Scorecards, US & India editions. Indian data subset is shorter (2018+) but consistent in shape with the longer US series. Carhart (1997) was the first paper to formally document this collapse.
This is the deepest statistical fact in fund selection. The thing investors most want past performance to mean — "this manager is better than the others" — is the thing past performance is least able to demonstrate at the horizons retail investors actually look at.
There is a second, structural problem with chasing recent track records — one that is specific to the Indian small-cap and mid-cap segments where retail enthusiasm runs hottest.
A fund earns a 3-year track record. Inflows arrive. AUM doubles, then triples. The strategy that produced the track record cannot deploy the new capital without violating its mandate — the small-cap universe is finite.
Indian small-cap fund AUM expanded ~5x between September 2020 and September 2025. The hunting ground did not. SEBI's 2025 stress tests now show that the largest small-cap funds need 36 to 57 days to liquidate just half their portfolios under stress. The funds with the most attractive 3- and 5-year returns are the same funds that have, in 2024–2025, capped fresh SIPs and suspended lump-sum subscriptions in their flagship schemes.
Performance attracts capital. Capital erodes performance. The retail investor who buys the chart-topper is buying the fund at the moment its strategy is most strained — and selling the underperformer at the moment its strategy is freshest. The recency-chaser is structurally on the wrong side of the AUM Capacity Curve.
Before going further, a clarification that quietly does most of the analytical work in this paper.
Every fund-rating site, every screener, every monthly factsheet shows trailing returns. The number depends entirely on the day you look. A fund that ended its 5-year window in late 2007 looked extraordinary; the same fund six months later looked terrible. Nothing about the strategy had changed — only the endpoint had.
Every chart and table in this paper uses rolling returns. That is what allows the question "is a 1-year track record meaningful?" to have an answer at all. Asked of trailing returns, the question collapses to "meaningful relative to which day?" — and there is no way to escape the endpoint without rolling.
The same Nifty 50 dataset that produces the wild 1-year distribution produces something very different at longer horizons. Holding period matters more than fund choice.
| Holding Period | # Rolling Windows | % Positive | % Negative | Worst Window |
|---|---|---|---|---|
| 6 months | 7,460 | 67.2% | 32.8% | ~−42% (2008–09) |
| 1 year | 7,334 | 73.8% | 26.2% | −57% (2008–09) |
| 3 years (CAGR) | 6,830 | 91.7% | 8.3% | −17% CAGR (2000–03) |
| 5 years (CAGR) | 6,326 | 95.4% | 4.6% | negative, narrowly |
| 7 years (CAGR) | 5,822 | 98.8% | 1.2% | ~0% CAGR |
| 10 years (CAGR) | 5,066 | 100.0% | 0.0% | positive in every window |
Source: NLE backtest engine, daily Nifty 50, Nov 1995 → May 2026. Each row is the full set of overlapping rolling windows of that length. % Positive is the share of windows ending with a positive total return (or positive CAGR for ≥3y).
The implication is uncomfortable but unambiguous.
A 6-month return is one-third noise. A 1-year return is one-quarter noise. A 3-year return is one-twelfth noise. A 10-year return finally has the market doing the same thing in every realisation — only at that horizon does fund-level dispersion show up clean enough to interpret.
Look at 7- and 10-year track records. Be sceptical of 3-year. Ignore 1-year. Treat 6-month and 3-month figures as noise dressed up as data.
None of these five inputs is on the front page of a fund-screener. All five matter more than the number that is.
"A fund return is not a manager return. A 12-month fund return is roughly eighty percent market, fifteen percent factor, five percent noise, and a sliver of manager skill that is the smallest piece and the noisiest to detect. Reading the whole number as if it were the sliver is the recency mistake. The cost is real. The capacity trap is real. The behaviour tax is real. What looks like signal is mostly the regime. Look at ten years. Look at the mandate. Look at the team. Look at the cost. Then ignore everything that fits in a 1-year column."
Carhart, M. M. (1997). On Persistence in Mutual Fund Performance. Journal of Finance, 52(1), 57–82. The seminal paper showing that persistent fund returns are mostly explained by momentum-factor exposure, not stock-picking skill.
Fama, E. F., & French, K. R. (2010). Luck versus Skill in the Cross-Section of Mutual Fund Returns. Journal of Finance, 65(5), 1915–1947. Bootstrap evidence that the cross-sectional distribution of true alpha is centred at zero, with the right tail nearly indistinguishable from luck.
Berk, J. B., & Green, R. C. (2004). Mutual Fund Flows and Performance in Rational Markets. Journal of Political Economy, 112(6), 1269–1295. The theoretical model: skill exists, but inflows compete it away. The reason persistence collapses.
Berk, J. B., & van Binsbergen, J. H. (2015). Measuring Skill in the Mutual Fund Industry. Journal of Financial Economics, 118(1), 1–20. Refines the Berk-Green framework using gross value-added; finds skill exists but is captured by the fund family, not the investor.
S&P Dow Jones Indices — SPIVA Scorecards (US & India). Annual reports tracking active-fund underperformance and persistence across 1, 3, 5, 10-year horizons. Indian data published since 2014.
NLE backtest engine. Daily Nifty 50 closes, 06-Nov-1995 to 07-May-2026, used for the rolling-return distributions in this paper. Data refreshed daily via the refresh-nifty-daily cron.