Executive Summary
This study measures how often lump-sum investments in Small Cap Direct Growth schemes ended with a negative annualised return, across rolling windows of 1, 3, 5, 7 and 10 years. Every statistic is a lump-sum, point-to-point return computed from daily NAV data.
Findings in the observed windows:
- Loss probability declined at each longer horizon tested: 19.46% (1Y), 5.73% (3Y), 2.79% (5Y), 0.25% (7Y) and 0.0% (10Y).
- The zero-loss horizon in this dataset is 10 years. No shorter horizon tested was loss-free.
- Median annualised CAGR was 19.28% at 7 years and 19.13% at 10 years.
- The worst 7Y window returned -2.72%. The worst 10Y window returned +3.37%.
- The 10Y result rests on 14,686 daily rolling windows across 12 Direct Growth schemes, with window start dates only from 2013-01-01 to 2016-10-01, and on currently active schemes only. It should be read as a narrow sample, not a general law.
Past performance is not indicative of future results. Other holding periods were not tested.
Data and Methodology
| Parameter | Specification |
|---|---|
| Source | nav_history daily NAV |
| Plan selection | Direct plan, Growth option, one series per scheme |
| Return metric | Annualised CAGR per window |
| Window step | Daily (calendar), NAV forward-filled over non-trading days |
| Data window | Window start dates from 2013-01-01 to each horizon's data_end |
| Data floor | 2013-01-01 (Direct plans launched Jan 2013) |
| Universe | Currently active schemes only (survivorship bias: closed/merged schemes excluded) |
| Dropped windows (gaps) | 0 |
| Excluded schemes | 3, for having no Direct Growth plan |
Each horizon's window count is the number of daily N-year rolling windows pooled across the schemes with enough history for that horizon. Scheme counts therefore fall as horizons lengthen. The methodology records 36 schemes in total. The category label reflects current SEBI classification, and earlier windows use today's membership. The label dates from 2017-10-06 (category_since). No claim is made here about the category before that date or about periods before the 2013-01-01 data floor.
The survivorship-bias note matters for interpretation. Closed or merged schemes are excluded, so the observed loss probabilities describe schemes that remain active. The data is internally validated.
Loss Probability by Holding Horizon
| Horizon | Daily rolling windows | Direct Growth schemes | Loss probability | Median CAGR | Share of windows above 6.0% inflation threshold |
|---|---|---|---|---|---|
| 1Y | 79,852 | 31 | 19.46% | 14.71% | 68.1% |
| 3Y | 59,156 | 24 | 5.73% | 22.01% | 88.91% |
| 5Y | 43,163 | 21 | 2.79% | 20.2% | 91.27% |
| 7Y | 29,217 | 17 | 0.25% | 19.28% | 95.27% |
| 10Y | 14,686 | 12 | 0.0% | 19.13% | 98.25% |
The inflation threshold of 6.0% is the RBI CPI upper tolerance band recorded in the snapshot's macro benchmarks. The final column reports the share of windows whose annualised return exceeded that threshold.
The loss-probability curve declines at every step in the table. The largest drop sits between the 1Y and 3Y readings. From 3Y onward, loss probability remains positive through 7 years (0.25%) and reaches 0.0% only at 10 years. The data cannot show why the curve falls. It records only the outcomes.
Return Distribution by Horizon
| Horizon | P5 | P25 | Median | P75 | P95 | Std dev | Worst window | Best window |
|---|---|---|---|---|---|---|---|---|
| 1Y | -13.44% | 2.76% | 14.71% | 37.99% | 92.37% | 31.27 | -44.61% | 222.83% |
| 3Y | -1.19% | 14.15% | 22.01% | 30.22% | 41.96% | 12.84 | -23.62% | 71.59% |
| 5Y | 2.7% | 13.77% | 20.2% | 27.76% | 36.36% | 10.24 | -13.99% | 54.64% |
| 7Y | 6.17% | 15.64% | 19.28% | 21.98% | 26.34% | 5.61 | -2.72% | 36.25% |
| 10Y | 8.12% | 16.88% | 19.13% | 21.54% | 26.38% | 4.77 | 3.37% | 31.8% |
Three features stand out in the observed windows:
- The 5th percentile rises from -13.44% at 1Y to 8.12% at 10Y, so the lower tail moves above zero at 5 years and stays above zero at longer horizons.
- Standard deviation falls from 31.27 at 1Y to 4.77 at 10Y.
- Median CAGR is similar at 7Y (19.28%) and 10Y (19.13%). Longer horizons narrowed the spread of outcomes rather than lifting the typical result.
The worst 7Y window returned -2.72%, and the 7Y loss probability was 0.25%. The worst 10Y window returned +3.37%, consistent with the 0.0% loss probability.
Loss Probability by Window Start Year
The pooled figures conceal variation by when a window began. At the 1Y horizon, loss probability by start year ranged from 0.0% (2020 and 2023) and 0.05% (2014) to 79.51% (2018). Other start years included 32.6% (2017), 42.38% (2019) and 39.38% (2024).
At 3Y, the 2017 start year showed a 51.23% loss probability, while start years 2019 to 2023 each showed 0.0%. At 5Y, the 2015 start year showed 17.35%. At 7Y, only the 2013 start year showed a loss probability above zero (2.42%). At 10Y, each start year (2013, 2014, 2015 and 2016) showed 0.0%.
Start-year results are highly clustered because daily windows overlap heavily. They should be read as regime-dependent samples, not independent observations.
The Zero-Loss Horizon
The hypothesis tested is whether any holding period in the dataset was loss-free. The result: zero_loss_horizon_years is 10. At 10 years, 14,686 daily rolling windows across 12 Direct Growth schemes produced no negative annualised return, and the lowest observed result was +3.37%.
Several constraints limit what that statement supports:
- Narrow start-date range. The 10Y windows start only between 2013-01-01 and 2016-10-01, because longer histories are not available for Direct plans before the data floor.
- Small scheme sample. 12 Direct Growth schemes contribute to the 10Y result, against 31 at 1Y.
- Overlap. Daily windows share most of their underlying NAV path, so 14,686 windows do not represent 14,686 independent outcomes.
- Survivorship. Only currently active schemes are included.
- Lump-sum only. Results do not describe staggered or periodic investing patterns.
A 0.0% loss probability is a statement about this sample. It is not a statement about future periods.
Limitations
- Other holding periods were not tested.
- Earlier windows use today's category membership, so the series is not a record of historical category composition.
- Loss probability measures frequency, not magnitude. The 1Y worst window of -44.61% shows the size of single-window outcomes at the shortest horizon.
- Returns are pre-tax and the study does not model taxation.
- Results are descriptive. They identify no causal mechanism.
Data Integrity Statement
All figures derive from the internally validated FundSageAI Quant Snapshot, built from daily NAV history for Direct plan, Growth option series. Window counts represent daily rolling windows pooled across the stated number of schemes, not a count of schemes. The universe is limited to currently active schemes, so closed or merged schemes are excluded and survivorship bias applies. No statistics were derived beyond those in the snapshot, and no claims are made for periods before 2013-01-01.
Disclaimer
This is educational content, not investment advice; consult a SEBI-registered investment adviser. Past performance is not indicative of future results.