How Long Should You Hold a Stock? What 20 Years of Nifty 500 Data Shows
We ran the same question across every holding period we track, from one year to twenty. The share of stocks that made money climbs from about half to nearly all — and the range of outcomes collapses. Here is the measured picture, including what it does not prove.
The short version
- Across 2,94,149 measured holding periods: 64% of one-year windows made money, rising to 94% at 15 years.
- Just 1 of 458 companies never had a losing one-year window. At 15 years, 239 of 280 — 85.4% — never had one.
- The average company's worst window improves from -56.0% a year to +10.0% — while its median return barely moves.
- Part of that improvement is survivorship, not skill. We quantify it below rather than glossing over it.
“Invest for the long term” is advice almost always given without a number. We have the numbers: the same universe of Indian stocks, measured at every holding period from one year to 15. This is what actually changes as you hold longer — and, just as importantly, what does not.
What we measured
Every possible start date, not one. For each company and each window length we stepped through the record a month at a time — a five-year window beginning in January is a different investment from one beginning in February — and measured what each returned. Across the universe that is 2,94,149 separate holding periods.
That is the important difference from how this question is usually answered. Asking what the last five years did samples one start date and calls it the answer. Asking what every five-year stretch did removes the start date from the question entirely.
A window length is only published for a company once at least twelve complete windows sit behind it, so nothing here is a range drawn from a handful of observations. Returns are measured on the adjusted close, so dividends are treated as reinvested throughout. These figures are computed live from the same data behind every stock page here.
| Holding period | Companies | Windows | Windows that made money | Companies that never lost |
|---|---|---|---|---|
| 1 year | 458 | 76,610 | 64% | 1 of 458 (0.2%) |
| 3 years | 420 | 65,675 | 78% | 45 of 420 (10.7%) |
| 5 years | 377 | 55,560 | 83% | 75 of 377 (19.9%) |
| 7 years | 357 | 46,476 | 88% | 163 of 357 (45.7%) |
| 10 years | 309 | 34,012 | 90% | 198 of 309 (64.1%) |
| 15 years | 280 | 15,816 | 94% | 239 of 280 (85.4%) |
The share of winners climbs with time
Of every one-year window in the record, 64% ended above where it started. At 5 years that rises to 83%, and at 15 years to 94%.
The sharper way to see it is per company. Only 1 of 458 companies got through every single one-year window without a loss. At 15 years, 239 of 280 did — 85.4% of them, against 0.2% at one year.
Nothing about the businesses changed between those measurements. The only variable is time. Short windows are dominated by sentiment, liquidity and news; long windows give earnings growth enough room to overwhelm them.
The range of outcomes collapses
The more striking pattern is not the average — it is the spread. The average company's one-year windows ran from -56.0% at worst to +284.6% at best: a spread of 341 percentage points. At 15 years the same measure narrows to 15 points, from +10.0% to +25.4% — a 95% collapse in dispersion.
Holding longer did not raise the typical return. It narrowed the range of returns you were likely to get — which is a different, and more useful, kind of improvement.
This cuts both ways, and the optimistic reading usually omits half of it. Long holding periods compress the downside, but they compress the upside just as hard. A +284.6% stretch has no counterpart at 15 years, where the average company's best window managed +25.4% a year. Spectacular short-run results are real, but they are not something you can plan around.
Why the median barely moves
Across every window length we measured, the average company's median return sits in a narrow band — 15.6%, 20.4%, 19.1%, 17.4%, 17.6%, 17.3% at 1y, 3y, 5y, 7y, 10y, 15y respectively. The centre of the distribution is remarkably stable.
That is the point worth internalising. Time is not a return multiplier. A longer hold does not make the typical Indian stock compound faster; it makes your actual result more likely to resemble the typical one.
What this does not prove
- This is not a prediction. It describes a period that included specific cycles and a specific rate environment. The next 15 years are not obliged to rhyme.
- Overlapping windows are not independent trials. A five-year window shares fifty-nine of its sixty months with the next one, so these are a description of what happened rather than a sample from which odds can be estimated. And a distribution tells you the shape of the field, not which company you hold.
- Returns here include dividends but not costs. Every figure is measured on the adjusted close, so dividends are treated as reinvested. Brokerage and taxes are not modelled and would reduce them. Demergers are not credited, because our data source does not adjust for them.
- “Made money” means the window finished above where it started. It does not mean the ride was comfortable — a window that ended up+17.3% a year could have been -50.0% at its low point. The drawdown pages show what those falls looked like.
How to use this
The practical implication is about matching your holding period to the outcome you need. If money is needed within a year or two, this data says the range of outcomes is very wide and roughly symmetric. If it can genuinely stay invested for a decade or more, the historical range narrows sharply — but toward the middle of the distribution, not the top of it.
You can check any of this yourself, company by company. Every stock has its own rolling-return page showing the same windows for that one business, and its drawdown page shows the falls along the way. For the long end of the distribution see the 10-year compounders. The exact calculations are on our Methodology page.
HELP CENTER
Frequently asked questions
Everything you need to know about this page and the data presented.
How long should I hold a stock?+
Measured across every rolling holding period in our Nifty 500 data — roughly 294,000 of them — the share of windows that made money rises from about 64% at one year to 94% at fifteen, and the share of companies that never had a single losing window rises from under 1% to about 85%. That argues for matching your holding period to when you actually need the money rather than to a fixed rule. It is a description of the past, not advice, and part of the long-horizon improvement is survivorship bias.
Does holding longer increase returns?+
Not for the typical stock. The median annualised return stays within a narrow band across every window length we measured. What changes is dispersion: longer holds narrow the range of outcomes, cutting the worst case and the best case alike. The average company’s worst fifteen-year window still compounded at about 10% a year, while its worst one-year window lost more than half. Time made results more predictable, not larger.
Is this data affected by survivorship bias?+
Yes, and materially. Long window lengths only include companies with enough listed history to produce twelve complete windows, so businesses that delisted or failed are missing from them entirely. The number of companies measured falls as the window lengthens, which is visible in the table in this guide. The true historical success rate for an investor buying at the start of a long window was lower than the figure shown.
Do these returns include dividends?+
Yes. Every figure here is measured on the adjusted closing price, which folds in dividends and stock splits, so dividends are treated as reinvested. Brokerage and taxes are not modelled and would reduce these numbers, and demergers are not credited because our data source does not adjust for them. Roughly a quarter of the typical twenty-year total return came from dividends rather than price — we measure that separately in our guide on dividend contribution.