Rolling Returns
Every holding period, not the flattering one
Our own top-performer pages carry a warning: the same stock can look outstanding over one horizon and ordinary over another. This is the answer to that warning. Pick a company and you will see every one-, three-, five-, seven- and ten-year stretch it has ever been through — how often each ended in profit, what the worst one did, and which month you would have had to start in to get it.
Start with a widely held name
Or browse every stock we track and open the rolling-returns view from any company page.
HELP CENTER
Rolling returns — common questions
Everything you need to know about this page and the data presented.
What is a rolling return?+
A normal return figure picks one start date and one end date. A rolling return uses every possible start date instead. If a stock has twenty years of history, there are roughly 180 different five-year stretches inside it, each starting a month after the last — and they do not all say the same thing. Publishing the whole set is the difference between describing a stock and advertising its best window.
Why does this matter more than CAGR?+
Because a headline CAGR is a single sample, and which sample you got depended on when you happened to buy. A stock quoted at "18% a year over ten years" may have delivered 31% to someone who started in one month and 4% to someone who started eight months later. The spread between those two outcomes is the part of the result that had nothing to do with the company, and it is invisible in the single number.
Does this tell me the probability of a future return?+
No, and it should not be read that way. Rolling windows overlap heavily — a five-year window shares 59 of its 60 months with the next one — so they are not independent observations and cannot be treated as a sample from which to estimate odds. They are a description of what one company actually lived through, over the specific years it happened to be listed.
Are dividends included?+
Yes. Every window is measured on month-end closing prices adjusted for dividends and stock splits, so dividends are treated as reinvested. What is not included: demergers and spin-offs, which our data source does not adjust for, and brokerage and taxes, which are not modelled.
Why do some stocks show fewer window lengths than others?+
A window length only appears once there are at least twelve complete windows behind it. A ten-year window needs eleven years of history before it can show twelve of them, so a company listed in 2020 shows one- and three-year windows and nothing longer. Fewer than twelve observations describes an anecdote, not a distribution, and publishing it as a range would overstate what the data supports.
Is a stock that never had a losing decade a safe investment?+
It is a stock that has not yet had one. That is a meaningfully different statement, and the distinction matters more than any figure on these pages. ReturnScreener is an educational research platform and is not a SEBI-registered investment adviser; nothing here is a recommendation, and a record of the past carries no promise about the next window.
Go Deeper
Other ways to read the same history
A rolling return is one question the price record can answer. Here are the others.
ReturnScreener is an educational research platform and is not a SEBI-registered investment adviser. Rolling returns are records of what historical prices did, not projections, and overlapping windows are not a sample from which probabilities can be estimated. Past performance does not guarantee future returns.