Explainer
Rolling Returns Explained
A single backtest window can be misleading. Rolling returns look at many overlapping windows to show a fuller range of outcomes and how timing changes the story.
Published: December 26, 2025 · Updated: December 26, 2025 · By FinToolSuite Editorial
Disclaimer
- Educational purposes only, not financial advice.
- Examples are illustrative and simplified.
- Past performance is not a reliable indicator of future results.
- Market returns can be negative.
- See the Privacy Policy for data handling details.
Open the Investment History Checker
Run overlapping windows to see how returns change with timing.
Quick answer
- Rolling returns measure the same window length across many start dates.
- They reduce cherry picking by showing a range, not one result.
- You can approximate them by running overlapping windows in the tool.
What rolling returns are
Pick a window length (e.g., 3 or 5 years), slide that window forward over time, and calculate the return each time. Instead of one answer, you see many overlapping results.
Why rolling returns reduce cherry picking
- No single start date dominates the story.
- Shows good, average, and bad windows.
- Helps you see how sensitive results are to timing.
Simple illustrative example
| Window length | Start (illustrative) | End (illustrative) | Return percent |
|---|---|---|---|
| 3-year | 2015 | 2018 | 12% |
| 3-year | 2016 | 2019 | 25% |
| 3-year | 2017 | 2020 | -5% |
| 3-year | 2018 | 2021 | 18% |
Illustrative only; overlapping windows can show very different outcomes.
How to approximate rolling returns in the tool
- Choose a window length (e.g., 5 years).
- Run the first window and save as a scenario.
- Move the start and end dates forward by 6 or 12 months and rerun.
- Repeat until you have several overlapping scenarios.
- Compare the spread of CAGR, total return, and drawdown.
See choosing start and end dates for date tips.
Using rolling windows to compare two tickers
Run the same set of overlapping windows for both tickers. Keep currency and date ranges consistent, and compare the distribution of returns instead of one headline figure.
Learn more in how to compare two tickers.
Limitations and safe interpretation
- Still historical, not a forecast.
- May exclude dividends, fees, taxes, and inflation unless stated.
- Data gaps and corporate actions can affect long histories.
- The step size (6 months vs 1 year) changes how many windows you test.
Quick checklist
- [ ] Chosen a window length
- [ ] Tested at least 5 to 10 overlapping windows
- [ ] Looked at return and drawdown together
- [ ] Avoided treating the best rolling window as normal
FAQ preview
What window length should I use?
Pick a length relevant to your question and test multiple windows.
How many rolling windows are enough?
Use several overlapping windows (e.g., 5–10 or more) to see the range.
Is rolling return the same as CAGR?
Each window can be shown as CAGR, but rolling returns are many windows, not one.
Can rolling returns predict the future?
No. They summarize history; they are not forecasts.
Can I do this for two tickers?
Yes, if you keep currency and dates consistent across both sets of windows.
Run overlapping windows
Choose a window length, shift dates, and compare the spread of results to see how timing changes the story.