A few hands-on tools to make compounding and long-term risk feel real, not theoretical.
What would recurring contributions to the S&P 500 (with dividends reinvested) be worth, using actual historical annual returns?
—
Returns are S&P 500 total returns (price + dividends reinvested), 1928–2025. The model places each of your N contributions at the middle of its period (e.g., monthly → 12 contributions, one mid-month each), grows each to year-end by the appropriate fraction of that year's S&P return, and compounds the running balance year over year. Frequency now matters: weekly modestly outperforms monthly which outperforms annual, because dollars in the market earlier get more time to grow (Jensen's inequality on convex compounding). You choose the end year (default = the latest year with data) and the number of years; the start year is computed as End year − Years + 1. That's why you can't pick a future year — there's no data past 2025 yet. Inflation-adjusted mode deflates each year's return by that year's actual CPI (Dec/Dec, same data file), so the whole simulation runs in constant purchasing power, anchored to today (the latest data year): the contribution amount means that many of today's dollars every year (back in 1996 that was a smaller nominal amount), ending balances read directly in today's dollars, and Return/yr and IRR become real rates — the nominal hockey stick flattens to what you could actually buy.
Runs everything in today's purchasing power: your contribution means today's dollars every year, each year's return is deflated by that year's actual CPI, and cash under the mattress visibly loses to inflation.
| Option | Ending balance | Total growth | Return/yr | Your IRR |
|---|
One bar per starting year. Bar height is the annualized return you'd have earned over the next N years. Drag N and watch the red bars disappear.
Each bar shows the geometric annualized return (CAGR) over the N-year window starting at that year — exactly what a buy-and-hold investor would have earned if they bought the S&P 500 at the start of that year and sold N years later, dividends reinvested. The x-axis stays anchored to the full dataset (1928 onward); starting years for which a full N-year window doesn't fit simply have no bar. The headline callout computes the smallest N at which no starting year in the dataset ended negative — the empirical answer to "how long until time wins?" With the inflation-adjusted toggle on (shared with Tab 1), each year's return is deflated by that year's actual CPI, so the bars are real CAGRs — and the fully-green threshold moves out: in purchasing-power terms, red bars survive much longer than the nominal chart suggests.