On this page
The overview projection applies your assumed rates every year, so it shows one smooth path. The Success odds tab runs the same plan through many market histories and counts how many trials run out. A plan can be funded in the projection and still fail in a large share of trials. A projection that runs out can also succeed in many trials.
Neither result is wrong. They answer different questions: what happens if markets match your assumptions every year, and how often the plan survives when they don’t.
The most common cause. In the projection every year earns the same return. In a trial, a run of losses while you’re withdrawing forces sales at low prices, and the money sold never recovers. A trial with the same long-run average return as your assumption can still run out if its bad years come first.
$40,000 of spending
The projection is funded through the plan’s end with $771,797 left. 69.3% of trials succeed.
$45,000 of spending
The projection runs out in its final year. 58.3% of trials still succeed.
Both use the same example plan: $1,000,000 in a brokerage account at 60, 40% bonds, plan default rates, 1,000 trials. A thin margin at the end of the projection usually means a success rate near half, in either direction.
- 90th percentile
- Median
- 10th percentile
- Steady-return projection
The projection stays inside the middle of the spread, but the 10th percentile runs out at 84.
Show data
| Year | 10th percentile | Median | 90th percentile | Steady-return projection |
|---|---|---|---|---|
| 2026 | $903,695 | $1,050,855 | $1,191,426 | $1,016,947 |
| 2027 | $844,993 | $1,076,355 | $1,279,472 | $1,033,646 |
| 2028 | $849,125 | $1,112,791 | $1,382,473 | $1,050,045 |
| 2029 | $826,110 | $1,108,057 | $1,418,480 | $1,066,089 |
| 2030 | $815,884 | $1,159,919 | $1,533,379 | $1,081,717 |
| 2031 | $765,045 | $1,154,955 | $1,650,942 | $1,096,865 |
| 2032 | $755,285 | $1,176,233 | $1,744,086 | $1,111,465 |
| 2033 | $736,868 | $1,209,473 | $1,861,436 | $1,125,440 |
| 2034 | $726,794 | $1,262,392 | $1,960,457 | $1,138,712 |
| 2035 | $700,499 | $1,236,553 | $2,052,218 | $1,151,193 |
| 2036 | $665,616 | $1,253,231 | $2,177,830 | $1,162,791 |
| 2037 | $649,097 | $1,287,465 | $2,339,571 | $1,173,406 |
| 2038 | $608,497 | $1,305,235 | $2,500,296 | $1,182,933 |
| 2039 | $582,184 | $1,337,811 | $2,608,519 | $1,191,256 |
| 2040 | $546,603 | $1,356,916 | $2,718,399 | $1,198,253 |
| 2041 | $509,830 | $1,368,980 | $2,827,490 | $1,203,793 |
| 2042 | $470,624 | $1,412,642 | $2,927,024 | $1,207,736 |
| 2043 | $421,581 | $1,421,444 | $3,097,985 | $1,209,931 |
| 2044 | $372,605 | $1,435,419 | $3,245,577 | $1,210,218 |
| 2045 | $324,680 | $1,427,118 | $3,385,770 | $1,208,425 |
| 2046 | $264,203 | $1,462,917 | $3,572,345 | $1,204,368 |
| 2047 | $207,954 | $1,462,463 | $3,698,620 | $1,197,852 |
| 2048 | $146,521 | $1,465,654 | $3,963,867 | $1,188,667 |
| 2049 | $53,558 | $1,464,605 | $4,109,049 | $1,176,588 |
| 2050 | −$56,554 | $1,496,442 | $4,384,975 | $1,161,379 |
| 2051 | −$131,409 | $1,444,973 | $4,627,238 | $1,142,784 |
| 2052 | −$215,355 | $1,452,018 | $4,801,611 | $1,120,532 |
| 2053 | −$314,718 | $1,472,230 | $5,053,376 | $1,094,334 |
| 2054 | −$419,476 | $1,469,517 | $5,246,729 | $1,063,880 |
| 2055 | −$507,483 | $1,464,069 | $5,670,521 | $1,028,842 |
| 2056 | −$618,184 | $1,404,668 | $5,733,979 | $988,870 |
| 2057 | −$747,828 | $1,417,931 | $6,136,271 | $943,590 |
| 2058 | −$884,473 | $1,403,716 | $6,435,965 | $892,604 |
| 2059 | −$1,032,042 | $1,438,848 | $6,915,241 | $835,490 |
| 2060 | −$1,169,270 | $1,384,414 | $7,286,081 | $771,797 |
The median line is the middle value of all trials in each year, not a trial of its own. It can end well above or below the projection. In the example it ends at $1,368,356 against the projection’s $771,797, yet 30.7% of trials run out. Judge risk by the success rate and the lower percentiles, not by where the median ends. See Success odds.
- Calibration off. Trials then use unadjusted history, which returned more than the 7% default. The example rises from 69.3% to 90.7%.
- Failure definition. Both tests fail a trial in any year spending goes unpaid. The default also fails it when cash and investments reach zero. The net-worth test instead fails it when debts exceed everything owned, which can happen while spending is still paid.
- Targets. A final net worth target or counting flexible-spending cuts affects only the target success rate, not the funding success rate.
- Trials and seed. The same seed repeats the same result. More trials move it by about a point: the example is 69.3% at 1,000 trials and 68.5% at 5,000.
Settings are in Success oddsRun settings and success criteria.
To see how one bad historical period plays out year by year, replay it in the Historical stress test.