25× Expenses Lasted 30 Years in 73% of Our Simulations, 50 Years in 49%
The 25× rule, and the 4% withdrawal rate behind it, comes from the 1998 Trinity Study, which tested historical US stock and bond returns against retirements of up to 30 years. Most early retirees plan for 45 or 50.
We used the same model as our FIRE calculator: yearly returns averaging 7% with 15% volatility, inflation averaging 2.5%, spending that rises with inflation every year. Then we retired someone on day one with $1.25M (25× of $50K) and ran 5,000 paths for each horizon.
What 25× buys at different horizons
| Retirement length | 25× (4%) | 28.6× (3.5%) | 30× (3.33%) | 33.3× (3%) |
|---|---|---|---|---|
| 30 years | 73% | 82% | 84% | 89% |
| 40 years | 59% | 69% | 72% | 78% |
| 45 years | 53% | 64% | 68% | 74% |
| 50 years | 49% | 60% | 64% | 70% |
Two things stand out. First, our 30-year number for 25× (73%) is lower than Trinity’s historical result for a 50/50 stock and bond portfolio (95%). Our model treats the whole portfolio as one asset with 15% volatility and a normal distribution; keep the 7% average but cut volatility to 10% and the same 30-year case survives 90% of the time. The volatility assumption matters as much as the multiplier.
Second, the drop from 30 to 50 years is steep at every multiplier: about 20 to 25 points. Early retirement isn’t just a longer version of normal retirement; it doubles the window in which a bad decade can hit.
Sequence risk is where the failures come from
We took the 50-year, 25× paths and split them by the average real return of their first five years:
| Opening five years | Survival over 50 years |
|---|---|
| Worst quarter | 16% |
| Best quarter | 80% |
Seven in ten of the paths that ran out had below-median returns in those first five years. When the portfolio falls early, every withdrawal sells more shares at low prices, and the recovery later works on a smaller base. Bengen’s 1994 paper (PDF) found the same pattern in historical data: the order of returns predicts failure better than the average.
Fixes, and how much each one moves the 50-year number
| Change (50-year retirement, $50K spending) | Survival |
|---|---|
| Baseline: $1.25M (25×) | 49% |
| $15K/year part-time income, first 5 years | 53% |
| $15K/year part-time income, first 10 years | 57% |
| $1.43M (28.6×, a 3.5% withdrawal) | 60% |
| $1.5M (30×) | 64% |
| $1.67M (33.3×, a 3% withdrawal) | 70% |
“Save to 30×” and “withdraw 3.33% from the same money” are the same ratio, so they give the same survival; the difference is only whether you get there by saving more or spending less. Part-time income is real insurance against a bad start, but $15K a year for five years is $75K in total, against the $250K extra that 30× represents. It shouldn’t be expected to do the same job.
The calculator itself doesn’t have a part-time income input. For those two rows we used a copy of its simulation loop with that one change.
Where the simple math breaks further
- Healthcare before Medicare. If $1.25M has to support $68K instead of $50K (an extra $18K of premiums), 50-year survival falls to 25%. Put insurance in the spending number before you multiply.
- Taxes. The simulation withdraws gross amounts. Withdrawals from traditional accounts are taxed, so the spending you can actually use is lower than the number in the table.
- Spending isn’t flat. Real retirees tend to spend more in the early years and on healthcare late (Center for Retirement Research). The model keeps spending flat in real terms.
Where this framework doesn’t hold
- Pensions and Social Security. Guaranteed income covering part of spending later in life shrinks what the portfolio has to do, and these survival numbers become conservative.
- Flexible spending. Guardrail strategies (Guyton-Klinger and similar) cut withdrawals after bad years. That changes the math entirely, because the portfolio is never forced to sell a fixed amount at the bottom.
- A different return model. A portfolio with bonds has lower volatility and a lower average. Run your own allocation’s assumptions through the calculator’s settings instead of borrowing ours.
What to actually do
- Multiply your real spending (including healthcare) by 30 to 33 if you plan for 45 to 50 years, not by 25.
- In the FIRE calculator, set the withdrawal rate to 3% and 3.5% and compare the survival rates with your own numbers.
- Decide on a rule for bad years before you retire: how much you’d cut spending, or how much part-time work you’d take on.
- Re-run the numbers every year for the first five years; that’s when a bad sequence is still fixable.
- For a worked example from age 25, see retiring at 40 on $80K.
Quick answers
Why doesn't the 25× rule work as well for early retirement?
The 4% withdrawal behind 25× was tested on 30-year retirements. In our simulation, $1.25M supporting $50K of inflation-adjusted spending lasted 30 years in 73% of paths, 40 years in 59% and 50 years in 49%. Longer retirements give a bad market more chances to land early.
How many times expenses do I need to retire at 45?
For a 50-year retirement in our model: 25× gives 49% survival, 30× gives 64% and 33× (a 3% withdrawal) gives 70%. No multiplier makes it certain; going from 25× to 30× adds 15 percentage points, and 30× to 33× adds 6 more.
Does part-time income in the first years fix sequence risk?
Only partly. Covering $15K of the $50K for the first 5 years lifted 50-year survival from 49% to 53%; for the first 10 years, to 57%. It helps most when the bad years come early, but it's a smaller lever than a bigger portfolio.