I computed a safe withdrawal rate for every one of the 165 strategy backtests we track. One engine, one rule set: Bengen's rolling cohorts, a 30-year retirement started in every possible month, withdrawals raised with actual inflation, and the safe rate is the highest starting rate where every single cohort survives. The median came out at 6.83%, and 95% of strategies beat the famous 4% rule. That sounds like the 4% rule is dead and retirees have been leaving money on the table for 30 years. It isn't dead. The reason it survives is the most important number in retirement planning that nobody puts on the label.
2 rows from the same table
The classic 60/40 portfolio, backtested from 1922 across 104 years, prints a safe rate of 2.70%. The 100% US stock row, whose data starts in 1987, prints 7.07%. Same engine. Same math. The 60/40 number had to survive a retirement that started into 1929 and another into 1966. The stock market number has never met anything worse than the dot-com bust. A withdrawal rate is mostly a statement about which disasters your data contains.
The pattern holds across the whole catalog. Strategies with century-scale backtests print floors between 2.7% and 5%. Every double-digit safe rate belongs to a backtest that starts after 1970, usually after 1990. Bengen's original 4% was itself the worst-cohort floor over about 70 years of American history. Most numbers above it are not proof the rule was wrong. They are proof the backtest lived through friendlier weather.
What survives the correction
Here is what survives, and it is not nothing. The cleanest comparison we track is a static Permanent Portfolio and its tactical variant on identical 66-year windows. Static: 3.99%. Tactical: 5.01%. Same assets, same cohorts, so the 1-point gap is the drawdown management itself, which attacks sequence-of-returns risk exactly where it hurts a retiree. Century Momentum tells the same story with 1929 inside its sample: 4.99% against the 60/40's 2.70%, with a perpetual rate of 4.74%. Matched windows say tactical management is worth roughly 1 to 2 points of safe withdrawal rate. That is a huge edge. It is just not 7%.
The number we trust least
The single highest safe rate in the catalog is 15.11%, and it belongs to a strategy our own robustness score flags as statistically fragile, meaning its results are within reach of pure luck given how many variants we tested. Spectacular withdrawal floors and weak statistics come from the same place: short, friendly histories. When a withdrawal number looks like a typo, check the years column before you believe it.
How to read a withdrawal rate
3 rules cover most of it. Never compare safe withdrawal rates computed on different history lengths, on our site or anywhere else. Prefer the perpetual rate, the stricter version that preserves capital (the 60/40's 1.41% against its 2.70% shows how much stricter), when your plan runs longer than 30 years. And treat any double-digit rate as a description of a lucky era until the strategy has survived at least one real bear market with its drawdown control working.
The full table, with the safe rate, the perpetual rate, and the years of history side by side for every strategy, is on the leaderboard. The distribution runs from a 10th percentile of 4.66% to a 90th percentile of 10.0%, and now you know exactly what that spread is measuring: mostly, the years column.