Historical backtest

What If You Retired in 1937? The 4% Rule, Backtested

A $1,000,000 60/40 portfolio, retiring in 1937 and spending $40,000/yr (inflation-adjusted), made it the full 30 years against real market history.

By · Updated July 15, 2026
3.5% withdrawal
$35,000/yr
PASSED
Funded all 30 years. Ended with $1,401,000 (today's real dollars).
4% withdrawal
$40,000/yr
PASSED
Funded all 30 years. Ended with $936,000 (today's real dollars).
5% withdrawal
$50,000/yr
PASSED
Funded all 30 years. Ended with $5,000 (today's real dollars).

Year by year: the 4% plan

YearAgeStocksBonds60/40WithdrawalEnd balance
1 65 −35% +2% −20% $40,000 $766,000
2 66 +31% +6% +21% $40,000 $879,000
3 67 0% +6% +2% $40,000 $859,000
4 68 −10% +5% −4% $40,000 $786,000
5 69 −16% −8% −13% $40,000 $650,000
6 70 +13% −6% +5% $40,000 $643,000
7 71 +25% 0% +15% $40,000 $694,000
8 72 +19% 0% +11% $40,000 $728,000
9 73 +36% +3% +23% $40,000 $845,000
10 74 −16% −18% −17% $40,000 $670,000
11 75 −3% −12% −7% $40,000 $589,000
12 76 +3% +2% +3% $40,000 $563,000
13 77 +19% +7% +14% $40,000 $597,000
14 78 +25% −2% +14% $40,000 $636,000
15 79 +18% −6% +8% $40,000 $646,000
16 80 +17% +2% +11% $40,000 $673,000
17 81 −2% +3% 0% $40,000 $633,000
18 82 +52% +7% +34% $40,000 $795,000
19 83 +30% −1% +18% $40,000 $887,000
20 84 +5% −6% +1% $40,000 $852,000
21 85 −13% +5% −6% $40,000 $765,000
22 86 +43% −4% +24% $40,000 $901,000
23 87 +10% −3% +5% $40,000 $902,000
24 88 −1% +10% +3% $40,000 $891,000
25 89 +26% +1% +16% $40,000 $988,000
26 90 −10% +5% −4% $40,000 $910,000
27 91 +21% +1% +13% $40,000 $983,000
28 92 +15% +3% +10% $40,000 $1,039,000
29 93 +10% −1% +6% $40,000 $1,055,000
30 94 −13% 0% −8% $40,000 $936,000

What this sequence teaches

Over the first five years of this retirement (1937–1941), a 60/40 portfolio's cumulative real return was −17%. The single worst year in the tested window was 1937, when the 60/40 blend returned −20% in real terms.

Under the 4% withdrawal plan, the real portfolio balance bottomed out at $563,000 in 1948, before recovering in later years.

Because the first five years were net negative, this retirement faced early sequence-of-returns risk: withdrawals were drawn from a shrinking pool before growth had a chance to rebuild it — the single biggest driver of historical 4%-rule failures.

What RetireOdds actually simulates

The table above is the transparent skeleton: one portfolio, one withdrawal rule, one sequence of real historical returns, before taxes. It's meant to be checkable by hand.

Inside RetireOdds, the same year-by-year loop runs against your plan and adds what a real retirement actually has to deal with: federal and state taxes with account buckets (taxable, tax-deferred, Roth) drawn in order, Social Security claiming and its partial taxability, Required Minimum Distributions, healthcare costs (ACA subsidies before 65, Medicare and IRMAA after), Roth conversions, and one-time life events.

It also runs three engines instead of one: Monte Carlo (1,000 lognormal real-return paths calibrated to this same 1928–2023 dataset), a block bootstrap of this history, and the historical replay shown on this page. A plan fails if any year is unfunded — including the last one.

Read the full method on /methodology, walk through the product in the user guide, or try your own numbers in the free calculator.

Returns are approximate, rounded, planning-grade real (inflation-adjusted) totals for US large-cap stocks and 10-year Treasuries — this is educational modeling, not financial advice.

Run this against your own plan

This page tests one fixed portfolio against history. RetireOdds tests your numbers — your accounts, your Social Security, your taxes — across three simulation engines.