Free primer: 25× and the 4% poster. Calculator: FIRE number. Tax order is a different file: tax-efficient withdrawals. Paid steps assume you already know a crash in year one is not a crash in year fifteen, and you want a sleeve + a cut rule, not a second Trinity citation.
Do this in order
Write honest spend and the nest egg. The rate is the remainder, not a personality. Annual checks you will actually write, after a pension you already have, after a paid-off house if it is paid off. Nest egg = taxable + traditional + Roth, not the house. Then rate = spend ÷ nest. Run it in the FIRE calculator. $1.2M and $48k is 4%. $1.2M and $80k is 6.7% — that is not a 4% plan with a travel hobby. Under-counting tax, insurance, and the roof is how the poster lies. The 4% / 25× slogan is Bengen (1994) and Trinity (1998): 30-year, 50/50 US stocks/bonds, inflation-adjusted, historical US sample. It is not IRS, SSA, or a 50-year paycheck.
Draw the two-path picture before you pick an allocation. Same returns, opposite order, withdrawals in between. Worked (educational, no inflation): $1.2M, $48k at each year-end. Path A (crash first): −25%, −15%, +20%, +15%, +10% → year-5 pile $864,952. Path B (the reverse): +10%, +15%, +20%, −15%, −25% → $967,722. Same five percentages, $102,770 apart because Path A sold shares while they were cheap. A 25% drop on $1.2M is $900k; after the $48k check you have $852k. A 25% bounce is $1,065k, not $1.2M. Back to $1.2M from $852k needs 40.8%. “The market always recovers” skips the check you cashed.
Name years 1–5 as the risk window. Year 15 is a different file. Sequence risk is selling a fixed dollar after a hole, then owning fewer shares for the bounce. After a recovered pile and a smaller remaining horizon, a crash is mostly a paper loss — you can cut, wait, or die with a smaller heir pile. The first five years are where a 4% plan becomes a 6% plan without anyone changing the lifestyle. If you cannot fund those five years without a forced stock sale, you are not retired. You are one statement away from a smaller life.
Size a 2–3 year cash / short-Treasury sleeve before the last paycheck. 2× spend is a mild year. 3× is a two-crash window. On $48k that is $96k–$144k, parked where you will not “invest it because the rate looks low.” Type your HYSA APY and T-bill discount in the I-bonds vs T-bills vs HYSA tool — defaults on that page are examples, not today’s auction. I-bonds: $10k/year/person, 12-month lock, 3-month penalty if cashed before five years (TreasuryDirect). Do not put the whole sleeve in I-bonds; Tuesday money cannot wait out the lock. The sleeve is what you spend in a hole so the stock lot stays put. Same-year: do not empty the emergency fund because a blog said “taxable first” — that file is a water-heater, this sleeve is a bear market.
Write a cut rule the year you retire, not the year the statement is red. Guyton–Klinger shape (not a law): start near 4% of the starting nest (or 3.3% if the horizon is FIRE-long — the calculator’s conservative twin). Skip the inflation raise after a down year. If this year’s withdrawal ÷ current pile exceeds the starting rate × 1.2 (4% → 4.8%), cut the check ~10%. If it falls below starting × 0.8 (3.2%), give yourself a raise ~10%. Worked: after Path A year 1, $48k / $852k = 5.63%. That is above 4.8%. Cut 10% → $43,200. Flexibility is the actual edge versus a pensioner. A fixed real $48k through a hole is how Trinity’s failures happen. Inflation on the check makes the hole deeper — 2026 SSA COLA is 2.8% (SSA COLA fact sheet); that is a floor on a Social Security check, not a required raise on a portfolio draw.
Delay Social Security if the sleeve exists. The delayed credit is a sequence hedge. Born 1960+: FRA 67, 62 = 70% of PIA, 70 = 124% (8%/year delayed credit, stops at 70 — SSA retirement-age / delayed-retirement pages). 2026 COLA 2.8%; average retired-worker check after COLA $2,071/mo; max at FRA $4,152/mo (SSA 2026 COLA fact sheet). A delayed check is an inflation-adjusted annuity that does not sell stocks in year one. Fund the 67→70 gap from the sleeve, a pension, Roth, or part-time work after FRA (no earnings test). Earnings test 2026: $1 withheld per $2 over $24,480 if under FRA all year; $1 per $3 over $65,160 in the FRA year (SSA). Do not claim at 62 then earn $80k. Who delays in a couple is the household file — high earner’s delay is the survivor check. If Social Security is the rent and the sleeve would be a card, claiming earlier can be rational. That is cash this year, not a failed IQ test.
Do not buy 80% stocks the day you retire because a glide path said so. Years −5 to +5 are a bond tent, not a victory lap. Hold more short bonds / T-bills in that window; let the equity share rise later if the pile recovered (rising-equity glide). A target-date “to” 2026 dumps you at a date; “through” keeps gliding. Read the glide, not the year on the label. A 7% first lien is still a 7% first lien — extra principal vs investing is that playbook, not a reason to lever the tent. Do not cash-out a 3% mortgage to “have more in the market” the year you stop working. Sequence risk plus a new 7% payment is two problems.
Taxes are part of the withdrawal rate. Harvest the hole; do not IRA-draw it. A down year in taxable is a tax-loss year (harvest file) and a year to spend from the sleeve or Roth basis so you are not selling the recovery and booking a large traditional withdrawal. MAGI still feeds IRMAA two years out and ACA this year. Conversion math lives in tax-efficient withdrawals and the conversion calculator — fill 12% on purpose in gap years, not in the same December you just cut the lifestyle 10%. Do not take 4% from the IRA because “that is the retirement account” while brokerage lots are down 25% and the Roth is sitting there. The rate is dollars out the door, after tax.
RMDs are a floor on flexibility, not a suggestion. Born 1951–1959: first RMD at 73. Born 1960+: 75 (SECURE 2.0; IRS Pub. 590-B). Uniform Lifetime Table (Treas. Reg. §1.401(a)(9)-9): age 73 factor 26.5; age 75 24.6. Worked: $800,000 traditional, 7% for 11 years to 75 → about $1.68M ÷ 24.6 ≈ $68k that must come out, crash or not. Missed-RMD penalty 25%, 10% if corrected in the window. You cannot skip an RMD because the S&P is red. A QCD (2026 cap $111,000 per IRA owner — IRS Notice 2025-67) can satisfy the RMD without MAGI if you give anyway. Roth IRAs have no lifetime RMD for the original owner. Run today’s balance through the RMD calculator. The paid RMD sequence is ira-rmd-playbook — do not copy it here.
Rewrite spend after pensions, Social Security, and a paid-off house. Then rerun the rate. $48k from the portfolio plus $2,071 × 12 ≈ $24,852 of average Social Security (SSA 2026) is not an 8% withdrawal. It is a 4% draw on $1.2M plus a pension-shaped check. Put the Social Security / pension dollars in the “already have” column of the FIRE tool, not as a hope. A mortgage that still exists is a required check — either pay it from the sleeve math or keep working. Part-time W-2 after FRA has no earnings test; before FRA it does (step 6). Healthcare before 65 is often the hole in the FIRE primer — price ACA at your MAGI, then decide whether the 10% cut in step 5 still feeds the premium.
Annuities are insurance. Price a quote; do not donate the pile. A SPIA can replace part of the year-1–5 stock-sale risk with a check you cannot sequence-break. A QLAC (qualifying longevity annuity) can take up to $210,000 (2026 premium limit, IRS Notice 2025-67 — same $210k as 2025; the 25% cap is gone after Dec 28, 2022) out of the RMD base, with income starting no later than the month after 85. That is a longevity hedge, not a sequence hedge. Do not put the whole IRA in one carrier. Do not skip a quote because a blog said annuities are a scam, and do not buy one with the sleeve. If the quote’s payout is worse than delaying Social Security (step 6), delay Social Security first — SSA’s 8% delayed credit is inflation-linked and has no commission.
Hard-stop list: a fixed real $48k through a −25% year because “4% is safe”; retiring into 80/20 with a $0 sleeve; claiming Social Security at 62 then working over $24,480; funding the 67→70 gap with a 22% card; cashing out a 3% mortgage to “have more invested” the year you stop working; taking the RMD as a stock sale in a hole when a QCD or a Roth/sleeve spend was available; converting $250k in the same year you just cut lifestyle 10%; putting the emergency fund and the sequence sleeve in I-bonds inside the 12-month lock; treating Trinity’s 30-year 50/50 sample as a 50-year FIRE paycheck; skipping Part B at 65 because you delayed claiming. If the sleeve in step 4 does not exist, you wanted more work, a smaller spend, or a later date — not a 4% slogan.
One-page decision
Honest spend ÷ nest egg is the rate. Draw the two-path picture. Fund 2–3 years of that spend in T-bills/HYSA before the last paycheck. Write the 10% cut at 1.2× starting rate the year you retire. Delay the high earner’s Social Security if the gap is not a card. Bond-tent the first five years; do not 80/20 on the way out the door. Harvest the hole in taxable; do not IRA-draw it. RMDs at 73/75 still fire. Annuities are a quote, not a personality. Nobody is required to retire on the 4% poster the year the statement is red.
Worked example (educational, not advice): $1.2M, $48k/year, five calendar returns. Crash-first (−25 / −15 / +20 / +15 / +10) ends at $864,952. Reverse ends at $967,722. Same average, $102,770 less because shares were sold in the hole. After year-1 crash the remaining pile is $852k and the implied rate is 5.63% — above the 4.8% Guyton-style trigger — so the next check is $43,200, not $48k. A $144k T-bill sleeve (3× $48k) means year 1 spends cash, not the stock lot: stocks fall 25% on $1,056k → $792k, cash $96k, total $888k vs $852k all-in. You still own the recovery.