Sequence-of-returns risk

This is the single most important concept for early retirees to understand. While you're working, sequence doesn't really matter. You're a net buyer of stocks, so a market crash early in your career is actually helpful (you buy cheap). The day you retire, sequence flips: you become a net seller, and an early crash means you're selling at the worst possible time. Recovery years don't help if you've already withdrawn the capital that would have ridden them up.

Conventional retirement calculators that average annual returns hide this risk entirely. They report a tidy number based on the long-run average. Monte Carlo simulations expose it: 10,000 different sequences with the same average produce a wide range of outcomes, and the worst 10-25% are dominated by sequences where the first 5-7 retirement years went badly.

Worked example. Two retirees, both starting with $1M at age 65, both withdrawing $50,000/year inflation-adjusted for 30 years. Both face the exact same 10 annual returns: -15%, -5%, -10%, +5%, +12%, +20%, +25%, +18%, +12%, +8%. Long-run average return: about 7%. The only difference between the two retirees is the order.

Retiree A gets the bad years first (the -15%, -5%, -10% sequence early), then the recovery. After year 3, her portfolio is worth roughly $600K, already cut nearly in half by the combination of withdrawals and market losses. Even though the recovery years are strong, she's now compounding from a much smaller base. She runs out of money around year 22.

Retiree B gets the good years first. The 20%, 25%, 18% sequence pushes his portfolio to about $1.4M after three years even after withdrawals. The bad years arrive later, but by then he's compounded so much extra base that the losses are manageable. He ends the 30 years with roughly $800K still left over.

Same retirees. Same total withdrawals. Same average annual return. Wildly different outcomes, purely because of the order. That's sequence-of-returns risk, and it's why a single-projection retirement calculator (which inherently picks one sequence, usually a smooth one) is dangerously incomplete. A Monte Carlo simulation runs 10,000 different orderings and reports the share that ended badly.

Mitigations. Three approaches are well-supported: keep 1-3 years of expenses in cash so a market crash doesn't force you to sell at the bottom; have a part-time income stream during the first 5 retirement years that can flex up if markets crash; build flexibility into your spending plan so a -25% portfolio year triggers a 10-15% spending cut for that year. None of these eliminate sequence risk, but each meaningfully reduces it.