Stress-testing a retirement plan against a bad first decade
Take two retirements with identical average returns over thirty years. In one, the bad years come early; in the other, the same bad years come late. The first portfolio can run out while the second finishes with a surplus. Same average, opposite outcomes. That gap is sequence-of-returns risk, and it is the single most important thing a retirement projection can hide from you.
Why withdrawals change everything
For an untouched lump sum, the order of returns doesn’t change the ending value at all, because compounding is commutative when no money moves. Once cash flows enter, order starts to matter: with ongoing contributions, a late crash hits a bigger balance than an early one. But it is withdrawals that make sequence dangerous rather than merely relevant. Selling shares into a downturn converts a temporary loss into a permanent one: those shares are gone and can’t participate in the recovery. A bad first decade forces years of selling low; the identical decade arriving late hits a portfolio that already grew and is nearly done being needed.
The classic illustration: someone who retired into the 2008 financial crisis spent their first years withdrawing from a falling portfolio, while someone who retired in 2013 enjoyed a long expansion first. Same strategy, same spending rule, very different trajectories, purely from the start date.
What an average-return projection hides
A single deterministic line, “6% every year, forever,” has zero sequence risk by construction. Every year is average, so no year forces selling low. That’s why a straight-line projection is systematically the rosiest picture of a given plan, and why a plan that merely survives the straight line hasn’t yet been tested at all.
What a real stress test looks like
Three complementary methods, in increasing order of coverage:
- Shift the start year through history. Replay the plan as if retirement began in each historical year and look at the worst cohorts. The classic worst cases cluster around retirements that walked into deep bear markets or extended high-inflation stretches. This tests against sequences that actually happened, including the inflation that accompanied them.
- Run a Monte Carlo. Simulate thousands of return sequences and look at the distribution of outcomes, not one line. The output worth reading is more than the success percentage: when the failures happen, how deep the near-misses go, and what the spread of ending balances looks like. A 90% success rate whose failures all happen at 95 is a different plan from one whose failures start at 78.
- Test the response, not just the shock. The realistic question isn’t “what if I keep spending obliviously through a crash?” but “what happens if I cut spending when the plan calls for it?” Guardrail-style rules (trim withdrawals when the portfolio falls behind, raise them when it runs ahead) can be modeled directly, so the stress test shows the size and duration of the cuts a bad decade would demand. Knowing “the worst case asks for a 12% spending cut for four years” is actionable in a way a bare failure percentage is not.
Things worth modeling (not prescriptions)
Different households blunt sequence risk differently, and each mitigation has a cost that belongs in the model rather than in a slogan: keeping a cash buffer lowers expected growth; flexible spending trades certainty of lifestyle for durability; delaying Social Security buys more inflation-indexed floor at the price of larger early withdrawals; a part-time income bridge shrinks early-year selling. None is universally right. A good tool lets you model each and read the trade-off from the ledger instead of arguing about it in the abstract.
The takeaway
An average-return line doesn’t test sequence risk at all. The informative question is what the bad sequences do to a plan, and whether the plan has a designed response for when one arrives.
The RetireGolden planner runs deterministic, historical, and Monte Carlo projections over the same year-by-year ledger, with selectable guardrail spending rules and an adjustment outlook that summarizes the cuts a bad sequence would demand: depth, duration, and frequency, not just a success percentage. The return models and their assumptions are documented and tested against independent implementations.
RetireGolden guides are educational, not financial, tax, legal, or investment advice. Historical performance does not guarantee future results; simulations are models, not predictions. Decisions with real-world consequences deserve a qualified professional who knows your full situation.