Retirement Plan Optimizer

Why together. Claiming later opens years with little income, which conversions can fill at low rates. Conversions shrink the RMDs that later make Social Security taxable and lift Medicare premiums. The withdrawal order decides which bracket you are in while both happen. Each solo search holds the other two decisions where you set them, so it can't see those trades.

What is searched. Whole-year claim ages from today, or 62, through 70 for each spouse; no conversions, filling the 12%, 22% or 24% bracket, or staying under the first or second IRMAA line; and five withdrawal orders, plus anything different you entered. Claim ages are searched every other year against every combination, then year by year around the five leading plans.

Benefits. From the benefit at full retirement age: reduced 5/9% a month for the first 36 months early and 5/12% beyond, raised 2/3% a month after it to 70. A lower earner is topped up to half the other's full benefit once both have filed. After a spouse's death the survivor keeps the larger benefit, under the widow(er)'s limit. Benefits start in the plan year you reach the claim age.

Ranking. Plans that leave spending unmet rank last. The rest rank by after-tax wealth at the end of the plan in today's dollars: Roth and taxable at full value, pre-tax less your heirs' tax rate. Your plan-through age and your spouse's age at death decide how much delay is worth, so try more than one.

Stress test. The ranking uses the steady return your average and volatility compound to. The five headline plans are then run through the same seeded random markets, so their survival rates and worst-tenth estates can be compared without the luck of the draw deciding it.

Tradeoff search and retention. The optional explorer evaluates every combination in the existing bounded menu, at most 3,402. It retains up to 32 distinct policies: the entered policy and separate legacy, tax and shortfall leaders, then alternating objective rankings. Legacy and tax rankings prefer plans without steady-return spending gaps. Screened-out policies do not receive market evaluation; this is best among evaluated candidates, not a global optimum.

Frontier and ties. A policy dominates another only when it is at least as good on every frontier objective and strictly better on one: steady-return after-tax legacy, lifetime tax, spending shortfall and common-market success when available. Monetary comparisons use cents; success uses six decimal places. Equal objective vectors share a stable representative, and the entered policy remains retained. P10 and P50 are diagnostic estimates, not additional frontier objectives. No objective weights are assigned.

Annual ledger and units. Annual rows reuse the household engine's steady-return projection and calendar years, capped at 70 plan years. Ending wealth and lifetime tax summaries use today's dollars; annual tax, conversions and spending gaps use nominal dollars. Shortfall is the sum of nominal unmet spending, not a discounted cost. Pre-tax legacy uses the entered heirs' tax rate. Charts show steady annual rows, not Monte Carlo percentile paths; this report includes all four annual metric tables for the selected pair.

Common markets and separate validation. Every retained policy faces the same seeded lognormal return paths, using the entered mean and volatility and at most 300 runs. Success means spending was met every year. Zero runs means unavailable, not zero success. Separate validation evaluates only captured finalists with the next unsigned 32-bit seed. Its results do not select policies or alter the original frontier. Distinct seeds provide a separate reproducible path set, not proof of statistical independence or a guarantee of future outcomes.

Reference sources. These sources support the component rules. Model choices and exclusions are described in the notes above.SSA: period life tableIRS: retirement distributions