How Beagle calculates your retirement projection

Accounts, expense items, debts, and property are shared across scenarios. Income assumptions can be overridden within a scenario, including amount, start age, and end age. Scenario profiles and Roth conversions are independent. Changing a claiming age does not automatically recalculate a Social Security benefit; enter the corresponding benefit amount yourself.

Income starts in the year you reach its start age and is paid through the year you reach its end age. Salary and part-time income with a blank end age stop at the scenario's retirement age, so retiring at 65 still includes a full year of that income at age 65.

Holding market values are used when available; otherwise the manual account balance is used. 401(k) and traditional IRA balances form the pre-tax bucket. Roth and HSA balances form the tax-free bucket. Brokerage and cash are tracked separately. A portfolio override scales starting balances, bond coupons, and maturities together.

Equity returns use a starting-balance-weighted average within each tax bucket: holding override, then account override, then the profile rate. Bond tickers beginning UST, BOND, CORP, or MUNI are separated. Cash uses the blended holding yield, falling back to the profile cash rate when that yield is zero.

Regular projections, historical paths, Monte Carlo paths, and Roth comparisons use one annual accounting engine. Each year, income, required distributions, conversions, expenses, taxes, and withdrawals are applied first; the year's return then applies to what remains. This start-of-year withdrawal timing is the convention used by Bengen (1994) and the Trinity Study, and is more conservative than withdrawing after growth. Historical paths change annual returns and inflation; Monte Carlo changes annual returns.

Bond principal stays at its modeled value until maturity and returns to its original tax bucket. Coupons in retirement accounts stay inside those accounts. Taxable Treasury interest is federally taxable and state-exempt; municipal coupons are included in cash income and excluded from modeled income tax. Coupons stop in the maturity year. Property equity is illiquid and never funds spending.

RMDs start at 75 for birth years 1960 onward and 73 for 1951-1959; older cohorts use their earlier start ages. The prior balance includes pre-tax bonds. If liquid pre-tax funds are insufficient, required distributions liquidate bonds at their modeled value and reduce their later coupons and maturities proportionally. Spouse-specific or inherited-account tables are not modeled.

The engine uses 2026 IRS ordinary-income and long-term capital-gains thresholds, with basic standard deductions of $16,100 single, $32,200 joint, and $24,150 head of household. These nominal tables are held constant in future years; no future inflation indexing is assumed. Age-65 additions cover only the modeled person. The enhanced senior deduction phases out with income and expires after 2028 in projected calendar years. Itemized deductions and most other credits are not modeled.

Taxable brokerage dividends are treated as qualified dividends at capital-gains rates, stacked above ordinary income. Unused deductions offset gains. Brokerage sales consume proportional cost basis and positive realized gains are taxed at long-term rates. Holding basis is shares times basis per share; balance-only accounts use their total cost basis. Missing basis uses your estimated unrealized gains percentage when provided, clearly flagged as an estimate; otherwise it is modeled as zero and flagged. Known holding basis takes precedence. New savings add basis; investment growth does not. This is a planning approximation, not tax-lot accounting. Short-term gains, capital-loss deductions/carryforwards, and net investment income tax are not modeled. Social Security inclusion is a flat 85% approximation, not the provisional-income calculation.

States use fixed effective-rate approximations, not current graduated state tax schedules. Treasury interest is excluded from their tax base. Tax on pre-tax withdrawals and brokerage gains is funded by additional withdrawals using an iterative calculation with a one-cent tolerance.

Base expenses, enabled expense items, recurring large purchases, loan payments, and scenario expense adjustments determine the budget. Expenses inflate from today; historical paths compound each historical inflation observation. Disabled and expired expenses are excluded consistently.

This replaces the fixed budget and adjusts spending as the portfolio changes. It is not the conventional rule that starts at 4% of an initial portfolio and then inflation-adjusts that dollar amount. Success in this mode does not establish that your fixed lifestyle budget is affordable.

Cash above the reserve floor funds a shortfall first, followed by brokerage, pre-tax, Roth, and finally reserved cash. Before age 59.5 the pre-tax/Roth order is reversed. This is a fixed sequence with estimated taxes, not an optimizer. Bonds are held to maturity for discretionary spending. Penalties and Roth qualification rules are not modeled.

A shared random shock is added to each equity bucket return each year; returns are floored at -100%. Cash yields and bond coupons stay fixed. The default run count is 1,000. Portfolio mix comes from recognized holdings and defaults to equity when holdings are absent. This is a simplified normal model, not historical resampling or an asset-class correlation model.

A positive ending balance cannot hide an earlier unpaid expense. A later recovery does not erase failure. Exactly funding the final year counts as success even with a zero balance. The percentile curves describe balances, not guaranteed outcomes or a guarantee of any particular spending level.

Historical returns and inflation are replayed from the selected year. Once available history ends, a disclosed tail uses 7% returns and 3% inflation. Taxes still use the 2026 planning baseline. Benchmark modes replace non-cash asset returns and do not separately model bond coupons or maturities.

Six presets compare no conversions, $10K, $25K, $50K annually, and filling the 12% or 22% bracket. Presets operate through the year before the applicable RMD age. Applying a strategy atomically replaces the scenario conversion schedule and saves the actual bracket rule, recalculating its amount each year. It does not save a flattened annual average. Manual conversions can continue after RMD age; the RMD is distributed first.

Bracket room uses modeled income, gains, and deductions before additional withdrawals required to fund spending and conversion taxes. Those withdrawals can push final income above the target bracket. Conversions use liquid pre-tax funds; there is no optimization of bond sales, capital gains, or future healthcare premiums.

IRMAA checks final modeled MAGI, including withdrawal taxes and municipal interest, against 2026 thresholds. Income at age 63 concerns premiums at age 65. Future thresholds are held constant and actual premium costs are not charged. ACA warnings use 400% of 2025 poverty guidelines for a 2026 coverage baseline ($62,600 single / $84,600 joint); non-taxable Social Security is included. Dependents are not modeled. No ACA threshold warning is calculated for Alaska, Hawaii, or married-separate filing. These warnings do not establish subsidy eligibility.

When a fixed-spending plan fails, a whole-dollar search reruns the same full engine with lower base spending. Conversions, RMDs, recurring expenses, and debt payments remain included. If even zero base spending cannot fund mandatory costs, no safe amount is reported.

The search reuses up to 200 sampled paths and returns a tested lower bound that funds every year in at least 90% of that sample. The search range ends at five times current base spending. It is a sample estimate, not a guaranteed maximum. Itemized costs in additive mode remain extra, and results can change between runs.