Model scope and assumptions
Prelaunch model versions: connected plan 1.1.0; foundation 1.1.0; investment scenarios 2.1.0; life models 2.0.0. Code checks cover specified cases; independent professional validation and real-device/screen-reader testing are still required.
A connected plan, in a clear order
Household records assets and debt terms. Budget sets income, living costs and distinct allocations. Debt payoff and Emergency fund use those figures; Financial freedom and Plan overview share the resulting portfolio projection. Shared inputs are edited at their source. Changed figures require review before saving a dependent result.
Each asset has one purpose. Emergency cash and other goal assets stay outside the retirement projection. Pension pots have their own access ages: gross balances count in net worth, while the projection uses an explicitly tax-adjusted value only from access. Freed debt payments and unused emergency contributions become investments only when you choose this in Budget. Other goal allocations remain reserved; costed milestones from Your future can be explicitly assigned an ongoing Life goals savings amount in Budget after a preview. The amount continues until the user changes it; it does not stop automatically at a goal date or model the purchase itself.
Connected projections use annual-average monthly EUR amounts over up to 720 months. Budget separately shows the actual selected bill/income months and any shortfalls requiring advance cash. Financial freedom shows accumulation towards an illustrative target, not the sustainability of retirement withdrawals. Continuing debt costs need to be included in expected retirement spending.
How pensions connect
Household owns pension inputs and optional provider, country, native-currency, manual FX and verification details. Its EUR amount funds the model; native amounts change it only through an explicit conversion action and a subsequent save. Pensions & policies shows these central records. Pension pots and pension income are different records. Do not include both a pot and the income that same pot is expected to buy. Defined benefit and state income have no capital value in this plan.
Budget owns pension saving paid from take-home income. Household assigns a percentage of that amount to each pot. Employer payments and deductions already outside take-home pay are entered on the pot separately. Unassigned Budget pension saving stays reserved and is explicitly excluded from the forecast.
Financial freedom and Plan overview use the same model. Retirement & mix uses the same pension growth and access assumptions when copied, but tests monthly spending through a chosen end age and return sequence, so its target differs from the Free withdrawal-rate target. While a pension is locked, its stated return applies; once available in Premium retirement drawdown, the selected common retirement sequence applies. Earlier source changes do not automatically overwrite a Premium copy.
All pension estimates use today’s money and EUR in the connected plan. Pension income is assumed to maintain buying power; eligibility, indexation caps, progressive withdrawal taxes and survivor benefits are not calculated. For a couple, translate each partner’s pension date into the same anchor person’s age. Record tax estimates and access ages from your own scheme and situation; no country entitlement is inferred.
General pension distinctions checked on 23 September 2026: MoneyHelper: pension pots, MoneyHelper: defined benefit income, and Your Europe: claiming pensions across countries. These sources explain the distinctions; they do not validate this calculator.
Your first retirement illustration
Your future offers an optional illustration from rough estimates before Household is completed. The visible 3% real return and 3.5% withdrawal assumption are editable examples, not predictions or guaranteed safe rates. Unknown pension benefits leave a partial picture. The full connected result uses saved steps 1–6; the provisional inputs never replace them. One-off milestones and inheritance are excluded from this early spending comparison.
How to interpret a result
Calculators produce deterministic illustrations, not forecasts, probabilities or suitability assessments. Check periods, currency, inflation basis, access restrictions and tax assumptions. A saved result keeps its inputs; changing another tool does not silently change that scenario.
| Tool | Supported model and main limit |
|---|---|
| Budget, debt and emergency cash | Recurring bill frequencies, explicit saving allocations, monthly amortization and cash accumulation. No bank feeds. Separate scenarios default to 600 months; connected scenarios use 720 months. Unresolved balances are disclosed. |
| Financial freedom · Free | Annual effective real returns compounded monthly. Connected pension pots have individual returns, contributions, access ages and flat withdrawal-tax assumptions. Pension income reduces retirement needs from its start age; it is not automatically added to working Budget contributions. Available money must fund each month before later pensions can help. The Free heuristic deducts discounted locked-pot value from the withdrawal-rate target and applies a separate monthly liquidity floor before access. No contributions after a candidate stopping date are credited in that bridge check. It does not require the same target to remain intact after each future withdrawal. Deterministic accumulation and a withdrawal-rate target; not a lifelong success probability. |
| Retirement & mix · Premium | Per-holding pre-retirement returns, timed lump sums, monthly contributions and withdrawals. Retirement sequences start with the same projected capital. Only selected accessible holdings count immediately. An explicit Household pension copy adds dated access and multiple income streams; duplicated manual pension rows are excluded while that copy is active. State pension is income. Tax drag is an effective scenario assumption, not a country tax engine. Mix changes do not receive an automatic extra-return credit. |
| Portfolio risk | Local CSV / spreadsheet-paste import; ticker or ISIN plus dated EUR statement values need no exchange. Unit-priced quotations need exact listing and FX. Concentration and group stress scenarios work before allocation comparisons. Risk-group assumptions are editable illustrations; historical analysis requires 36–600 consecutive monthly total returns in EUR for every security and a named benchmark. See the detailed portfolio method below. |
| Rent or buy | Equal initial cash and monthly budgets; retained cash or investments, monthly mortgage payments and sale equity. Includes entered purchase and selling fees. Fixed-rate, stable running-cost assumptions; local rules need user inputs. |
| Lease or buy | Loan payments within the horizon plus settlement of the outstanding loan on sale. Lease deposit reduces capital cost. A partial final lease includes remaining payments and return fees as an illustrative exit liability; actual termination quotes may differ. Taxes use a simplified percentage basis. |
| Wedding and trip | Wedding contributions arrive in their stated month; initial cash counts once. Percentages drive supplier costs, not guest reductions. Trip costs distinguish travellers, rooms, nights and days. Goals must not reuse the same cash or monthly savings. |
| Pensions & policies · Premium | Native-currency account balances and annual income remain separate. Life-cover comparison uses the selected insured person and explicit conversion to EUR. State pensions are excluded from capital and beneficiary-nomination checks. |
| Stress test | Independent cash-flow scenarios with entered benefits, waiting periods and duration. N/A cases are excluded from the descriptive score. No market recovery date or combined-shock forecast. |
| Tax | AT and DE 2026 general individual income-tax tariffs; UK 2026/27 ordinary non-savings case outside Scotland with allowance taper; US 2026 single ordinary-income case with standard deduction. Eligibility, other taxes and cross-border relief are outside the calculation. Official sources are linked on the tool. |
| Estate | Inventory, intended shares, country context and printable briefing. Reserved shares and tax are supplied assumptions; blank means unknown. Automated legal entitlements and estate/per-heir tax for the twelve contexts are not validated or activated. |
Records and access
Household fields use EUR. Explicit imports preserve scenario control. Financial records remain in browser storage; no account database, bank link or payment entitlement is connected. The data controls export or erase records for this browser.
Review before release
Production authentication, database access controls, recovery, billing and signed webhook handling remain to be connected. The operator’s legal details and final policies require completion. Every specialist model needs independent sign-off before customers rely on it.
Portfolio risk and allocation comparisons
Model portfolio-1.0.0 supports 2–80 distinct positive holdings for optimization, with up to 250 rows in the worksheet. Duplicate ticker / exchange / quote-currency combinations are combined. Missing exchanges are explicitly unverified; the same ticker on different markets may identify different securities. Holdings can be stocks, funds, bonds or cash, valued in EUR. Individual bonds use supplied market values; coupons, accrued interest, duration, default, maturities and price-as-percent-of-par conventions are not calculated.
The assumption mode uses editable nominal EUR annual arithmetic returns and volatilities by risk group. Initial examples are shares 6% / 20%, bonds 3.5% / 7%, property 5.5% / 23%, and cash 2% / 1%. They are uncalibrated illustrations, not ticker estimates, forecasts or recommended inputs. Within-group correlation starts at 0.8; between-group correlation at 0.35. The model combines a common factor, group factors and independent factors; requiring 0 ≤ between ≤ within ≤ 1 ensures positive semidefinite covariance. It does not measure sector, issuer, country or underlying fund overlap.
Historical inputs must be complete monthly total returns in EUR, including distributions and consistent FX conversion, for all positive holdings and the selected benchmark. Missing months, duplicate columns and ambiguous ticker mappings are rejected. At least 36 observations are required; a longer sample still may not represent future conditions. Arithmetic sample means are annualized ×12 and sample covariance ×12. Comparison covariance shrinks off-diagonal terms towards zero by the user’s percentage (25% initially). This is fixed diagonal shrinkage, not an implementation of the estimated optimal Ledoit–Wolf shrinkage coefficient. The observed historical volatility uses the unshrunk sample covariance.
Current weights are held constant with monthly rebalancing for the historical illustration. Growth compounds monthly portfolio returns; annualized growth uses geometric compounding. Drawdown is measured at month ends, so intramonth losses can be larger. The benchmark is supplied by the user and should fit the portfolio’s opportunity set; a stock-only index may be unsuitable for a mixed stock/bond portfolio. Beta and annualized alpha come from ordinary least-squares regression of monthly portfolio excess returns on benchmark excess returns. The annual cash input is converted to a constant monthly effective rate. Alpha is the monthly intercept ×12, not a forecast or proof of skill. The approximate interval uses ±1.96 ordinary regression standard errors ×12; it is not robust to serial correlation, changing exposures or small-sample effects. Constant benchmarks produce undefined alpha/beta.
Risk contribution is each holding’s contribution to portfolio variance: weight × covariance-with-portfolio ÷ portfolio variance. Contributions sum to 100% when variance is positive; a negative contribution can represent a hedge. Effective holdings (1 ÷ sum of squared weights) measures weight concentration only. Group stress scenarios are explicit shocks, not probabilities, worst-case bounds or historical loss estimates.
Allocation comparisons are fully invested, long-only and restricted to existing positive holdings. Every weight observes the user’s maximum weight and maximum absolute change per holding. A projected-gradient convex quadratic solver generates 25 sampled points by varying return preference, including the minimum-variance solution; a nonconverged result is rejected. “Higher model Sharpe” selects the highest ratio among these points, not a guaranteed global maximum. Sharpe is annual arithmetic return less the annual cash input, divided by annual volatility. No separate cash allocation is introduced unless cash is already included as a holding.
One-way turnover is half the sum of absolute weight changes. Estimated costs apply to both purchased and sold amounts at the entered rate, and are reported separately from gross model returns. Trades balance before costs; costs require extra cash or smaller purchases. Taxes, minimum lots, account eligibility and trading restrictions are excluded. The solver is not a trade engine. Results never overwrite holdings or Household. In historical mode the same sample is used for fitting and comparison, introducing in-sample optimism. Today’s holding selection also creates hindsight and survivorship bias. Independent validation and out-of-sample evaluation are necessary before a commercial investment-decision service.
Method references: Ledoit and Wolf: covariance estimation and shrinkage; CFA Institute: estimation risk in portfolio optimization. Provider constraint: Twelve Data price adjustments — price history is not automatically a dividend-inclusive total-return series. References inform model design and limitations; they do not validate this implementation.