Model integrity
These models assert their own accounting and refuse to report a result when one of the checks fails. Those checks run every time this site is built, and this page is their output — not a list somebody typed.
All 28 checks passed in this build. Last run November 1, 2026.
Each one re-derives its quantity independently rather than reading a number the engine already computed and comparing it against itself. That distinction matters: a model checking its own arithmetic against its own arithmetic will always agree with itself.
Rent vs. Buy
PASSPrincipal is conserved
Every dollar borrowed is either repaid or still owed. Repaid principal plus the remaining balance must equal the original loan, at every point in the schedule.
residual $0.00
PASSBoth households spend the same
The rule the whole comparison rests on. If either side were handed money the other did not get, every figure downstream would be meaningless.
difference over 30 years $0.00
PASSMortgage cash reconciles
Total cash paid on the loan equals cumulative interest plus cumulative principal.
residual $0.00
PASSThe portfolio adds up
The renter's final balance is rebuilt from the starting cash and every monthly contribution, compounded independently of the engine's own running total.
rebuilt $1,159,898.66 vs reported $1,159,898.66
PASSThe loan retires on schedule
A fixed-rate loan left alone must amortise to exactly zero on its final scheduled payment.
balance at month 360 $0.00
Car: Lease vs. Buy
PASSEvery dollar is dated correctly
Each option's cost is rebuilt by compounding every outflow from the month it is paid to the horizon. A dollar paid today and a dollar paid in five years must not be treated as equal.
worst deviation across five options $0.00
PASSThe chart and the table agree
The final point of the committed-cost chart is the same number as the net cost in the summary table.
worst deviation $0.00
PASSCash and financing converge
A loan of nothing at zero percent must cost exactly what paying cash costs. If the two ever diverge, the financing arithmetic is adding or losing money.
difference $0.00
PASSLease obligations are carried
A lease cannot be walked away from. Any contract still running at the horizon is charged as the present value of what is still owed, rather than vanishing.
outstanding at horizon $0.00
Rental Property vs. Investing
PASSBoth sides invest the same
The landlord and the investor start with identical capital, and every month the property needs feeding the investor contributes the same amount. Lifetime out-of-pocket must match.
difference $0.00
PASSPrincipal is conserved
Repaid principal plus the outstanding balance equals the original loan.
residual $0.00
PASSHigher appreciation never hurts the landlord
The sensitivity grid is checked for monotonicity in both directions. Linking operating costs to market value can break this, and when it does the grid says so rather than presenting a misleading picture.
grid is monotone in both directions
Health Plan: HSA vs. Copay
PASSWhat you pay matches the plan rules
Each plan's bill for a year of care is rebuilt from its copays, deductible, coinsurance and maximum by a separate function, at a quarter, one, two and a half and ten times your care, and compared to the engine's.
worst deviation $0.00
PASSThe out-of-pocket maximum holds
No plan ever charges more than its maximum, and a five-million-dollar year charges exactly the maximum on every plan whose hospital stays go through the deductible.
a $5,000,000 year: HDHP + HSA $3,750.00, POS $3,000.00, Kaiser HMO $445.00
PASSMore care never costs less
Every option's yearly net cost is checked at forty-one levels of care, from none to five times yours. Needing more care must never make a plan cheaper.
0 decreases across 41 levels
PASSEvery household spends the same
The rule the comparison rests on. Each year every option's household spends the same take-home pay on health, and the cheaper options invest what is left. Lifetime spending must match across all of them -- one person's four and a couple's fourteen -- and each taxable account is rebuilt by compounding its yearly deposits independently.
spending gap $0.00 over 30 years; account residual $0.00
PASSThe HSA adds up, through a switch
The HSA of the option that leaves the HDHP is rebuilt from every deposit before the switch -- the full IRS limit each year, catch-up included, read from the limits table rather than the engine -- none after, every withdrawal for care on either plan and every year's growth, independently of the engine's running total.
rebuilt $257,406.17 vs reported $257,406.17
PASSOne year is the familiar yearly cost
With one year and nothing growing, each option's net cost -- the number every result on the page is given in -- must equal the yearly net cost everybody quotes: premiums, plus care, minus the HSA money. For one person and for a couple on two employers' plans. The two views of the same plan cannot disagree.
worst residual $0.00
PASSMedical inflation and age move the bill as they should
Prices, deductibles, maximums and copays rise together, and the care itself grows with age, so a later year is priced directly at that year's prices, plan terms and amount of care and compared to the bill the model carries.
year 10 worst deviation $0.00
PASSA couple's care is priced on the right policies
Two people on their own plans pay two separate bills, each with its own deductible; one plan covering both pools their care against the family deductible and maximum. All twelve arrangements are rebuilt from the plan rules and compared.
worst deviation $0.00
PASSA couple on two employers' plans, and an HRA
With your spouse on their own employer's plans, each of the nine pairs is rebuilt from the two plans' rules every year, and the spouse's plans covering both in year one. The HRA's balance is carried by hand: each year's allocation added, cut to the deductible, spent on everything but prescriptions -- in a year light enough for it to roll over.
worst deviation $0.00
US Tax
PASSTax never falls as income rises
Progressive brackets must be monotone: earning one more dollar can never reduce the tax bill. A table entered out of order, or one that stops short of the top rate, breaks this and produces a plausible-looking number for everybody below the break.
smallest step $0.00, highest average rate 33.11% against a 37% top marginal rate
PASSA deduction is worth what it actually saves
The benefit of itemising is rebuilt by computing the whole tax bill twice -- once taking the standard deduction, once itemising -- and differencing them. Pricing the deduction at a single marginal rate instead is the common shortcut, and it is wrong whenever the deduction spans a bracket, which a mortgage routinely does.
residual $0.00
PASSA deduction under the standard deduction saves nothing
Since the standard deduction was doubled, most households get no benefit at all from the mortgage interest deduction. A calculator that quietly credits them with one overstates the case for buying in the most common case there is.
benefit on $31,200.00 of deductions is $0.00
PASSTax settings do nothing while tax is off
The tax model ships switched off, and every figure published here is a pre-tax figure. Changing a tax input while the model is off must therefore change nothing at all -- otherwise the pre-tax answer is quietly a partly-taxed one.
difference with and without tax inputs set: $0.00
PASSBoth households still spend the same, with tax on
The deduction is modelled as a reduction in what owning costs, not as a payment to the buyer. If it were the latter the buyer would be handed money the renter never gets, and the comparison would no longer be between two uses of the same income.
difference over 30 years $0.00
PASSDepreciation claimed is depreciation recaptured
Depreciation shelters income now and is clawed back at sale, at a rate above the capital-gains rate. On a sale into a gain, every dollar deducted over the hold must reappear as unrecaptured section 1250 gain. A model that takes the deduction and forgets the recapture reports a tax break nobody receives.
claimed $212,160.00 vs recaptured $212,160.00
PASSAll three calculators tax a portfolio identically
The same brokerage account appears on one side of all three comparisons, and all three ask us_tax what it costs. If they ever disagreed, a reader moving between the pages would see the models differ where only the decisions should.
drag at a 7% return: rent vs. buy 4.3069%, rental 4.3069%, car 4.3069%
What this does not tell you
That the arithmetic is sound is not the same as the model being right for your decision. These checks cannot tell you whether 3% appreciation is a sensible assumption, whether the tax treatment left out of a rental model matters in your bracket, or whether the thing you are really deciding is financial at all. Each calculator’s methodology page sets out what it deliberately does not model, and those omissions are usually a bigger deal than any rounding.
Beyond these, the project runs a test suite of several hundred cases covering the same invariants across edge scenarios — zero-rate loans, hundred-percent deposits, one-month horizons, price crashes — plus an independent reference implementation of the rent-vs-buy model, written from its published economics rather than its code, which reproduces every headline figure to the cent.