Independent verification
Check our arithmetic yourself
This site was built by one engineer with heavy use of AI tooling, and it publishes money figures. You should not take our word for those figures — so here is how to reproduce every one of them in a spreadsheet you already trust.
The whole exercise takes about ten minutes. Five formulas, one government calculator, no account and no download. Two of the six results deliberately disagree with us, and the size and cause of each disagreement is stated below before you run it.
Why a spreadsheet is the right referee
Excel and Google Sheets ship the same set of financial functions — PMT, NPER, FV. They implement the standard formulas, they have been scrutinised by accountants for three decades, and they were not written by us, by our tooling, or by any AI. If our engine agrees with them, that is evidence that came from outside this project.
That last point is the whole argument. A test comparing our code to our own expectations proves only that we are self-consistent. It says nothing about whether we are right. Every check below is anchored to something published by somebody else.
The five spreadsheet checks
Open a blank sheet. Paste each formula into a cell and compare.
Check 1 of 5
The monthly payment on a loan
=PMT(0.06/12, 60, -10000)
- The spreadsheet says
- 193.3280…
- QuickOper says
- $193.33· agrees
The payment our debt fixture uses for $10,000 at 6.00% over 60 months. We round up to the cent, as a lender does.
Check 2 of 5
How long it takes to clear
=NPER(0.06/12, -193.33, 10000)
- The spreadsheet says
- 59.99…
- QuickOper says
- 60 months· agrees
The fraction below 60 exists because $193.33 is a rounded-up payment. It is also why our final payment is $193.21 rather than a full one.
Check 3 of 5
Total interest over the term
=193.33*60 - 10000
- The spreadsheet says
- 1,599.80
- QuickOper says
- $1,599.68· differs, on purpose
Twelve cents lower, and the twelve cents has a specific cause: we charge the reduced final payment of $193.21 rather than a full $193.33. Subtract that $0.12 and the two agree exactly.
Check 4 of 5
Investment growth over thirty years
=FV(0.07, 30, 0, -100000)
- The spreadsheet says
- 761,225.50
- QuickOper says
- 761,225.08· differs, on purpose
Forty-two cents apart across 360 monthly steps — about five parts per billion. The spreadsheet carries full precision to the end; we round to the cent every month, the way a statement does.
Check 5 of 5
The coast number
=1000000/1.05^30
- The spreadsheet says
- 231,377.45
- QuickOper says
- $231,377.45· agrees
Exactly what the Coast FIRE calculator shows for a $1,000,000 target thirty years out at a 5% real return.
A sixth check, from the US government
The Securities and Exchange Commission publishes a compound interest calculator at investor.gov. It is about as disinterested a source as exists: a regulator with no product to sell and a statutory interest in the arithmetic being right.
Enter $100,000 initial, $0 monthly, 30 years, 7% interest, compounded annually. It returns $761,225.50 — the same figure the spreadsheet gives, from a third independent implementation, and the figure our Coast FIRE projection lands within 42 cents of.
When a figure disagrees, find out why
Do not assume the calculator is right. Do not assume the spreadsheet is right either. There is always a reason, and it is almost always one of three stated assumptions rather than a mistake.
1. Compounding convention
Our debt tool divides the annual rate by twelve, because that is how a lender quotes an APR and how a statement computes one month's charge. Our investment tool uses (1 + r)1/12 − 1, because a stated annual return compounds to that figure over a year. At 7% those are 0.5654% and 0.5833% a month — close enough to look like a rounding error and different enough to move a thirty-year projection by several percent, always flatteringly. Excel's PMT uses the first convention and FV the second, so comparing across the two will disagree, correctly.
2. When rounding happens
We round to the cent every period, because that is what a statement does and what you will be charged. A spreadsheet carries full precision and rounds once at the end. Over 360 periods that produces cents of difference — the 42 cents in check four — and never dollars. If you ever see dollars, something else is wrong.
3. The final payment
Ours is reduced to exactly what remains, because a lender does not take a full payment when less is owed. Most closed-form answers assume every payment is identical. That is the entire twelve cents in check three.
If a difference is larger than a few cents and none of those three explain it, that is a bug. Please tell us, with the inputs you used.
What this proves, and what it does not
It proves that our arithmetic matches independent, long-established implementations of the same formulas — and that where it differs, it differs by a stated amount for a stated reason.
It does not prove that the formula is the right one for your situation, that our assumptions match your lender's contract, or that a projection will resemble the future. Nothing can prove those. It is why every calculator here states its assumptions on the page, and why this site computes rather than advises.
The four things that make a silently wrong figure unlikely
- The calculation is isolated. Every engine is a pure function with no access to the page and no framework attached. It can be wrong arithmetically, but not through a rendering accident.
- Fixtures are anchored outside the codebase. Each engine is tested against a figure published by somebody else — the five above.
- Assertions are exact, not tolerant. Two early drafts used loose bounds — "under a dollar", "within 50 cents" — that were fifty times wider than reality and would have absorbed a genuine regression silently. Both were replaced with the exact figure.
- A human merges every calculation change. Automated checks can prove the code matches the fixture. They cannot prove the fixture matches reality. That is a person's job, and it is why changes under the calculation directory require one.
Sources
- US Securities and Exchange Commission — Compound Interest Calculator — retrieved 2026-08-08
- Microsoft — PMT function (Excel) — retrieved 2026-08-08
- Microsoft — NPER function (Excel) — retrieved 2026-08-08
- Microsoft — FV function (Excel) — retrieved 2026-08-08
Written and maintained by Vikash Singh. Last verified 2026-08-08. Every figure on this page is also asserted by a test in the public repository, so it fails the build before it can go stale here.