Methodology
Every figure this calculator uses, where it came from, and an explicit list of what is out of scope. Nothing is filled in from memory.
The principle
Wrong tax numbers are the one failure this site cannot recover from. So the rule throughout is: a figure is either read from a primary source or it is not used. Where a number cannot be verified it is left null and the calculator declines to compute rather than guessing. Where a figure is carried forward from a prior year deliberately, the calculator still computes but emits a warning naming the year it came from.
Every line of every calculation carries the rule applied and the authority for it. Open Show your work on the calculator to read the full derivation.
Federal sources
Inflation-adjusted figures for tax year 2026 come from IRS Revenue Procedure 2025-32 (Internal Revenue Bulletin 2025-45). The 2026 adjusted items are in SECTION 4 of that document — not Section 3, which is taken up by modifications to the prior year’s Revenue Procedure. The widely repeated “§ 3.03” citation for the capital gains table is the previous year’s numbering and is wrong for 2026.
| Figure | Source | Indexed? |
|---|---|---|
| Seven ordinary rate brackets, four filing statuses | Rev. Proc. 2025-32 § 4.01 (IRC 1(j)(2)) | Yes, annually |
| 0% / 15% / 20% capital gain ceilings | Rev. Proc. 2025-32 § 4.03 (IRC 1(h), 1(j)(5)) | Yes, annually |
| Standard deduction | Rev. Proc. 2025-32 § 4.14(1) (IRC 63(c)(2)) | Yes, annually |
| Net investment income tax, 3.8% | IRC 1411(a)–(b) | No — statutory since 2013 |
| Capital loss offset, $3,000 / $1,500 | IRC 1211(b) | No — unchanged since 1978 |
| Indefinite loss carryforward, character preserved | IRC 1212(b) | n/a |
| Unrecaptured § 1250 gain, 25% maximum | IRC 1(h)(1)(E), 1(h)(6); IRS Pub. 544 | No |
| Collectibles / § 1202 gain, 28% maximum | IRC 1(h)(1)(F) | No |
| Principal-residence exclusion, $250k / $500k | IRC 121(b); IRS Pub. 523 | No |
| Reduced (partial) exclusion proration | IRC 121(c); Treas. Reg. 1.121-3(g) | n/a |
| Short/long netting and loss ordering | IRC 1222; IRS Pub. 550 | n/a |
How the federal figures were checked
Rev. Proc. 2025-32 prints, for each bracket, a cumulative tax base — the $X plus 24% of the excess over $Y figure. That base is fully determined by the breakpoints beneath it, so recomputing it from the data file and comparing against the printed figure is a genuine independent check: one mistyped breakpoint changes the base. All four individual tables reconcile to the cent, and that check runs as a test on every build.
Three rules that are easy to get wrong
- MAGI for the net investment income tax is measured BEFORE the standard deduction. Using taxable income instead silently zeroes the tax for some taxpayers.
- The § 121 reduced exclusion prorates the maximum exclusion amount, not the gain (Treas. Reg. 1.121-3(g)). Prorating the gain overstates tax for anyone with a modest gain and a qualifying early sale.
- Unrecaptured § 1250 gain is the lesser of depreciation taken and total gain, and 25% is a maximum, not a flat rate — a taxpayer in the 22% bracket pays 22% on it.
State sources
State data is split into two halves with different verification calendars. The rate structure — the rate ladder, how gains are treated, exclusions, surtaxes, local add-ons — is statutory and verifiable at any time. The thresholds — bracket bounds and standard deductions — are indexed annually and mostly published in November and December.
39 of 51 jurisdictions have a rate structure read from a tax code or revenue department. The remaining 12 are shown in the calculator as not yet available rather than left out, so no user can mistake an unverified state for a state with no tax.
Where a figure turned out to be wrong, it is logged with its date and its primary source on the corrections page, permanently. The rules those cases produced are set out in the editorial policy; the six failure modes below are the ones that have actually occurred here.
When the published rate is not the operative rate
A rate that is written down is not always the rate in force. Three cases found while verifying states for 2026 are recorded here, because each one would have produced a wrong answer that looked entirely plausible.
| State | What the secondary or codified text said | What is operative |
|---|---|---|
| Georgia | Tax Foundation carried 5.19%, the rate HB 111 set in April 2025 and the rate Georgia did begin 2026 under. | 4.99%. HB 463 reduced it effective 1 January 2026. This is the second time a Tax Foundation figure has disagreed with the operative rule, after Washington at 9% against the statutory 9.9%. |
| Kentucky | A search result asserted 4% for 2026, citing HB 152 of the 2026 session. | 3.5%. HB 152 is an introduced bill, not an enacted one. KRS 141.020(2)(f) states the rate directly. |
| Michigan | MCL 206.51 defines a trigger rather than a rate: the rate falls whenever general fund revenue growth outpaces inflation. | 4.25%. The statute alone cannot answer the question. The Treasurer and the two fiscal agencies determined for 2026 that the trigger did not fire — revenue fell 1.56% while inflation rose 2.70% — so the citation is that determination, not the code section. |
Expired and superseded text
A different failure, and a subtler one. The text below is genuinely in the statute, from the statute’s own publisher. It has simply stopped applying — because it expired, because a later act replaced it, or because the copy retrieved was an old revision. A quotation carries no date with it, and nothing about these documents announces that they are out of date.
| Provision | What it says | Why it does not apply |
|---|---|---|
| Kansas zero-rate band | A zero-rate band at the bottom of the Kansas bracket structure, presented in a research pack as current. | K.S.A. 79-32,110 applied it only “for tax years 2018 through 2023”. For 2026 the rates are 5.2% and 5.58% with no zero band. |
| Connecticut bottom two rates | 3.0% and 5.0%, read from Chapter 229 of the General Statutes on the legislature’s own site. | That copy was the 2021 revision. P.A. 23-204 cut those rates to 2.0% and 4.5% effective 2024 and the amendment is absent from it. Nothing on the document said so. The current figures were later read from a 2026 DRS form. A revision year is part of a citation. |
| Rhode Island preferential capital gains rate | A preferential rate for capital gains, said to sit in R.I. Gen. Laws 44-30-2.6. | Reported to be limited by its own terms to “tax years ending prior to January 1, 2010”. We could not reach the Rhode Island statute to confirm that, so Rhode Island stays unverified and no preferential rate is implemented either way. |
The rule this produces: a quotation is not evidence on its own. The years a provision applies to are part of the figure, and a rate quoted without them has not been verified — it has been read.
Aggregated research output
The fourth mechanism, and the one that looks most like a source. Summaries assembled from many documents arrive with citations attached, in the register of something authoritative, and are wrong in ways that are invisible without going back to the text they point at. The citation is often real; it simply does not say what the summary claims.
One such summary, checked line by line against sources already read for this site, contradicted the primary text on nine states — among them Louisiana at 6% against a flat 3%, Kansas at 2.75% against 5.2%/5.58%, Nebraska at 6.84% against a rate the statute scopes to 2026 at 4.55%, Montana as a seven-bracket state against a 4.7/5.65% ladder with separate capital gains rates, Virginia as flat against a graduated ladder, and Missouri as having no income tax at all. Two independent summaries also both gave the Arkansas top rate as 3.7%; the Department’s own 2026 formula says 3.90%.
Nothing on this site is taken from such a summary. They are useful only as a list of documents to go and read. No figure is used without verbatim text we have read ourselves — which is also why the errors above are listed here rather than sitting in the data.
When the source declares itself incomplete
The fifth shape is not an error in the document at all. The source is accurate, current and authoritative — and says on its own face that it is only part of the calculation.
Connecticut’s DRS Form CT-1040ES carries the 2026 rate ladder in Table B, under a note that it “does not include personal tax credits, the 2% phase-out or tax recapture”. The recapture in Table D reaches $3,400 for a single filer and $6,800 filing jointly. Reading the ladder and stopping would understate a large Connecticut gain by thousands of dollars, and nothing about the figure would look wrong.
So Connecticut’s rates are verified and Connecticut is still not computed. Where a document tells you what it leaves out, that sentence is part of the source.
When a repeal document names the tax but repeals its structure
The sixth shape is the most dangerous one found so far, because reading only the title of an authentic, correctly-named legislative document produces exactly the wrong answer.
Louisiana’s legislature publishes a document titled “Repeals the individual income tax”. Taken at its word, Louisiana has no income tax and no capital gains tax, and a calculator would confidently return zero. Louisiana in fact taxes a capital gain at a flat 3%.
The statute settles it in a way no summary, news report or aggregator could. La. R.S. 47:32 has two subsections. Subsection B — the graduated bracket schedule — carries “Repealed by Acts 2024, 3rd Ex. Sess., No. 11, s4, eff. Dec. 4, 2024”. Subsection A, which imposes the tax, stands unrepealed and reads: “The tax to be assessed, levied, collected, and paid upon the taxable income of an individual shall be computed at the rate of three percent on net income.” What was repealed was the structure — the ladder of brackets — and it was replaced by a single rate in the same act.
The same act reached further than the rate. Louisiana’s statutory subtractions list at R.S. 47:293(9)(a) now contains no occurrence of “capital gain”, “net capital” or “equity interest”: the business-interest exclusion formerly at (9)(a)(xvii), and item (ix), both read “Repealed by Acts 2024, 3rd Ex. Sess., No. 11”. So the act that flattened the rate also stripped the gains exclusion, and a reader who took the title at face value would have been wrong twice over.
The rule this produced: a repeal is read at the subsection it names, never at the title. A document can be genuine, current and accurately titled and still describe something narrower than its name suggests — and the amendment history, which here ends at Act 11 of 2024, is what confirms nothing later moved it.
States where a determination decides the rate
In these states the statute defines a trigger rather than a rate, so reading the code tells you the mechanism and nothing about the answer. Each needs a third kind of source: the department’s own determination for the year.
| State | Trigger | Status for 2026 |
|---|---|---|
| Michigan | MCL 206.51(1)(c) — rate falls when general fund revenue growth outpaces inflation, determined jointly by the Treasurer and the two fiscal agencies. | Determined. It did not fire: revenue fell 1.56% while inflation rose 2.70%, so the rate stays 4.25%. |
| Oklahoma | 62 O.S. 34.103 — rates fall a further 0.25% across all brackets when revenue conditions are met, evaluated each December and certified in February by the State Board of Equalization. | 2026 brackets published by the Tax Commission under HB 2764. The certification must be read again before rolling forward to 2027. |
| North Carolina | S.L. 2023-134 — 3.99% for taxable years after 2025 is stated outright; further cuts from 2027 are contingent on General Fund revenue triggers. | 2026 stated by the Department. NCDOR’s own rate schedule reads “For Taxable Years after 2025, the North Carolina individual income tax rate is 3.99%”, which replaces the survey figure the entry rested on. ncleg.gov still serves a bot challenge, so the statute itself is unread; the 2027 trigger determination must be read before rolling forward. |
Washington is a special case: RCW 82.87.040 imposes a 7% excise on all Washington capital gains plus an additional 2.90% on the portion above $1,000,000 — a 9.9% marginal rate. Real property is exempt outright and only long-term gains are in scope. Its standard deduction is indexed under RCW 82.87.150 and published separately from any rate table; the 2026 figure was not published when this was last checked, so the 2025 amount is used under a mandatory warning.
What is out of scope
This list is deliberately explicit. If your situation involves any of these, the number this calculator gives you is incomplete — not merely approximate.
| Not modeled | Why it matters |
|---|---|
| Alternative minimum tax (IRC 55–59) | Can change the answer for high earners with large gains. |
| Qualified small business stock (IRC 1202) | Can exclude the entire gain. Only the 28% residual rate is modeled. |
| Collectibles | The federal 28% rate is modeled, and Massachusetts’ 12%-with-50%-deduction. No other state’s collectibles treatment is. |
| Wash sales (IRC 1091) | Disallowed losses are not detected. The engine takes your net gain as given. |
| Carryover in a low-income year (IRC 1212(b)(2)) | The full capital loss offset is treated as used even when taxable income was too low for all of it to reduce tax, so the carryover is understated for that year. |
| Like-kind exchanges (IRC 1031) | Deferral is not modeled. |
| Installment sales (IRC 453) | Spreading gain across years is not modeled. |
| Multiple lots and specific identification | One gain figure per sale. No per-lot basis tracking or FIFO/HIFO selection. |
| Opportunity zones (IRC 1400Z-2) | Deferral and step-up are not modeled. |
| Kiddie tax (IRC 1(g)) | A child’s unearned income above $1,350 is not handled. |
| Estates and trusts | Verified 2026 figures are recorded in the codebase but unreachable: only the four individual filing statuses are supported. |
| Net investment income beyond gains and qualified dividends | Interest, rents, royalties and passive business income are NII under IRC 1411(c). The engine derives NII from capital gains and qualified dividends only, and warns when it does. |
| State additions, subtractions and deductions | State tax is computed by stacking the gain on FEDERAL ordinary taxable income. Each state’s own base differs, and the engine warns when it approximates. |
| Local income taxes | New York City and Yonkers, Maryland counties, Ohio municipalities, Indiana counties, Michigan cities. Flagged in a warning, never silently omitted. |
| Non-residents and part-year residents | Every state figure assumes full-year residency. |
Corrections
If you find a figure that is wrong, please tell us and cite the source. Corrections to tax figures are treated as urgent. Source data for tax year 2026 is Rev. Proc. 2025-32 (I.R.B. 2025-45).