About the ACA cliff planner
Bracketsight is a planning tool for the households the 2026 ACA subsidy cliff actually hits: self-employed people, early retirees, and gig workers whose income sits near 400% of the federal poverty line. Other calculators tell you the number. Bracketsight tells you the move — how far you are from the edge, what crossing costs, and which legal levers pull you back, ranked by dollars recovered per dollar committed.
What does this engine decide that a premium calculator does not?
A calculator answers “what is my credit at this income.” This engine answers “how far can my income move before the credit disappears, and which legal moves pull it back.” For 2026 the premium tax credit ends at a hard edge rather than tapering: Rev. Proc. 2025-25 §3.01 publishes an applicable-percentage table whose top row reads “at least 300% but not more than 400%,” and there is no row above it. A table that terminates is a table with a cliff, so the useful output is a distance and a ranked set of levers, not a monthly figure.
Where exactly does the cliff fall?
At exactly 4.0 × the federal poverty line for your family size, and not a dollar higher. Form 8962 Worksheet 2 asks first whether household income is more than 4.0 × FPL; if it is, you enter 401 on line 5 and line 6 tells you that you are not eligible. Truncation to a whole percent is reached only in the “no” branch, so 400.9% is 401, not 400. Applying the 2025 HHS poverty guidelines that govern 2026 coverage — $15,650 for one person in the contiguous 48 states and DC, plus $5,500 for each additional person, at 90 Fed. Reg. 5917 — the last eligible income is $62,600 for a single filer and $128,600 for a family of four. Both are engine output pinned by tests, not typed into a page.
Which rules does the engine encode?
The IRC §36B credit and its six applicable-percentage bands; the cost-sharing reduction ledges of 94%, 87% and 73% actuarial value from 42 U.S.C. §18071(c)(2); Medicaid expansion status for all 51 jurisdictions; the 2026 retirement and HSA contribution limits; and the reconciliation rule. It also encodes a repeal: Pub. L. 119-21 §71305 struck IRC §36B(f)(2)(B) for tax years beginning after 31 December 2025, so for 2026 there is no cap on repaying excess advance credit at any income. A household at 250% of the poverty line repays every excess dollar. Each of those figures traces to a primary source on /sources, with its verification date, and each formula is set out on methodology.
Which rules does it deliberately leave out?
- The age 60–63 catch-up. IRC §414(v)(2)(E)(i) allows $11,250 for 2026 instead of the $8,000 age-50 catch-up, confirmed in IRS Notice 2025-67. It is not modelled, which understates the room available to exactly the pre-Medicare enrollees most likely to be near the cliff.
- The HDHP eligibility gate. The engine offers the HSA lever without checking the 2026 high-deductible plan definition (minimum deductible $1,700 self-only / $3,400 family; out-of-pocket maximum $8,500 / $17,000). Confirm your plan qualifies before you contribute.
- Income timing. It is listed advisory-only and never carries a computed recommendation, because shifting income across tax years changes more than a subsidy.
What is real data today, and what is placeholder?
Benchmark premiums are the placeholder. The six counties in the sample file — Travis TX, Los Angeles CA, Cook IL, Maricopa AZ, Miami-Dade FL and Denver CO — carry invented age-21 base premiums, written so the engine and interface could be built and tested. There is nothing in them to verify; they must be replaced from the CMS Marketplace public use files for plan year 2026. Any dollar figure that depends on a premium is therefore illustrative, and the sample-data labelling stays until real premiums land.
The age-rating curve beside them is real and was checked exactly: all 65 factors match Appendix I of CMS's 2016 age-curve guidance with zero mismatches, and 45 C.F.R. §147.102 caps age variation at 3:1. That curve is only a federal default: CMS's state-variation table, stamped December 2021, shows New York and Vermont with no age rating at all. The 2026 Form 8962 is also unpublished until roughly January 2027, so the cliff convention above rests on the 2025 and 2020 editions, which agree word for word.
How should you use this, and how should you not?
Use it to see how far your household sits from the edge and to compare levers while the tax year is still open — that comparison is the product. Do not use it as a filing position or as a premium quote: confirm your benchmark premium with the marketplace and confirm any contribution or deduction with a tax professional before you act. No enrolled agent or CPA has reviewed this section. That review is a launch gate and it has not happened yet.
The promises
- Deterministic math. Every figure comes from an open, tested formula — no AI computes anything. See methodology.
- Cited rules. Every threshold traces to a primary source on /sources, with its verification date.
- Your data stays yours. Everything runs in your browser; nothing is uploaded, stored server-side, or tied to an account.
- No insurance sales. Bracketsight never recommends a specific plan and earns nothing from enrollment.
Status
This is a pre-launch build. Benchmark premiums are sample data, a handful of 2026 indexed figures await verification against their primary sources, and the named EA/CPA reviewer will appear on the editorial policy page before launch. The changelog tracks every rules change, including the pending enhanced-credit restoration bill.