Skip to the content
LoanXray
Ver en español

Compare

The rate is not the price

At the same borrower profile — same county, same program, same LTV, same DTI, same loan size — closing costs run 3.71 times higher at the 90th percentile than at the 10th. That is the regulator’s own file, not our opinion. Mortgage insurance is opaque by design, and nobody tells you how long each piece of the payment lasts. So we compare total cost over the years you say you will stay.

Where that dispersion comes from

HMDA 2025, Harris County TX, conventional 30-year first liens on a primary residence, one unit, LTV 79–81, DTI 36–43, loan $300k–$500k. 531 loans, 523 with a cost reported. Tenth percentile $3,628, median $6,812, ninetieth $13,471. Borrowers an underwriter could not tell apart, paying $9,842 apart.

We never ask for your Social Security number

No SSN

Not out of politeness. Hand a lender your name, income, address, property value, loan amount and SSN and you have made an “application” under 12 CFR 1026.2(a)(3)(ii): a hard credit pull fires, a Loan Estimate obligation starts, and your file becomes a trigger lead. We give you the first five and keep the sixth, which is why you can compare here without your phone starting to ring.

Loaded with a real case

María, 40, Fort Worth. Household of four, two earners, FICO 691, $43,000 saved against the $434,100 median home. Forty is the median age of a first-time buyer in the United States and an all-time high; 691 is the band where lenders disagree most. Change anything below and everything to the right recalculates.

Twenty questions

Each one says what it decides. Change anything and the panel on the right recalculates before you look up.

What you are doing
1
How you will use the home

The first question, because it prunes the tree. Anything but a primary residence kills FHA, USDA and VA in one stroke, and on conventional it adds +1.125 to +4.125 points.

2
Purchase or refinance

Purchase, rate-and-term and cash-out are three different pricing grids with three different seasoning windows. Cash-out above 80% LTV simply does not exist.

No US comparison site asks this
This only bites on an FHA rate-and-term refinance, so it is switched off while you are buying. Switch question 2 to “Refinancing, no cash out” and it comes alive — and moves the maximum LTV from 97.75% to 85%.

On an FHA rate-and-term it moves the maximum LTV from 97.75% to 85%. It is one of the two fields no US comparison site asks.

The house
3
Tarrant County, TXFIPS 48439
Conforming limit here · 1 unit$832,750

Baseline county — 3,075 of the 3,235 are · 2026

The conforming limit, the FHA limit, USDA rural eligibility, property tax, insurance, transfer tax and title all come from here. The same $900,000 loan carries a full point of adjustment in one county and nothing in the one next door.

4
$
$

On a purchase, LTV goes on the lower of price and appraisal. Leave it empty and we use the price.

LTV is measured against the LOWER of price and appraisal. A low appraisal raises your LTV and can move you a whole pricing band.

5
$
3.50%LTV 96.50%HUD reads it at two decimals; Fannie rounds it up to 97%. Two regulators, two roundings, and the pricing band can differ.

LTV, the mortgage insurance factor, and FHA’s minimum required investment. Four hundred dollars either way can move FHA’s insurance from eleven years to thirty.

6
Units

Limits are published per unit count. Three and four units add a self-sufficiency test on FHA, mandatory reserves, and their own price adjustment.

7

An attached condo adds 0.750 points and a manufactured home 0.500. The condo charge does NOT apply to a detached condo or to a co-op — a footnote worth thousands.

The loan
8
Term and rate type

The FHA insurance table splits at fifteen years, an ARM with a short fixed period pays 20–30% more private mortgage insurance, and the qualifying rule changes with the fixed period. A 7/6 buys more house than a 5/6 at the same starting rate.

optional
pts

58.7% of 2023 purchases paid points, median $3,000. Comparing a rate with two points against a rate with none is the exact lie this product exists to undo.

optional
Mortgage insurance coverage

Minimum coverage cuts the premium and triggers an extra price adjustment. Which one wins depends on your horizon, and it is solved below with numbers, not with a rule of thumb.

optional
$

It counts against your DTI in full, and it does not go through escrow: you pay it separately.

Credit
9
Score model

VantageScore 4.0 bands sit twenty points above Classic FICO. The same person lands in a different pricing row depending on which model the lender pulls, and today nobody shows it.

10
Borrower 1
Borrower 2

Qualifying score

Conventional: average of the borrowers’ medians691
FHA: the lowest of the borrowers691

The rule differs by program: conventional averages the borrowers’ medians, FHA takes the LOWEST. Two people at 640 and 700 qualify at 670 on conventional and at 640 on FHA.

Money coming in and going out
11
Borrower 1
$
Borrower 2
$

Total monthly income from the borrowers$8,025

Seeded from NAR’s median income for the Hispanic homebuyer ($96,300), not from María’s own income, which the research does not give. It is an anchor for a segment, not her paycheck.

DTI, VA residual income and the USDA cap. Income type is not decoration: W-2 documents in one page and self-employment takes two years of returns.

12
$
At zero this DTI is a floor, not a fact. The research file does not give María’s debts, so we did not invent them — put yours in.

Car, cards, student loans, child support. USDA closes the list: childcare and transportation do NOT count as debt there.

13

The VA residual income table counts everyone who lives there, including a spouse who is not on the note and not on title.

14
$a year

At or below 100% of AMI with a first-time buyer, every price adjustment goes to zero. That is up to 1.125 points on this file. Household income and county are pricing fields, not courtesy fields.

USDA is the only program that asks for it, and it disqualifies files that pass on FHA. Against your area median income it can also waive every price adjustment down to zero.

What opens or closes a program
15
First-time buyer

It opens conventional at 97% LTV, it can waive the adjustments entirely at or below 100% of area median income (120% in high-cost areas), and it is the gate to down payment assistance.

16
No US comparison site asks this
Prior use of the benefit, the disability exemption and access to a nearby base only matter once there is qualifying service. Pick anything but “No military service” and the three questions appear — the first of them is worth $4,600 on a $400,000 loan.

Funding fee 2.15%, 3.30% or zero. Prior use of the VA benefit with less than 5% down is 115 basis points more — $4,600 on a $400,000 loan — and no US comparison site asks it.

17

FHA removed non-permanent residents outright on 25 May 2025. Fannie Mae does accept an ITIN; what actually blocks the loan is the legal-presence representation the lender signs.

18
$
Documented compensating factors (FHA manual underwriting)

None gives 31/43. One gives 37/47. Two give 40/50. Below a 580 score the 31/43 ceiling is hard and no factor lifts it.

Compensating factors on FHA manual underwriting, the 45% manual DTI ceiling on conventional, and mandatory reserves on three and four units.

19

CLTV, and 0.625 to 1.875 points — but only when CLTV exceeds LTV, never on the undrawn part of a HELOC, and never on a Community Seconds.

How long you stay
20
7 years

The axis of this entire page. It decides whether buying points pays back, whether single premium beats monthly mortgage insurance, and which program is actually cheapest.

Escrow assumptions

Aggregate accounting projects the year and funds the lowest point. Move the tax month and watch the initial escrow deposit move with it — that is why block G differs between two offers on the same house.

optional
$a year

Your own bill, at this county’s effective rate, would be $5,405 a year — 1.245% of the price. The gap between the two is the year-two shortage, and it is the only part of the escrow surprise we can compute without inventing a growth rate.

It is on the seller’s bill. The initial escrow is often set from it while your own bill arrives reassessed at what you paid. That gap is the year-two shortage.

Twenty numbered questions, plus 7 optional ones. Nothing here is stored, nothing is sent anywhere, and there is no field for a Social Security number.

Every program, priced side by side

Same borrower, same house, same day. What separates them is not the rate — it is the insurance, how long it lasts, and what you hand over at the table.

Order by
  • FHALTV 96.50%

    Loan $426,237

    Unresolved condition
    Payment$3,598.14today
    Rate6.766%no published observation for your exact segment
    Price adjustmentsnone apply
    Mortgage insurance
    $191.08never ends$15,454
    Total$320,8787 years · over your horizonAPR 7.506% · Cash at the table $22,083

    The qualifying debt-to-income ratio is 44.8% against this programme’s 43% manual ceiling, and the housing ratio is 44.8% against 31%. It can still clear through the automated underwriting system, but that is not automatic, and it is the first thing an underwriter looks at. HUD Handbook 4000.1 II.A.5.d.viii

  • ConventionalLTV 96.50%

    Loan $418,907

    Payment$3,736.45today
    Rate6.766%no published observation for your exact segment
    Price adjustments1.125 pts$4,713
    Mortgage insurance
    $377.02ends month 145 · October 2038$31,669
    Total$340,4527 years · over your horizonAPR 7.749% · Cash at the table $24,730
  • VANot eligible

    A VA loan needs a Certificate of Eligibility, and that needs qualifying service: veteran, active duty, Guard or Reserve with enough time in, or an eligible surviving spouse. Nobody on this file has it.

    Rule: 38 U.S.C. 3702; 38 CFR 36.4302

  • USDANot eligible

    The qualifying debt-to-income ratio is above this programme’s ceiling, and this ceiling is hard: no compensating factor lifts it and no automated underwriting system overrides it. Your qualifying ratio is 43.7% against a ceiling of 41%, and the housing ratio is 43.7% against 29%.

    Rule: 7 CFR 3555.151(h)(1)(i)

  • JumboNot eligible

    You are below the conforming limit, so you do not need a jumbo — and the conforming loan has public rules and a comparable price, which a jumbo does not.

    Rule: FHFA CLL FAQs 2026, pregunta 1

Ordering rule

Ordered by total cost — every charge included — measured over the tenure you declared, ascending. HUD blessed this exact presentation as neutral in 1996 (CLO Policy Statement 1996-1, 61 FR 29255, 29258). Ties break by program name, alphabetically. Ineligible programs go last, never hidden.

Programs you do not qualify for are not hidden and not moved to a second page. They sit at the bottom, dimmed, each with the rule that excludes it quoted in full.

Where the order flips

The same five programmes, priced at every horizon from one year to thirty. Nothing else on this page moves the ranking as much as this slider does.

$0$322.8K$645.7K$968.5K$1.3M1510152025307 yrConventionalFHA
  • FHA is cheaper than Conventional up to year 25, and dearer by year 30. Somewhere in between, the order of the table flips — and that is the whole argument for asking how long you plan to stay before showing anyone a price.

The chosen programme

What you would actually be signing

Pick any row above and this whole section rebuilds around it. Nothing is hidden behind a paywall, a phone call or a form.

Taken apart

FHA

Every line of the payment, how long it lasts, and what the engine assumed to get there.

Cash at the table

Down payment plus what you pay for, plus the initial escrow deposit. Anything financed into the loan is not here — it shows up in the balance instead.

Total out of pocket

$22,083

7 year horizon

Total over the horizon

$320,878

Down payment$15,194
Closing costs you pay$4,037
Initial escrow deposit$2,853
Financed into the loan instead$7,331

FHA’s upfront premium. It does not come out of your pocket — it is added to the balance on day one, and it is excluded from the LTV and from the base of the annual premium by regulation.

Where the total goes

Principal and interest$232,60572%
Mortgage insurance$15,4545%
Property tax$37,83212%
Homeowners insurance$15,7575%
Closing costs you pay$4,0371%
Down payment$15,1945%
Total$320,878

Closing costs are anchored to what lenders actually charged, from the regulator’s own file. Where a component is fixed by public rule — a promulgated title rate, FHA’s upfront premium — we pull it out and price it exactly, and what remains is the lender’s own charge.

FHA’s upfront premium sits inside HMDA’s total loan costs when the borrower pays it, and the engine already charges it separately. We subtract it here ($7,331) so it is not counted twice.

Texas is one of only two states where title insurance carries a rate promulgated by the regulator, so this line is exact to the cent rather than estimated. The appraisal is anchored to the VA fee schedule, the only public, primary, county-level source of what an appraisal costs in this country.

What this measures: 3 blocks of the Loan Estimate (A, B and C) and nothing else. Transfer taxes, prepaids, the initial escrow deposit and the owner’s title policy are all outside it.

The full payment, component by component

Nobody shows the third column. A payment made of pieces that die on different dates is not one number: it is five numbers with five different clocks.

  • Principal and interest$2,769.10the whole loan$232,605
  • Mortgage insurance premium (MIP)$191.08the whole loan$15,454

    Charged on the average balance of each year, so this instalment falls every year. Year one is a ceiling.

  • Property tax$450.38the whole loan$37,832

    The state effective rate on the existing housing stock, not your own bill — that is set by the county and the school district. In the median home it is bigger than the mortgage insurance.

  • Homeowners insurance$187.58the whole loan$15,757

    An interpolated median of what households WITH a mortgage report paying, not a quote. Flood insurance is a separate policy and is not in here at all.

  • Full payment today$3,598.1484 months$301,647
Debt-to-income, as you would pay it44.8%
Debt-to-income the underwriter looks at44.8%

Above this programme’s 43% ceiling.

Mortgage insurance, with the date it dies

You will pay FHA mortgage insurance for all 360 months of the loan — it never cancels — and $15,454 of that falls inside the years you plan to stay.

This one never dies. The only way off it is to refinance out of the program entirely.

Premium today$191.08
Months you pay it360 months
Total over your 7 years$15,454
Total over the whole loan$45,589

Annual premium of 55 basis points, falling

Year 1$191.08
Year 5$181.73
Year 10$165.89
Year 15$143.69
Year 20$112.58
Year 30$7.90

FHA charges on the average balance of each year, so the instalment falls. Conventional PMI is flat on the original amount. They are two different engines, and every calculator that draws them alike is falsifying the one comparison that matters.

Standard coverage against minimum coverage

Minimum coverage cuts the monthly premium and fires an extra price adjustment that is charged even when every other adjustment is waived. There is no rule of thumb here. There is arithmetic.

The programme you have selected carries no private mortgage insurance, so this choice does not arise on it. What follows is the Conventional row, which is the one on this file where the choice is real.
Standard coverage
Monthly premium$377.02

1.08% a year of the original loan

Extra price adjustment, once—
Cost over your horizon$31,669
Minimum coverageCheaper at your horizon
Monthly premium$251.34

0.72% a year of the original loan

Extra price adjustment, once$7,331
Cost over your horizon$28,444

Over your 7 years the two options are $3,226 apart. Change the horizon slider and the answer can flip, because one side is a monthly premium and the other is a one-time charge.

Break-even points, measured against your horizon

Discount points

You are not buying points, so there is nothing to break even on.

Single premium against monthly mortgage insurance

Shown on the Conventional row, the only one on this file that carries private mortgage insurance.

Single premium, paid once$15,416
Monthly premium over your horizon$31,669
Breaks even at month41 · February 2030

Paid once, the single premium wins at your horizon by $16,254.

The single premium is non-refundable. Sell or refinance before the break-even and you lose the whole thing.

The year-two escrow jump, with both increases separated

This is one of the four confusions the CFPB documents in its own 2025 complaint report, and it is the only one you can see coming before you sign, because it is arithmetic.

Initial escrow deposit at closing

$2,853

Cushion $1,276

The two together, hitting the same month

+$10.58

per month, from month 14

Shortage found at the analysis$18
Shortage spread over twelve months+$1.46
Higher base deposit, because the annual bill grew+$9.11
The two together, hitting the same month+$10.58
Principal, interest and escrow, year one$3,407.06
Principal, interest and escrow, year two$3,417.64

At the end of the computation year the servicer re-analyses. If taxes or insurance went up, two things happen at once: the shortfall gets spread over twelve months or more, and the base deposit rises because next year’s bill is bigger. They land on the same payment.

The commonly told version of this story is $180 a month. We cannot source it. With the only growth rate we can cite — insurance at 2.4% real, the floor of the NAIC range — and no primary source for property tax growth, the engine gives what you see. The Texas case that produces the big number is reassessment at the purchase price, and no primary source quantifies it. Give us the seller’s current tax bill above and the shortage becomes exact.

APR, and what it leaves out

Two offers with the same APR can cost thousands of dollars apart at the table, because the legal APR excludes exactly the charges that vary most between lenders.

APR

7.506%

Note rate

6.766%

Amount financed$418,036
Prepaid finance charge$8,201
Total interest over the full term$570,639

Excluded from the finance charge by 12 CFR 1026.4(c)(7) when bona fide and reasonable: title examination, title insurance, survey, document preparation, appraisal and credit report. And by (c)(5), seller’s points.

Legal accuracy tolerance: 0.125 percentage points (12 CFR 1026.22(a))

Rate band for this profile

6.766%6.766%6.931%
OBMMIC30YF6.766%
OBMMIC30YFLVGT80FLT6806.931%

Your segment is OBMMIC30YFLVGT80FB680A699, and it has no published observation. Rather than seed the nearest lookalike series — which would move your rate for narrative convenience — we anchor to OBMMIC30YF and show you the gap: 16.5 basis points between the two observations that do exist. That gap is the observed price of risk, not an estimate.

Source: OBMMI · 2026-09-08 · Checked Sep 10, 2026

How much house each program’s rulebook allows

Solved by bisection, because the mortgage insurance depends on the LTV which depends on the price. Not a generic 43%: each ceiling comes from its own regulation.

  • Conventional$467,000

    DTI ceiling 50%Fannie Mae B3-6-02. Manual underwriting starts at 36% and reaches 45% with the score and reserves of the Eligibility Matrix, but a DU casefile reaches 50% — the highest ceiling in the whole catalogue. This page assumes a DU Approve/Eligible.

  • Jumbo$467,000

    DTI ceiling 50%There is no published rule. By the FHFA’s own definition a jumbo is the loan the Enterprises cannot buy, so there is no Selling Guide, no common AUS and no price matrix — each investor writes its own. The figure shown is the agency ceiling as a declared reference, not a verified limit.

  • VA$416,000

    DTI ceiling 41%38 CFR 36.4340(c). The 41% is a documentation trigger, not a cap. What actually decides a VA file is residual income, and if residual exceeds the guideline by 20% or more, no supervisor justification and no second review are required at all.

  • FHA$292,000

    DTI ceiling 43%HUD Handbook 4000.1 II.A.5.d.viii. Manual underwriting is 31/43 with no compensating factor, 37/47 with one, 40/50 with two, and 40/40 with no discretionary debt. Between a 500 and a 579 score the 31/43 pair is a hard ceiling that no factor lifts. With an Accept from the TOTAL Scorecard the AUS governs instead of this matrix.

  • USDA$279,000

    DTI ceiling 41%7 CFR 3555.151(h)(1)(i). 29/41, the strictest pair in the catalogue, and it is hard. Note that USDA closes the debt list: childcare and transportation do not count.

The order of generosity inverts the folk wisdom that “FHA forgives more”: conventional through DU reaches 50%, FHA manual reaches 50% only with two documented compensating factors, VA’s 41% is a documentation trigger rather than a cap, and USDA’s 29/41 is the hardest pair in the catalogue.

What the engine assumed, and how sure it is

Printed as the engine returned them. Anything not marked verified is a figure you should ask a lender to confirm before you rely on it.

  • SIN_AVALUOVerified

    No appraisal was supplied, so LTV runs on the price. On a purchase the value is the LOWER of price and appraisal, so a low appraisal would raise the LTV and could move the pricing band.

    No se recibió avalúo: el LTV se calcula sobre el precio. En compra el valor es el MENOR entre precio y avalúo, así que un avalúo bajo subiría el LTV y podría mover la banda.

  • BANDA_SCORE_MAYOR_O_IGUALVerified

    Score bands are applied as “greater than or equal to”. Fannie prints “>740” and Freddie “>=740”, so a score of exactly 740 falls in no Fannie row at all — a literal hole in the document. We follow Freddie’s reading and say so.

    Las bandas de score se aplican como «>=». Fannie imprime «>740» y Freddie «>=740»; un score de exactamente 740 no cae en ninguna fila de Fannie. Se sigue la lectura de Freddie.

  • PMI_PROXY_ENACTVerified

    Private mortgage insurance is modelled on Enact’s public rate card, updated 17 July 2025. MGIC, Radian, Essent, National MI and Arch quote only through a proprietary engine inside the lender’s pricing system, and the borrower does not choose the insurer. This is a declared proxy.

    Prima de MI modelada con la rate card pública de Enact (act. 17-jul-2025). MGIC, Radian, Essent, National MI y Arch cotizan solo por motor propietario dentro del LOS/PPE del prestamista, y el prestatario no elige aseguradora. Esto es un proxy declarado.

  • HPA_SOBRE_VALOR_ORIGINALVerified

    All three cancellation dates run against the ORIGINAL value — the lower of contract price and appraisal. Appreciation does not bring cancellation forward under the Homeowners Protection Act. Using current value means going through the GSE rules, which are different and were not verified in this project.

    Las tres fechas van sobre el valor ORIGINAL (menor entre precio y avalúo). La apreciación de la vivienda no adelanta la cancelación bajo HPA; para eso están las reglas del GSE, que son distintas y no están verificadas en este proyecto.

  • HPA_BUEN_HISTORIALVerified

    Cancellation on request requires “good payment history”, and that is two separate windows, not one: nothing 60 or more days late in months 13 through 24 back, and nothing 30 or more days late in the last 12. The intuitive version denies cancellations the law grants.

    La cancelación a petición exige «good payment history», que son DOS ventanas distintas: sin mora de 60+ días en los meses 13 a 24 hacia atrás, y sin mora de 30+ días en los últimos 12. La versión intuitiva («60+ en los últimos 12») deniega cancelaciones que la ley concede.

  • APR_EXCLUSIONESVerified

    Charges excluded from the finance charge under 1026.4(c)(7) when bona fide — title, appraisal, credit report, survey, document preparation — and under 1026.4(c)(5) for seller’s points. The legal APR leaves out precisely what varies most between lenders.

    Excluidos del finance charge 2 cargo(s) por 1026.4(c)(7) —título, tasación, credit report, levantamiento, preparación de documentos, si son bona fide— y por 1026.4(c)(5) —seller points—. El APR legal deja fuera lo más variable entre prestamistas.

  • APR_TOLERANCIAVerified

    The law treats an APR as accurate if it is within 0.125 percentage points. Differences smaller than that are not informative, and a ranking must not pretend to a precision the regulator does not require.

    La ley considera exacto un APR que no se desvíe más de 0.125 puntos porcentuales (1/8). Diferencias menores NO son informativas y el ranking no debe fingir una precisión que la ley no exige.

  • LIMITE_FHA_SIN_CONDADOVerified

    FHA’s national floor was applied, which is what governs in most counties. The real FHA limit is per county and is NOT in the FHFA file: it comes from HUD’s CHUMS files. In a high-cost county it runs far higher, so this cut-off may exclude a loan that would be eligible there.

    Se usó el PISO nacional de FHA ($541,287 en 1 unidad), que es el que rige en la mayoría de condados. El límite real es por condado y NO está en el CSV de la FHFA: sale de los ficheros CHUMS de HUD (entp.hud.gov). En un condado de alto costo puede llegar a $1,249,125, así que este corte puede excluir un préstamo que allí sí sería elegible.

  • LLPA_NO_APLICAVerified

    Loan-level price adjustments belong to Fannie and Freddie, so they do not apply to FHA, VA or USDA, which each have their own premium and fee structure. And a jumbo carries none either: by definition the Enterprises cannot buy it, so neither matrix governs it. The risk adjustment does exist, each investor sets it, and it is not public — this engine does not estimate it.

    Los LLPA / Credit Fees son de Fannie y Freddie: no aplican a FHA, VA ni USDA, que tienen su propia estructura de primas y comisiones.

  • FHA_UFMIP_EXCLUIDOVerified

    FHA’s financed upfront premium is excluded from the LTV and from the base of the annual premium, per 24 CFR 203.284 — “excluding the portion of the remaining balance attributable to the premium”. Almost every internet calculator adds it to both, and the error gets paid twice.

    El UFMIP financiado se excluye del LTV y de la base del MIP anual, por 24 CFR 203.284 («excluding the portion of the remaining balance attributable to the premium»).

  • FHA_MIP_DECRECIENTEVerified

    FHA’s annual premium is charged on the average outstanding balance of the year, so the payment FALLS every year. Year one is a ceiling, not a 30-year instalment. Conventional PMI, by contrast, is flat. Two different engines, and showing them alike falsifies exactly the comparison this page exists to make.

    MIP anual de 55 bps sobre el saldo medio del año: la cuota BAJA cada año. La cifra del año 1 no es una cuota a 30 años.

  • FHA_CONVENCION_PROMEDIONot verified

    HUD says “average outstanding balance” but does not publish the exact averaging convention in any source we could read, so the engine averages the twelve monthly balances of each year and declares it.

    Promediado con la convención «MEDIA_DE_SALDOS_MENSUALES». HUD dice «saldo pendiente medio» pero no publica la convención exacta de promediado en fuente que se haya podido leer.

  • FHA_UMBRAL_EN_DISPUTAIn dispute

    The threshold that splits FHA’s annual premium table is genuinely in dispute, and both sides are verified in primary sources: the current Appendix 1.0 prints $726,200 while Mortgagee Letter 2023-05 redefines it as the national conforming loan limit. The engine applies what the current table prints and refuses to present the gap between the two as a product finding until HUD clears it up.

    Umbral de tarifa aplicado: $726,200 (interpretación APPENDIX_1_0). Appendix 1.0 (rev. 20-mar-2023, vigente en Update 18 de 12-ago-2026) imprime $726,200; la ML 2023-05 lo redefine como «the national conforming loan limit» ($832,750 en 2026). Ambas verificadas en fuente primaria y mutuamente incompatibles.

  • FHA_MIP_NO_MUEREVerified

    At this LTV, FHA’s mortgage insurance runs for the full term. It never cancels. The only way off it is to refinance out of FHA entirely. At 90% LTV or below — a 10% down payment — it would drop to eleven years.

    Con LTV 96.5% el MIP dura el plazo COMPLETO (360 meses): no se cancela nunca. Para quitárselo hay que refinanciar. Solo con LTV <=90% (10% de enganche) bajaría a 11 años.

  • VA_RESIDUAL_DEDUCCIONES_NO_MODELADASNot verified

    VA residual income is computed on NET income and also deducts federal and state taxes, social security and an estimate of home maintenance and utilities. Those deductions were not verified in this project and are not modelled, so the residual shown is a ceiling.

    El residual del VA se calcula sobre ingreso NETO y descuenta además impuestos federales y estatales, seguridad social y una estimación de mantenimiento y servicios de la vivienda. Esas deducciones no están verificadas en este proyecto y no se modelan: el residual mostrado es un TECHO. Pásalas en `deduccionesAdicionales` cuando se verifiquen.

  • USDA_TARIFA_NO_VERIFICADANot verified

    The USDA fee for this fiscal year comes from a Rural Development bulletin and from lender bulletins, not from the Procedure Notice, and has to be confirmed before any demonstration. The statutory caps — 3.5% upfront and 0.5% annual — are verified in 7 CFR 3555.107.

    Tarifas del FY2026 tomadas de boletines de prestamistas, no de la Procedure Notice de USDA. No deben salir en una demo sin confirmar contra USDA RD. Los topes estatutarios (3.5% upfront / 0.5% anual) sí están verificados en 7 CFR 3555.107.

  • AMI_SIN_RESOLVERVerified

    You did not tell us where your income sits against the area median. At or below 100% of AMI — 120% in a high-cost area — a first-time buyer has every price adjustment waived. On this file that is up to 1.125 points, so the conventional row may be priced higher than it should be. Household income and county are pricing fields, not courtesy fields.

Verified: Read in the primary source, with its effective date. · In dispute: Two primary sources say different things and both were checked. We show which one we applied. · Not verified: We could not reach a primary source. Treat it as a working figure, not a fact. · Market convention: What the industry does, not what any regulator published.

Warnings this file triggered

The engine records its warnings in the language it is written in, and we print them verbatim rather than paraphrase — a paraphrased warning is a weaker warning, and this trail has to be auditable. Every figure they mention appears in English above.

  • Sin observación para el segmento OBMMIC30YFLVGT80FB680A699. Se usa el OBMMI general de 30 años conforming, que promedia todos los perfiles: para un LTV alto con score bajo SUBESTIMA la tasa (la diferencia observada entre el general y el cruce LTV>80/FICO<680 fue de 16.5 pb).
  • No hay diferencial verificado de FHA frente a convencional. OBMMI publica una serie propia (p. ej. OBMMIFHA30YF) y esa es la vía correcta: observar el diferencial, no postularlo. Se cotiza con el ancla convencional, sin restar nada inventado.
  • El seguro hipotecario de FHA lo va a pagar los 30 años completos. Para quitárselo hay que refinanciar. Con 10% de enganche (LTV <=90%) bajaría a 11 años.
  • DTI de 45%, por encima del 43% de FHA. HUD 4000.1 II.A.5.d.viii, suscripción manual con 0 factor(es) compensatorio(s). Con Accept del TOTAL Scorecard manda el AUS, no esta matriz.
  • Ratio de vivienda de 45%, por encima del 31% de FHA.
  • Primer uso, enganche 3.5%.
  • No hay diferencial verificado de VA frente a convencional. OBMMI publica una serie propia (p. ej. OBMMIVA30YF) y esa es la vía correcta: observar el diferencial, no postularlo. Se cotiza con el ancla convencional, sin restar nada inventado.
  • DTI de 43%, por encima del 41% de VA. 38 CFR 36.4340(c): 41% es DISPARADOR DE DOCUMENTACIÓN, no tope. Lo que decide es el residual income.
  • El DTI pasa de 41% pero el residual income excede la guía en 20% o más: por 38 CFR 36.4340(c) no se requiere ni justificación del supervisor ni segunda revisión.
  • No hay diferencial verificado de USDA frente a convencional. OBMMI publica una serie propia (p. ej. OBMMIUSDA30YF) y esa es la vía correcta: observar el diferencial, no postularlo. Se cotiza con el ancla convencional, sin restar nada inventado.
  • DTI de 44%, por encima del 41% de USDA (TOPE DURO). 7 CFR 3555.151(h)(1)(i): 29/41, el par más estricto del catálogo. Cuidado infantil y transporte NO cuentan como deuda.
  • Ratio de vivienda de 44%, por encima del 29% de USDA.
  • No hay diferencial verificado de JUMBO frente a convencional. OBMMI publica una serie propia (p. ej. OBMMIJUMBO30YF) y esa es la vía correcta: observar el diferencial, no postularlo. Se cotiza con el ancla convencional, sin restar nada inventado.

What lenders actually charged here

The gap between two lenders is bigger than the gap between two programmes

Everything above prices the rulebook. This prices the people who apply it — with the regulator’s own file, ordered on a criterion anyone can reproduce, and never on a rate.

The spread at your profile

Same county, same program, same term, same occupancy, same unit count, same LTV band, same DTI band, same loan size. Borrowers an underwriter could not tell apart.

3.71×times more, at the same profile
Projected, not counted here
$6,055$11,368$22,480

The mark is where this page put you, and it sits at the median because nobody has quoted you yet — we anchor to what lenders actually charged rather than to a number we made up. The width of the band is the entire argument: at this same profile, someone paid the left end and someone paid the right end.

10th percentile
$6,055
25th
$8,665
Median
$11,368
75th
$17,020
90th percentile
$22,480

Source: Mediana nacional de CFPB, «2023 Mortgage Market Activity and Trends» (diciembre 2024), pp. 24-27, Tablas 3D y 3H, con la forma de la distribución tomada de la cohorte contada de Harris County TX 2025. Proyección, no medición. · Medianas de 2023; forma de 2025 · Checked Sep 10, 2026

What lenders actually charged here

Not estimated rates. Closing costs reported to the regulator, counted inside one county, one program and one narrow profile band, ordered cheapest first.

We have not counted this county yet. Ranking lenders here would be inventing, so we do not. What you see is the national median for this program, shaped by the one cohort we did count. Here is that cohort, so you can see the mechanism working on real ground.

This is history, not an offer. HMDA arrives three to fifteen months late, carries no origination date and no credit score, so a lender that closed when rates were high cannot be told apart from a lender that is expensive. We rank on closing costs, which do survive that limitation, and never on rate.

What that number measures, exactly

Inside

  • A. Origination Charges (comisiones del prestamista, incluidos los discount points)
  • B. Services You Cannot Shop For (tasación, credit report, flood certification, y el UFMIP de FHA cuando lo paga el prestatario)
  • C. Services You Can Shop For (settlement agent, seguro de título del prestamista)

Outside

  • E. Taxes and Other Government Fees (registro, impuestos de transmisión)
  • F. Prepaids (interés prepagado, primas, impuestos)
  • G. Initial Escrow Payment at Closing
  • H. Other (póliza de título del propietario)
Transfer taxes are outside, and they are not small: on a $500,000 house in Philadelphia they run $22,890 — three times the national median total loan cost — and they appear in neither this file nor any comparison site.

Everything above describes Tarrant County, TX on the FHA programme. Change the county or the programme in the form and this whole section changes with it.

45-day window

What turns all of this into a price

Nothing on this page is an offer, and no page anywhere can be one: lenders do not publish their pricing. What does exist is the Loan Estimate — free, legally owed to you within three business days, and the only comparable artefact in American mortgage lending. Every comparison site in the country builds on advertised rates instead. Build on this.

1

Take the five figures, not six

Name, income, property address, estimated property value and the loan amount. Hold back the Social Security number and no hard pull fires, no trigger lead is sold, and no Loan Estimate obligation starts until you decide it should.

2

Ask three to five lenders inside forty-five days

Multiple mortgage credit checks inside a 45-day window land on your report as a single inquiry. That is the CFPB’s own wording, and it is the reason shopping around costs you nothing.

3

Line the Loan Estimates up against these numbers

Block by block. Where a lender’s figure and this page disagree, the lender’s figure is the real one and the gap is a question worth asking out loud.

“Within a 45-day window, multiple credit checks from mortgage lenders are recorded on your credit report as a single inquiry”A lender may not charge you anything for a Loan Estimate except the cost of the credit report. 12 CFR 1026.19(e)(2)(i)(B).
We never sell your information, and we never introduce you to a lender. Nothing on this page leaves your browser: you make the calls, to the lenders you chose.
FHA · 7 years$320,878See the table