Collateral Finding — Credit File Exhibit

§1Finding

1004 subject-to-completion · Rehab · Essex County, NJ · 1247 Maple Avenue, Newark, NJ 07104

Value the file supports

$402,100

Appraised $455,000
Correction −$52,900 · −11.6%

Supported range

$366k–$438k

80% coverage: four times in five, the value the evidence supports falls inside this range.
Confidence 68/100 · moderate

What moved it · by dollar effect

RTL-V-01⚑ Budget versus comparable finish level−$28,300
RTL-V-04ARV versus neighbourhood ceiling−$11,000
RTL-C-15Market conditions adjustment−$8,000
RTL-A-02Gross adjustment ratio−$5,600
RTL-C-03RTL-A-07widen the range+4.0 pts

4 corrections, 2 range wideners, 0 flags, 0 checks not run, of 56. The largest item is the budget: it funds a lower finish than the comparables were sold at. A scope of $146,000 clears it (§7).

§2Subject

Address

1247 Maple Avenue, Newark, NJ 07104

Report

1004 subject-to-completion · 14 pages · Rehab

Living area

1,850 sf

Condition and quality, as-is

C5 / Q4

As-is value, appraised

$295,000

After-repair value, appraised

$455,000 ($246/sf)

Scope of work, submitted

$96,200 ($52/sf) ⚑ reconstructed from 12 line items · §8

Lender reference

LN-2026-08-0142

Appraiser licence

NJ-42RG00891

Census tract

34013001700 · Essex County, NJ

§3Corrections

Appraised after-repair value $455,000

RTL-V-01 Budget versus comparable finish level ⚑ reconstructed input Sphinx-Original −$28,300

RTL-V-04 ARV versus neighbourhood ceiling Sphinx-Original −$11,000

RTL-C-15 Market conditions adjustment Sphinx-Original −$8,000

RTL-A-02 Gross adjustment ratio Industry Baseline −$5,600

Gross corrections −$52,900

Overlap suppressed (RTL-V-01 ∩ RTL-V-04) $0

Supported value $402,100

Overlap: RTL-V-01 and RTL-V-04 can both respond to the same overvaluation. When they do, the smaller effect is reduced by the shared portion, and the line above shows the reduction. On this file the shared portion was nil. Provenance is shown per check; §5 gives the arithmetic behind each figure and §11 the values used.

§4Supported range

Supported range · 80% coverage

$366,000 – $438,000

Supported value $402,100 · appraised $455,000

Confidence 68/100 · moderate — how firmly the file supports its own number. Not a probability: the coverage figure is the probability.

±9.0%

range width

Four times in five, the value this file's comparable evidence supports falls inside the range. The range measures how firmly the file supports its number. It is not a forecast of sale price.

Base range (Essex County rehab, $350–500k) ±5.0%

RTL-C-03 Comp set depth (2 qualifying sales) +2.5 pts

RTL-A-07 Round-number clustering +1.5 pts

Range width ±9.0%

$366k

$438k

$402k

$350k $400k $450k $500k

The appraiser's four comparables produced adjusted values of $440k, $455k, $468k and $450k; the reconciled $455,000 sits inside that range (RTL-A-12 passed). The corrections in §3 act on the evidence behind those figures, not on the reconciliation.

§5Findings in detail

RTL-V-01 Budget versus comparable finish level −$28,300

The comparables supporting the after-repair value sold at a C2/Q3 finish. The submitted scope funds a finish consistent with C3/Q4. Nobody in the process is asked to check that the budget reaches the standard the comparables were finished to; this check does.

Finish level of the comparables: $79/sf
Finish level the submitted scope funds: $52/sf
Cost gap: ($79 − $52) × 1,850 sf = $49,950
Value recovery rate applied: 0.567 (provisional for this submarket and project type; replaced by the calibrated rate once realised sales exist)
Effect: −$49,950 × 0.567 = −$28,300, to the nearest $100
Fires when: The budget falls materially short of the finish level implied by the comparables

To clear: increase the scope to $146,000 ($79/sf), the finish level the comparables represent. See §7.

⚑ Input: the scope total of $96,200 is not printed as a total in the report; summed from the 12 line items on the scope of work. See §8.

Method: RTL-STD-001 v1.0 §V-01 · Sphinx-Original
RTL-V-04 ARV versus neighbourhood ceiling −$11,000

The appraised value works out to $246/sf. The tract's ceiling, the 90th percentile $/sf of closed sales, trailing 12 months, is $240/sf. Values above the ceiling face longer marketing time and more reversion risk.

Ceiling: $240/sf × 1,850 sf = $444,000 (census tract 34013001700)
Appraised after-repair value: $455,000
Effect: −($455,000 − $444,000) = −$11,000
Fires when: The ARV exceeds the neighbourhood ceiling

No borrower input changes this finding. It describes where the value sits against the neighbourhood's own sales.

Method: RTL-STD-001 v1.0 §V-04 · Sphinx-Original
RTL-C-15 Market conditions adjustment −$8,000

The grid applies +1.5% a year on the Date of Sale/Time line to each comparable. The county index moved −2.1% over the trailing year. Over each comparable's own lag (98 to 287 days) the grid moves the sales up when the index moves them down. Backward-looking only: nothing here forecasts.

Index (FHFA county series, 2026Q2), trailing year: −2.1%
Stated time adjustment, annualised: +1.5% · gap 3.6 pts
Lag-weighted mean sale price, (sale price × days ÷ 365) averaged over the 4 comparables: $222,728
Effect: −$222,728 × 3.6% = −$8,000, to the nearest $100. Equal weights: the reconciliation states none.
Fires when: Stated time adjustment is inconsistent with the index move over the comparable's own lag

No borrower input changes this finding. It compares the grid to the index over the months the comparables have been closed.

Method: RTL-STD-001 v1.0 §C-15 · Sphinx-Original
RTL-A-02 Gross adjustment ratio −$5,600

Gross adjustments average 19.2% of sale price across the four comparables, against a local median of 12.0% for Essex County rehab, $350–500k sales. Adjustments this large mean the comparables are further from the subject than the grid makes them look.

Gross adjustments by comparable: 18.9% · 16.6% · 23.7% · 17.6% · mean 19.2%
Firing level: 1.5× the local median = 18.0%, or any comparable above 25%
Effect: −$455,000 × (19.2% − 18.0%) = −$5,600, to the nearest $100
Fires when: Above 25% or 1.5× local median

No borrower input changes this finding. It describes how far the comparables had to be adjusted.

Method: RTL-STD-001 v1.0 §A-02 · Industry Baseline
RTL-C-03 Comp set depth +2.5 pts

2 qualifying sales within 0.5 mi in the trailing 180 days. This does not move the supported value. It widens the range to reflect thinner evidence.

Range widened by +2.5 pts
Fires when: Comparable depth is thin for the submarket

No borrower input changes this finding. It describes the market's depth at the effective date.

Method: RTL-STD-001 v1.0 §C-03 · Sphinx-Original
RTL-A-07 Round-number clustering +1.5 pts

17 of 17 adjustment lines end in 000, and the condition adjustment is the same +$40,000 on all four comparables. Habit rather than derivation is a documented marker of post-hoc support. It widens the range rather than moving the value.

Range widened by +1.5 pts
Fires when: Round-number adjustments predominate

No borrower input changes this finding. It describes how the grid's adjustments were derived.

Method: RTL-STD-001 v1.0 §A-07 · Sphinx-Original

§6Comparable sales from the appraisal

Property	Sale	C/Q	Adjustments	Adjusted
Subject · 1247 Maple Ave, Newark NJ 1,850 sf	as-is $295,000	as-is C5/Q4 budget funds C3/Q4	Comparables are C2/Q3 · RTL-V-01	ARV $455,000
87 Oak Street 0.3 mi · 1,780 sf	$412,000 142 days ago	C2/Q3	GLA +$7k · Cond +$40k · Qual −$25k · Age +$4k · Time +$2k Net +$28,000 gross $78,000 · 18.9%	$440,000
234 Elm Avenue 0.4 mi · 1,920 sf · closest in size, most recent	$445,000 98 days ago	C2/Q3	GLA −$7k · Cond +$40k · Qual −$25k · Time +$2k Net +$10,000 gross $74,000 · 16.6%	$455,000
512 Cedar Lane 0.5 mi · 2,100 sf	$498,000 203 days ago	C2/Q2	GLA −$25k · Cond +$40k · Qual −$49k · Time +$4k Net −$30,000 gross $118,000 · 23.7%	$468,000
19 Birch Road 0.6 mi · 1,800 sf · oldest sale	$425,000 287 days ago	C2/Q3	GLA +$5k · Cond +$40k · Qual −$25k · Time +$5k Net +$25,000 gross $75,000 · 17.6%	$450,000

All four comparables are C2. The subject is C5/Q4 as-is and the submitted budget funds C3/Q4; that gap is RTL-V-01. Time lines are the grid's +1.5% a year applied to each comparable's lag; that is RTL-C-15. Gross ratios are RTL-A-02.

§7Scenarios

Scenario	Supported value
As submitted ($52/sf scope)	$402,100
At comp-matched scope ($79/sf)	$430,400

Scenarios change the borrower's inputs only. They are not recommendations, and they say nothing about what the appraiser would need to change.

§8What was read from the appraisal

Fields read41, from 14 pages of a 1004 subject-to-completion Read cleanly40 Needs a human look1 · scope total Could not be read0 · 0 checks suppressed Comparable grids4 of 4 reconcile: sale price plus the adjustment lines equals the adjusted value on every comparable

Scope total $96,200 is not printed as a total in the report; summed from the 12 line items on the scope of work. Checks that use it: RTL-V-01, RTL-V-03. Each line was read cleanly, so they ran on the sum and are marked ⚑. Confirm the total or correct it and the finding re-issues.

Every field carries a confidence score and a source page; a check that cannot read what it needs is suppressed and counted above, never run on a guess.

§9Coverage

56

Checks run

6

Fired

50

Passed

0

Not run

Fired: 4 with a dollar effect, 2 widening the range, 0 flags. Not run means an input could not be read cleanly; it is shown even when zero so the reader knows the case was tested.

Check family	Result
RTL-C — Comparable selection (15 checks)	2 findings · 13 passed
RTL-A — Adjustments (12 checks)	2 findings · 10 passed
RTL-V — The after-repair value (5 checks)	2 findings · 3 passed
RTL-S — Subject & file integrity (16 checks)	16 passed
RTL-P — Appraiser (8 checks)	8 passed

Passed checks worth noting

RTL-C-02 Distance: farthest comparable 0.6 mi
RTL-C-04 Sale recency: oldest sale 287 days before the effective date
RTL-C-05 Bracketing — living area: 1,850 sf bracketed by comparables of 1,780 to 2,100 sf
RTL-A-01 Net adjustment ratio: net adjustments within ±6.8% of sale price on every comparable
RTL-A-12 Conclusion inside the adjusted range: $455,000 sits inside the adjusted range of $440,000 to $468,000
RTL-P-01 Licence valid at effective date: licence NJ-42RG00891 valid at the effective date

What this review does not examine

Hidden conditions behind drywall not visible in inspection photos
Whether the scope is achievable at the stated cost in this market
Execution risk — whether the borrower can complete the project
Anything requiring interior inspection beyond what the appraisal photographs show
Title, survey, or legal matters
Sponsor experience or financial capacity
Permit timeline or municipality-specific approval delays
Material cost fluctuations after the effective date

§10Lender response

Finding	Effect	Response	Note
RTL-V-01 Budget versus comparable finish level	−$28,300	No response recorded	—
RTL-V-04 ARV versus neighbourhood ceiling	−$11,000	No response recorded	—
RTL-C-15 Market conditions adjustment	−$8,000	No response recorded	—
RTL-A-02 Gross adjustment ratio	−$5,600	No response recorded	—
RTL-C-03 Comp set depth	+2.5 pts	No response recorded	—
RTL-A-07 Round-number clustering	+1.5 pts	No response recorded	—

The underwriter records agreed, disagreed or not material against each finding, with a note. Responses are stored against this finding version and travel with the exhibit; until loans exit, they are the only label the standard has.

§11Reproduction

Standard at reviewRTL-STD-001 v1.0

Extraction pipelineextract-1.4.2

Market indexFHFA county series, 2026Q2

Source PDFSHA-256 7f3a9c2b…4e81

Appraisal effective15 Aug 2026

PDF received28 Aug 2026

Finding produced30 Aug 2026 09:14 UTC

Verificationsphinx.credit/v/RTL-2026-0842-7F3A

Values used on this file

RTL-V-01	Value recovery rate 0.567, provisional; finish levels $79/sf against $52/sf on 1,850 sf
RTL-V-04	90th percentile $/sf of closed sales, trailing 12 months: $240/sf in tract 34013001700
RTL-C-15	FHFA county series, 2026Q2, trailing year −2.1%; lags from closing date; equal comparable weights
RTL-A-02	Local median 12.0%; fires at 1.5× the median (18.0%) or 25% on any comparable
RTL-C-03	Qualifying sales within 0.5 mi and 180 days; 2 found; range widened +2.5 pts
RTL-A-07	17 of 17 lines end in 000; range widened +1.5 pts
Range	Base ±5.0% for Essex County rehab, $350–500k; coverage 80%

This document states the version of the standard at review and nothing later. What has changed in the standard since, and whether this finding would differ under the current version, is answered by the verification page, which is live.

§12Reference

C2 Good condition — minor deferred maintenance only

C3 Well maintained — limited updating, no major repairs

C5 Fair condition — significant repairs needed

Q2 Good quality construction — above typical

Q3 Above average quality — better than tract standard

Q4 Average quality — typical tract construction

Condition (C) and quality (Q) ratings follow UAD (Uniform Appraisal Dataset) definitions.

Supported valueThe value the appraisal's own evidence holds up once the checks with a dollar effect are applied. Not a valuation.

Supported rangeThe range around the supported value; the coverage figure says how often the supported value falls inside it.

ConfidenceHow firmly the file supports its number, out of 100: 80 and above high, 60 to 79 moderate, below 60 low. Not a probability and not extraction accuracy.

CorrectionA fired check with a dollar effect. Range widener: a fired check that widens the range without moving the value. Flag: a fired check with neither.