Methodology and data sources

Last updated: May 12, 2026.

This page documents how protest.pearlandai.com selects comparable properties and computes the suggested relief amount that appears on each evidence Packet. We publish it because (a) ARB defensibility requires the customer to be able to explain where each number on the Packet came from, and (b) transparency about methodology is the difference between a defensible exhibit and an unexplainable spreadsheet.

1. Source data

All property data — owners, addresses, subdivisions, neighborhood codes, year built, living area, class codes, foundation, exterior, fireplaces, plumbing, garage, pool, lot acreage, and the full appraised / market / land / improvement / assessed value breakdown — comes from each appraisal district's own annual preliminary appraisal roll. We currently ingest data from four districts:

• Brazoria Central Appraisal District (BCAD) — the PACS-format export (layout v8.0.33). We read the APPRAISAL_INFO, APPRAISAL_IMPROVEMENT_DETAIL, APPRAISAL_IMPROVEMENT_DETAIL_ATTR, APPRAISAL_LAND_DETAIL, and APPRAISAL_ABSTRACT_SUBDV fixed-width files and join them on prop ID.
• Fort Bend Central Appraisal District (FBCAD) — the quoted-CSV export. We read PropertyProperty-E (master record), WebsiteResidentialSegs (segment detail: sqft, year, class, pool), PropertyOwner-E (owner names), and PropertyLand-E (lot size) and join them on PropertyID.
• Harris Central Appraisal District (HCAD) — the tab-separated public records export. We read real_acct.txt (master record + neighborhood codes + valuations), building_res.txt (heated area, year built, quality grade), structural_elem1.txt (foundation, exterior wall, GRD letter grade), extra_features_detail1.txt (pool detection + detached garage area), fixtures.txt (story count, bedrooms, baths, fireplace count), owners.txt, and real_neighborhood_code.txt (neighborhood-name lookup), joined on the 13-digit account number.
• Galveston Central Appraisal District (GCAD) — the PACS-format export (same layout v8.0.33 used by BCAD). We read the APPRAISAL_INFO, APPRAISAL_IMPROVEMENT_DETAIL, APPRAISAL_IMPROVEMENT_DETAIL_ATTR, APPRAISAL_LAND_DETAIL, and APPRAISAL_ABSTRACT_SUBDV fixed-width files and join them on prop ID.

Source dataset code, run date, and layout version are printed on every Packet's sources page. We do not enrich CAD data with photographs, on-site inspections, third-party MLS data, or proprietary valuation models. The only non-CAD calculation in the Packet is the median $/sqft of the selected comparable set and the suggested-value figure derived from it.

2. Comparable selection — base variant

Texas Tax Code §41.43(b)(3) lets a property owner argue the appraised value is unequal to the median appraised value of a "reasonable number of comparable properties appropriately adjusted." The base variant operationalizes this in a defensible, deterministic, CAD-data-only way:

1. Filter to residential properties in the same scope (subdivision by default; neighborhood as a fallback for sparse subdivisions — see §4).
2. Filter to properties with the same CAD class code as the subject (a class-code match is the strongest comparable signal the CAD records).
3. Filter to properties built within ±10 years of the subject.
4. Filter to properties whose living area is within ±20% of the subject.
5. Filter by story count and pool amenity match.
6. Sort by appraised $/sqft ascending; take the lowest 10.
7. Compute the median appraised $/sqft of those 10.
8. Multiply by the subject's living area to produce the suggested value.
9. Subtract from the subject's current appraised value to produce the relief amount.

The Packet prints the full filter chain, the population sizes at each step, the 10 selected comparables, and the median calculation in a format the ARB can audit line-by-line.

3. Comparable selection — pool-adjusted variant

Pool homes face a structural problem under the base variant: in many neighborhoods there are too few similar pool homes to produce a defensible 10-comp panel, and pool homes' $/sqft is systematically higher than non-pool homes' $/sqft. The pool-adjusted variant addresses both issues at once. It deducts a county-level estimate of the pool premium from the subject's market value, then compares the resulting "pool-stripped" subject value against the cheapest 10 non-pool comparables in the same neighborhood that match on size, age, story count, and class. The pool premium estimate is documented on the Packet's methodology page and is presented transparently so the ARB can substitute its own estimate if it disagrees. The pool-adjusted variant is offered only when (a) the subject has a pool and (b) the county has a configured pool premium estimate.

4. Scope: subdivision vs. neighborhood

Most subdivisions in our supported counties have enough homogeneous, class-matched, similarly-sized housing stock to produce 10+ comparables within the subdivision boundary. For those properties we default to subdivision scope, which is the strongest §41.43(b)(3) framing because subdivision is the most explicit comparability signal the appraisal district records.

For smaller or more heterogeneous subdivisions, fewer than 10 close matches exist within the subdivision. In those cases we expand to neighborhood scope (the CAD's broader neighborhood code, which groups adjacent subdivisions the district considers similar for assessment purposes). Neighborhood-scope Packets are clearly marked as such on the cover, in the methodology table, and in the §41.43(b)(3) argument paragraph, with an additional disclosure paragraph explaining why subdivision scope was insufficient. The customer chooses scope explicitly at unlock; we never silently substitute scopes.

5. What we do not do

• We do not visit, photograph, or inspect any property.
• We do not adjust comparables for differences the appraisal district does not record (interior condition, recent renovations, deferred maintenance, view, traffic exposure, school assignment, lot orientation, etc.).
• We do not use MLS data, recent sale data, or any non-CAD valuation source.
• We do not run a market-value appraisal, a hedonic regression, or any model the ARB would recognize as an appraisal under USPAP.
• We do not predict or guarantee what the ARB will do with the Packet.

6. Customer review obligation

Before submitting a Packet to the appraisal district or the ARB, the customer is responsible for reviewing it for accuracy and reasonableness. In particular: (a) confirm the subject information matches the property, (b) confirm the comparables are in fact reasonably comparable to the property as the customer knows it, (c) consider whether the customer wants to add condition photographs, repair estimates, or other evidence the Packet does not include, and (d) confirm the suggested value is one the customer is willing to argue for at the hearing. The Packet is the customer's exhibit; the customer's name is on it and the customer presents it.

7. Updates

We may refine this methodology between tax years. We will not change the methodology of an already-rendered Packet — your downloaded PDF is what you bought. If you re-download a Packet at a later date, you will receive the same PDF that was originally rendered against the snapshot of CAD data current at unlock time.