Auditable evidence.
Qualified decisions.

Research source inventory

The frozen discovery register has 47 files/access entries. The current source model catalogues 123 entries; these are not 123 independent research reports. The core research consists of two CRE Matrix transaction files and one PropEquity PDF with two workbook supplements. Third-party and public material retain original publisher lineage.

Source extraction is complete for 109 of 123 catalogued entries;14 remain partial / unavailable. “Reviewed” and automated field processing do not mean an attachment is fully legible or its facts independently verified.

Evidence: source_inventory.json · source_inventory.csv · verification_source_register_v5.csv.

The fresh full-population ledger reconciles the original 47 entries into 41 attachment entries plus 6 non-attachment evidence entries, and separately accounts for 76 later web entries. It retains 9 unsupported DWG files, 4 unreliable OCR documents and 1 failed original authority-access entry. The new media universe is separate and is not added to legacy attachment counts.

Evidence: source_population_audit_v5.json · source_population_audit_v5.csv · source_population_audit_v5.html.

Gmail attachment and report audit

Documented historical CRE and Tarush searches used no date restriction and exhausted their returned pages. CRE searches identified 7 relevant messages; the two relevant datasets were processed in full. Tarush’s relevant 15 Oct 2026 forward contains three PropEquity attachments, not three independent providers. The frozen broader audit examined 602 message metadata records,12 candidate messages and 27 attachment reads; this is bounded discovery, not proof of complete mailbox coverage.

Original attachment hashes, sender, sharer, publication/report dates, reporting period and extraction status are retained. Supplier invoices, signatures and irrelevant files are explicitly excluded. Drawings and incomplete OCR remain disclosed rather than claimed fully read.

Evidence: source_inventory.json · verification_source_register_v5.csv.

A fresh Gmail delta check returned 4 unique matches and found 0 new relevant market messages / 0 new relevant attachment versions. The population audit reopens original local files and reconciles raw counts, columns, sheets, formulas and source hashes. This does not certify the entire internet or mailbox exhaustive.

Evidence: source_population_audit_v5.json.

Full CRE Matrix population reconciliation

All 11,909 identities remain retained, including records excluded from a particular metric. V5 accounting statuses describe verification/usefulness and do not replace source rows. A consideration/date can support some observed counts even when its carpet area is unsuitable for a rate. Configuration is never inferred solely from size.

Evidence: record_audit.csv · verification_transaction_audit_v5.csv · transactions.csv.

PropEquity report and workbook reconciliation

All 93 PDF pages were text-extracted and visually reviewed in contact sheets, with key sales tables inspected individually. All 9 RERA workbook sheets and all 3 checklist sheets were enumerated. The RERA workbook has 6,699 nonempty rows, including 6,602 unit specifications; the checklist has 32 nonempty rows. These are not registered sales. Raster plan dimensions were not converted into numeric area evidence merely because a page was rendered.

The original checklist’s carpet/configuration objection remains substantive: e.g. Casagrand 2 BHK specification ranges extend above the PDF’s 852–967 sq ft range. A response or a FINAL filename does not prove correction. Source report strategy and forecast figures remain source claims, distinct from this user’s 25 lakh plan.

Evidence: source_conflicts.json · rera_inventory_specifications.json.

The full fresh workbook audit enumerates 5,580 formula cells across the source population. Cached original values are preserved; fresh native Excel recalculation was not performed. Reviewer comments and responses are extracted as worksheet text, including unresponded density, L-shaped living/dining, butterfly-layout and broader-catchment questions. No causal layout/density sales advantage is accepted without evidence.

Pristine O 2 World remains a material unresolved case: 18 original specification rows label 113.35 sq m as 4 BHK, while the final PDF ranges omit those rows and a secondary phase table calls matching-area units 3 BHK. Referral to the Excel file did not resolve the discrepancy. No configuration is relabeled from area alone.

Evidence: source_population_audit_v5.json · critical_evidence_research_v5.json.

Third-party research comparison

CRE registrations measure transaction records for explicit source geographies and periods. PropEquity supply/absorption tables use their own catchment, phase, inventory and time definitions. Their totals cannot be merged, averaged or substituted without reconciling scope. Developer sites and portals add product/marketing evidence; multiple reproductions of one publisher are one lineage.

Broader land, design, permission and operator materials are catalogued where relevant. A nearby JLL opportunity or an S. No.39 permission package is not assumed to be the proposed site’s entitlement. Neither creates additional apartment registrations.

Evidence: source_inventory.json · source_conflicts.json · competitor_register.json.

Project-level verification

The unified data retains 427 source project identities. V5 researches 46 public residential competitors, with 42 source-attributed representative points; the frozen V3/V4 baseline was 44 / 40. Stage 2 verification applies to individual fields, never an entire project badge. A located source point does not verify legal boundaries, every tower or all RERA phases.

Source claims remain available for qualified exploration. Missing authoritative or independent exact-scope evidence prevents a Verified label. A fresh local review date is not invented as the original source publication date.

Evidence: verification_fields_v5.json · verification_issues_v5.csv · competitor_register.json.

Evidence: critical_evidence_research_v5.json · critical_evidence_research_v5.csv.

Carpet and terrace area validation

RERA carpet, exclusive terrace and exclusive balcony are independent fields. Developer planning carpet=RERA carpet+known exclusive terrace. Extended usable=RERA carpet+known terrace+known balcony. A combined balcony/terrace,EBVT,utility,sundeck or marketed total is not silently split into a terrace. Unknown is not zero.

Within the current V5 phase-supported 5 km cohort, 2,373 records have known RERA carpet and 2,361 support a comparable RERA rate; 436 have known planning/model area and 426 pass existing area/rate comparability gates. The frozen V3/V4 baseline was 2,346 / 2,334 / 410 / 400 respectively.132 published typology/product rows lack a separately reconciled terrace, so modeled saleable is withheld. The V5 investigation population includes 160 typology records under its broader audit definition.

Evidence: area_model_summary.json · loading_typology_audit.csv · loading_area_audit.csv · verification_transaction_audit_v5.csv.

Rate and saleable-area comparison

The 40% is a standardized management assumption over developer planning carpet, not a competitor’s actual loading. Balcony stays separate. Loading controls are editable in the app; static report figures below use 40%. Actual documented saleable takes priority only where appropriate unit-specific evidence exists; none is documented in these transaction records.

These basis-specific medians can use different eligible homes. Same-cohort comparisons in the app explicitly hold the matching modeled-rate cohort constant. Registered tickets are unchanged when loading changes. Headline advertising is not joined to a typical floor plan or a historic registration; current unit-matched quotation rates remain unavailable.

Evidence: area_loading_definitions.json · loading_project_comparison.csv · loading_area_audit.csv.

Map-based competitor intelligence

The user-confirmed site is 18.551876,73.966402. Source-attributed points place 39 researched projects within 5 km straight-line distance, compared with 37 in the frozen V3/V4 baseline; this is not a complete geographic census. V5 adds MapLibre, geodesic 1/2/3/5/7/10 km rings, clustering, project cards, marker layers, radar and separate cumulative/band counts. Ring vertices and stored distances pass independent math checks.

The current browser lacks WebGL, so the MapLibre GL rendering path cannot be claimed browser-tested here. The Leaflet geographic fallback has been exercised with online road tiles, clusters, source pins and rings. Satellite/hybrid is disabled because no authorized imagery provider is connected. Source coordinates retain their actual confidence and mapped-phase scope.

Evidence: competitors_public.json · verification_fields_v5.json.

The two exact-RERA additions are Godrej River Crest P52100077127 and The Strand P52100050923. Their official developer-linked points add 27 registrations; no transaction, original source field or configuration label was replaced. Three configuration conflicts remain held out of supported configuration conclusions. Seven geographic source references are a separate V5 evidence universe, not a relabeling of the frozen 123-entry source catalog.

Evidence: geography_enrichment_audit_v5.json · current_geography_summary_v5.json · geography_sources_v5.json · transactions_v5.csv · record_audit_v5.csv.

Inventory and possession analysis

Historic launched/sold/booked/unsold snapshots remain dated evidence. Registered transactions,6,602 source unit specifications and developer-advertised project scale do not establish currently available apartments. Available stock is not total units minus registrations. Current independently verified available-for-sale inventory and future releases remain unavailable.

Original/extendedMahaRERA possession, developer-advertised dates, provider completion dates and achieved possession/occupancy are separate. Reported mapped-phase completion can support a provisional timeline; it does not verify statutory possession or future available stock. Direct current MahaRERA / OC verification remains open.

Evidence: competitor_register.json · source_conflicts.json · verification_issues_v5.csv.

Apartment demand distribution

Observed configuration, size and ticket bands are linked in the app. Missing configuration is retained as Unspecified. The 12-month recent source and 15-calendar-month older Mundhwa source are not one 36-month appreciation series. Monthly medians reflect changing registered product mix, not a controlled repeat-sales index.

What the data says: registered purchases show the distribution of completed source-recorded transactions. Why it matters: broad market volumes are not our achievable bookings. What we should do: use comparable size/ticket cohorts and obtain actual release/current-stock denominators before ranking true absorption.

Evidence: analysis_v2.json.

Premium apartment demand study

The V5 location overlay exposes three additional source-labeled 1400-band 3 BHK registrations. All three have contradictory configuration evidence and are held out; the supported 1400-band conclusion remains zero. Original 3 BHK labels are not relabeled as 4 BHK. The 1400/1500 sq ft 3 BHK questions use ±50 sq ft half-open RERA bands; the 2000 sq ft 4 BHK question uses 1900–<2100 sq ft. The two 3 BHK bands do not overlap. The 3.5 BHK cohort uses source classifications across all known sizes. The four 1500 sq ft 3 BHK records are resale registrations at Panchshil Towers, involving only two repeated project/building/unit identities; they are not four independent primary launch sales. No primary-launch comparable supports a fixed allocation for these tested premium size bands. Ticket medians use the displayed registered cohort; a modeled-rate median uses only the eligible area subcohort. These counts do not establish launch absorption or future stock.

3.5 BHK can be added to the design studio only after management enters actual proposed RERA carpet, terrace and rate inputs. No fabricated default 3.5 BHK size or demand forecast has been inserted.

Evidence: analysis_v2.json · loading_area_audit.csv.

Official Mahindra IvyLush material distinguishes Tower E 3 BHK RERA carpet 1,229.25 sq ft from aggregate 1,407.72 sq ft; balcony/utility explain the difference within source rounding and do not become exclusive terrace. Official Tower E 4 BHK RERA areas near 1,420–1,426 sq ft are not 2,000 RERA sq ft comparables. Source marketed aggregates are therefore excluded from these direct premium carpet bands.

Evidence: critical_evidence_research_v5.json.

Product opportunities and market gaps

The research supports provisional comparisons of configuration, ticket affordability and registered product ranges. It does not yet prove an unsupplied category, a causal amenity premium or the fastest-selling home after controlling for released supply. Opportunity scores remain transparent research proxies and cannot replace a reliable supply census.

What the data says: some proposed products have observed comparable transactions. Why it matters: existence of comparable sales does not identify the optimum inventory allocation. What we should do: combine validated comparable areas/tickets with current dated release/stock evidence, then retest pricing and mix before final investment sign-off.

Evidence: product_opportunity_scores.csv · verification_certificate_v5.json.

Provisional 25 lakh saleable inventory mix

Both 25 lakh development and 10 lakh launch objective are user-confirmed saleable totals. The balanced reference contains 2,012 whole homes, allocating 24,95,080 saleable sq ft plus 4,920 saleable sq ft of explicit design reserve. Preliminary own-project terrace=0 is a labeled design assumption, not competitor evidence. Blank terrace inputs gate inventory.

Conditional gross inventory revenue at those editable opening rates is₹2,352.39 Cr. Ticket=RERA size×opening RERA rate; equivalent planning/saleable rates share that ticket. The three velocity/balanced/premium strategies are available for comparison. No 1 BHK is proposed. The current default 4 BHK design is 1450 RERA sq ft; the 2000 sq ft demand test is a separate product question, not silently substituted inventory.

Combination-ready smaller units remain an uncompleted feasibility investigation. Structural, services, fire, sanction, purchaser/agreement and legal permissibility have not been established. No combination strategy is recommended as approved or lower risk without that evidence.

Evidence: product_allocations_three_strategies.csv · model_results.json · master_requirements_register_v5.json.

Six-to-eight-month sales strategy

The 10 lakh target requires 1.67 lakh saleable sq ft per month over 6 months or 1.25 lakh over 8 months. At this reference mix it is 807 weighted home equivalents. This is a requirement, not proven local demand. The default base 8-month case models 488 net bookings but 324 registrations; agreements, registration lags and collections remain separate.

The separate stretch preset assumes 9000 monthly leads and 1.35 release buffer. Its existing audited 8-month output models 839 registrations/10.40 lakh saleable sq ft with about₹26.12 Cr acquisition spend. This is an execution stress test, not a calibrated probability. Obtain CRM conversion, lead cost, team capacity, channel mix and dated competitive-release evidence before approving a funded launch target.

Evidence: launch_plan_base_44_months.csv · launch_plan_conservative_44_months.csv · launch_plan_aggressive_44_months.csv · stretch_8_month_plan.csv · model_results.json.

Thirty-six-month plan and longer inventory horizon

Month 36 means the first 36 months from launch. The complete model extends to 44 months:8 launch months plus 36 further months. Tables expose both horizons rather than calling them the same period. Long-run case ordering can reflect release and legal-stage timing; no case is assumed to be a guaranteed upper or lower bound.

All outcomes are conditional on editable conversion, cancellation, release, pricing and post-launch lead assumptions. The 44-month plans preserve cohort prices, pending agreements, pending registrations and collections. These are not verified future sales.

Evidence: model_results.json · launch_plan_base_44_months.csv.

Data verification certificate

Stage 1 validates extraction/accounting. Stage 2 requires exact-scope independent field evidence and source lineage. Stage 3 independently recalculates computation and recommendation consistency. The current engine has 10 critical test groups passed,0 failed and 1 not checkable.

Independent float / Decimal arithmetic, exact unit conversions and geographic recalculation strengthen the computations. They do not corroborate the original transaction deed or approved plan. Check H is Qualified against the final browser receipt and 17 independent client contracts. Browser review covers 18 layouts across six routes, with selected interactions tested; it does not certify physical mobile installation, hardware pinch gestures or the unavailable WebGL rendering path. Publishing a visibly provisional exploration app is separate from certifying final development conclusions.

Evidence: verification_certificate_v5.json · reconciliation_report_v5.html · verification_fields_v5.json · verification_issues_v5.csv · v5_application_verification.json.

Outstanding evidence and material limitations

Targeted independent research must continue where meaningful authorized avenues remain; an unavailable field is not a completed factual finding. Current discovery does not establish that all reasonable public/authority avenues are exhausted.

Evidence: verification_issues_v5.csv · project_visual_register_v5.json · requirements_compliance_v5.html.

Targeted authority research encountered normal CAPTCHA-gated project details and unavailable/time-out routes; no challenge was bypassed and no statutory fields were extracted from that gate. The critical research ledger records the next legitimate action and marks decision-critical avenues as not exhausted. Neco inventory/possession and Pristine unit/area conflicts remain open; document identity findings do not solve current stock or unit-area verification.

Evidence: critical_evidence_research_v5.json · source_population_audit_v5.json.

Implementation-status audit of earlier prompts

The master register preserves 1,797 distinct original instruction statements from 5 pasted briefs, the standalone loading addition and accepted user answers, grouped into 28 categories. Exact repeated wording is merged with all references; each item stores original section/line/message reference, implementation, dependencies, test method/result and remaining work under the required six statuses.

“Completed and tested” is restricted to its stated implementation/calculation scope. It never overrides independent factual gaps. The earlier 1.4×carpet loading base and blue / yellow palette are specifically linked to their later superseding instructions. The 25 lakh and 10 lakh saleable confirmations remain active and require no repeated question.

The register is not a generic completion percentage: paragraphs and field-list items are not equally weighted capabilities. The final V5 browser receipt is Qualified: 18 layouts across six routes at 360, 768 and 1440 CSS pixels passed without document overflow. Tested interactions include the geographic fallback, linked cards, fullscreen, selected filters, representative profile imagery/zoom, theme controls and accuracy drilldown. Publication outcome is recorded separately in the delivery receipt. This bounded UI evidence does not complete broad research or final-decision requirements. Packaging checks passed for 105 hosted files, a largest hosted asset of 24.316 MB, archive CRC integrity and 455 relative links in each archive. Missing sources, statutory facts, connected AI, satellite authorization and combination feasibility remain explicit.

Evidence: master_requirements_register_v5.json · master_requirements_register_v5.csv · requirements_compliance_v5.html · v5_application_verification.json.