twentyci-logo

Automated Valuation Model (AVM) for Mortgage Lenders

Our Decision AVM combines verified property foundations with real-time listing activity and homemover signals—detecting market shifts before completions hit the register. Model refreshed weekly (full signal updates); monthly recalibration; quarterly bulk test reporting against completed sales.

See our latest bulk test results and methodology in the governance pack—available on request.

Request Governance Pack & Demo
twentyci-avm-example

Deliver fast, reliable residential valuations for origination and portfolio monitoring—with transparent confidence grading built for lending policy.

Decision-grade accuracy: MDAPE 2.8% overall

99.6% coverage of residential properties in England, Wales, and Scotland

Up to 90% of valuations within ±10% of actual sale prices (bulk-tested)

Confidence bands: A (highest) to F/U (lowest), derived from Forecast Standard Deviation (FSD)

MDAPE and hit rates from quarterly bulk testing against market value sale prices at completion. Performance varies by property type, region, LTV band, and confidence level. Full methodology, inputs, assumptions, and evidence in documentation.

One AVM Family for Origination and Portfolio Risk Management Origination / Underwriting

  • Validate value at application with confidence-graded outputs to inform policy thresholds
  • Automate triage: straight-through for high-confidence cases; desktop or physical for exceptions
  • Dual sales and rental valuations—supports owner-occupied and Buy-to-Let (BTL)
blue-graph-in-shape

Portfolio / Risk Management

  • Refresh back-book values for LTV monitoring, IFRS 9 impairment, and proactive risk reviews
  • Capture early market movement via live listing and homemover signals (not just delayed completions)
  • Historic valuations for trend analysis, back-testing, and regulatory reporting

Interim Case Studies / Outcomes

  • Tier 2 lender reduced desktop valuation referrals by 40% via confidence-based rules
  • Building society achieved same-day decisions on 65% of applications, cutting the average turnaround from 3 days

The Challenge with Traditional AVMs –
and How We Address It

Most AVMs rely heavily on historic completion data, which lags market reality by weeks or months. In shifting conditions—rising/falling prices, regional hotspots—this creates uncertainty in underwriting decisions and delayed risk signals for portfolios.

twentyci-avm-report-example

When the Market Moves, Lender Risk Moves Faster

Completion records arrive late. TwentyCi Decision AVM layers in live listing activity and homemover behaviour signals that emerge pre-completion, delivering a timelier valuation signal for both origination and back-book oversight.

What Sets TwentyCi Decision AVM Apart

1. Live-Market Visibility

Verified base (HM Land Registry Price Paid, EPC, OS AddressBase, Royal Mail PAF) + real-time signals from >25 billion data points on current listings and buyer/seller behaviour.

 

2. Policy-Ready Confidence Grading

Every output includes an A-F/U confidence band (based on FSD), enabling clear rule-based acceptance, exception handling, and audit trails. 

 

3. Governance-Focused Model Design

Fully documented inputs, methodology, assumptions, and benchmarking—built with PRA model risk management expectations in mind. Quarterly bulk tests against actual completions provide independent validation.

How the TwentyCi Decision AVM Works

Layer 1: Factual Foundation (The “What”)

Core data from official sources: HM Land Registry Price Paid, EPC attributes, Ordnance Survey location context, Royal Mail PAF/AddressBase Premium (UPRN-verified). 99.6% residential coverage across GB.

Layer 2: Market Movement Signals (The “When”)

Overlay live listing activity, homemover intent, and behavioural data—signals that precede transactions, reducing lag in volatile markets.

Layer 3: Advanced Modelling (The “How”)

Proprietary, documented algorithms generate current/historic sales and rental valuations + confidence grading. Monthly recalibration for model stability; weekly signal refreshes for currency. 

Key Benefits for Mortgage Lenders

neon-house-data
  • Faster, Smarter Origination — Confidence-based straight-through processing where policy permits; reduce manual interventions.
  • Reduced Valuation Uncertainty — Graded outputs support consistent rules across channels, teams, and products.
  • Market-Responsive Portfolio Oversight — Earlier detection of LTV shifts and risk trends via forward-looking signals.
  • Single Solution for Sales & Rentals — Unified model family simplifies vendor management for owner-occupied and BTL.

Flexible Delivery for Lender Systems

data-in-shape
  • API (Real-Time) —Instant queries for value, range, confidence band, and supporting fields (<1s response). Ideal for application workflows.
  • Bulk Feed (SFTP/ETL) — Scheduled portfolio refreshes (e.g., monthly) across full or targeted address bases.
  • SaaS Platform — Embed outputs in TwentyCi tools for operational teams.
  • Integrations — Compatible with all sourcing platforms.

Request a demo, and we’ll provide

  • A live walkthrough using lender-style scenarios (origination and portfolio refresh)
  • A sample output showing value plus confidence rating and supporting fields
  • Integration options: API endpoints or bulk delivery routes
  • Governance pack: methodology summary and bulk test summary