Source provenance
A structured market-data pipeline with source and period metadata, keeping observations distinguishable from derived measures.
Research product · In development
HBF Intel
Homebuilding market intelligence that makes its sources, methodology and uncertainty part of the product.
This is a project brief, not a live market feed or a published score report. Hosted publication and live AI narrative generation remain unverified or pending approval.
The opportunity
Credit teams need more than another collection of housing statistics. They need a repeatable view of market conditions, an explanation of what changed, and enough evidence to challenge the conclusion.
HBF Intel is designed as an owned research product: a consistent methodology, reusable data pipeline and publication workflow that can support both internal decisions and external thought leadership. The value lies in a research capability that improves over time, rather than a one-off dashboard.
What I have built
Repeatable methodology
Scoring and historical-validation work designed to make assumptions inspectable and changes testable.
Publication controls
Snapshots and review boundaries that separate internal calculations from material approved for publication.
Evidence-led narratives
A rule-based narrative fallback. Live AI-generated narratives still need end-to-end verification before being represented as a working capability.
Why this belongs in an underwriting portfolio
The product challenge is the same one running through the working demos: make a complex analytical process understandable, expose the source of a conclusion, and decide where automation should stop for human review.
For a commercial real estate data and software business, this is an example of how I think about building a research product around an underwriting user’s decisions. It is not a proposal to reproduce CoStar’s licensed data or a claim of an existing integration.
Independent project by Rohail Abid. CoStar branding is illustrative; no affiliation or endorsement is implied.