Housing Data and Technology Innovation Challenge

A nine-week, fully remote, open-source challenge running October 14 to December 16, 2026. Teams build transparent tools that estimate how state-level housing reforms are likely to affect housing supply.

The U.S. housing shortage is estimated at 1.2 to 4.7 million homes. States are actively weighing reforms to close that gap, but the people making those decisions rarely have accessible, transparent tools to estimate what a given reform would actually do, which leaves high-stakes choices resting on anecdote and instinct. The Challenge exists to build those tools in the open and put them directly in decision-makers’ hands.

The Challenge is administered by The Turnout and sponsored by Arnold Ventures.

Participation is free. You’re eligible if you are a legal resident of one of the fifty United States or the District of Columbia, are 18 or older at registration, and have a U.S. bank account.

Government employees can participate, but ethics rules on outside activities and prizes may apply to you. Disclose your status at registration and we’ll review it early.

Teams are one to five people, and you can compete solo. If you’re looking for teammates, register as an individual and we’ll help you find them: we run team-formation webinars and a matchmaking channel in the weeks before the roster locks.

A transparent, replicable, explainable tool: a simulator, dashboard, model, or analytical tool that estimates how housing reforms affect housing production or aggregate housing stock at the state level, measured against a baseline. Every tool models a baseline plus at least three of five benchmark reform scenarios, which keeps submissions comparable to one another. The scenarios and their shared modeling assumptions are defined in the participant handbook. Teams are encouraged to add scenarios reflecting live policy debates in the state they choose.

We’re methodology-agnostic. Your tool might be an econometric model, a pro forma simulator, a rule-based policy engine, a geospatial analysis, or something we haven’t imagined. What matters is that it rigorously and transparently addresses real policy questions and is usable by the people it’s built for.

A working tool, a slide deck, a five-minute demo video, a Model and Assumptions Overview, your source code in a public repository, and a short non-technical comparison of the scenarios you modeled. Full specifications are in the handbook.

Nine weeks of building, plus two mid-point reviews and optional weekly office hours. Teams set their own pace, but the tool has to work by the submission deadline.

Yes, but you need to be able to explain, test, and secure what you submit, and disclose material use.

Your code is published under a permissive OSI-compliant license. Apache 2.0 is the default; MIT and BSD are approved alternatives. Ownership stays with you. A permissive license grants others rights to use your code, and that’s separate from who owns it. A separate ownership-transfer option applies only if you accept a prize.

Weekly mentor-led office hours run throughout the build period, along with two structured mid-point reviews that give teams focused feedback on progress, methodology, and policy relevance. Mentors come from economics, planning, real estate finance, and civic technology. They advise, but they do not build your submission.

You also get a starter kit: a pre-configured template repository, a reference example, document templates, and a curated dataset list.

The prize pool is $38,000: $20,000 for first place, $12,000 for second, $6,000 for third. Judging runs in two virtual rounds. Every submission that clears the eligibility gates is scored against the rubric, then the top five present live to the judges.

Scoring weights transparency and explainability most heavily, followed by replicability, usability, methodological soundness, and policy insight.

One thing worth knowing: applying your methodology to states with less complete data counts in your favor. Thin data is a real barrier for policymakers, so teams that show their approach still works where the data is sparse are judged more favorably than teams that model only data-rich states.

Yes. Partnership is in-kind. Partners amplify the Challenge across their networks, provide mentors, open up data, or supply infrastructure like cloud credits and API access. Write to us and we’ll set up a short conversation about the right fit.