Week 2 update: Why the board changed

When we published Week 1, we were upfront that the board was built on judgment. Experienced judgment — but judgment nonetheless. The goal was always to make the methodology more rigorous over time, and that's what this week represents.

Between Week 1 and Week 2, we built a quantitative scoring system and ran all 52 metros through it. The board looks different as a result. But the more important change isn't the rankings — it's that the framework now makes predictions we can actually test.

Here's the honest difference between Week 1 and Week 2.

Week 1: "Columbus feels hot. Sherman feels promising." There's no way to validate that. It's either right or wrong in retrospect, but there's no mechanism to learn from it systematically.

Week 2: "Our framework predicts Sherman will see 60–70bps of cap rate compression within six months. Columbus will see 0–10bps." That we can test. In six months, we pull the data. If Sherman compresses as predicted, the framework holds. If it doesn't, the framework needs adjustment. Either outcome is useful.

This shift — from judgment to hypothesis — is the most meaningful thing we've built. It means the methodology improves over time rather than just accumulating conviction.

Every metro is now scored on two dimensions.

Hero measures catalyst quality — what the investment is, how much capital is committed, how many jobs it creates, and whether it's the type of catalyst that actually moves housing markets. Not all investments are equal here. A pharmaceutical or semiconductor fab generates a concentrated, high-wage workforce that has to live somewhere. A data center generates 200 jobs and warehouses electricity. Same headline dollar figure, but very different effect on a local housing market. We excluded more than $50B in federal catalyst projects because the underlying wage fundamentals don't support repricing.

Progress measures timing — how far along is execution, and when do workers actually arrive? A project that breaks ground in 2026 and opens in 2030 has a very different investment profile than one where workers are showing up today.

The combination is what drives tier placement. A high Hero score alone is not enough.

Columbus was a NOW metro in Week 1, and it deserved consideration. Hero score of 8.0 — the catalyst is real. But Progress came back at 0.05. The investment has been operational for more than 18 months. The market has already absorbed most of the repricing, and the delta for new capital has compressed significantly.

Qualitative frameworks struggle to catch this because the story still sounds compelling. Until this week, it didn’t account for the delay in construction at the Intel hero catalyst (for years). The framework now makes the timing problem visible.

Sherman, TX is the mirror image. Qualitatively it can look like an afterthought — a small Texas city, not a marquee market. Scoring tells a different story: Hero 8.4, Progress 1.0. Workers are arriving now into a city of 45,000 that has never had a semiconductor workforce. There is no existing housing supply calibrated to that income band.

Indianapolis is a compounding story, where each development raises the cost of Lilly not following through.

Last week added the latest layer. On February 26th, Lilly cut the ribbon on LillyPod — the most powerful AI supercomputer owned by any pharmaceutical company, built with NVIDIA, running 1,016 Blackwell Ultra GPUs at over 9,000 petaflops. It is live now in Indianapolis.

Here's why it matters to the framework. LillyPod's job is to compress drug discovery timelines — Lilly's executives have publicly stated the goal is cutting the traditional 10-year cycle in half. The Lebanon API campus was built to manufacture whatever comes out of that pipeline. A faster pipeline means a faster hiring ramp. LillyPod is an accelerant on the Progress clock for Brownsburg and Lebanon.

The framework already scored Indianapolis well on Hero. What LillyPod does is compress the uncertainty band around Progress — and that's what moves a market from compelling to now. Lilly currently has 329 open roles posted across Indiana, $13.5B committed to Lebanon, and the world's most powerful pharma supercomputer running two miles from their corporate headquarters. Companies don't build that stack and walk away.

West Lafayette and Kansas City slide from NOW to NEXT. Nothing deteriorated. Both remain strong, well-supported markets with credible catalysts. But the scoring was honest about their execution arc, and both carry specific predictions:

West Lafayette: 40–50bps compression when SK hynix workers arrive, anticipated Q2 2026. Kansas City: 30–40bps compression around Q1 2027 as the Panasonic ramp matures.

When their Progress scores move, so will their placement.

We're tracking three things going forward.

First, forward predictions — the specific compression estimates above will be measured at 6 and 12 months against actual cap rate data.

Second, difference-in-differences studies — comparing metros like Indianapolis against similar cities that didn't receive major catalyst investment, to isolate the catalyst effect from broader market conditions.

Third, backtesting — running the framework against historical cases to see whether it would have identified past repricing events before they happened.

In 12 months, we'll publish what the framework got right and what it missed. The goal isn't to be right — it's to build something that gets more accurate over time.

The practical value is clarity on timing. A market can have everything going for it fundamentally and still be the wrong entry point today, because the repricing has already occurred or hasn't started yet. The framework tries to make that distinction explicit rather than leaving it to feel.

We're also being transparent about uncertainty. These are predictions, not guarantees. The difference is that predictions can be tested, refined, and trusted more over time. Judgment can only be trusted or not.

Week 1 pointed toward the right markets. Week 2 tries to be precise about the right moment — and commits to showing its work.

Scoring universe: 52 U.S. metros with qualifying catalyst activity as of March 1, 2026 Hero × Progress framework; tier boundaries: NOW 7.5–10.0 / NEXT 5.5–7.4 / LATER 3.5–5.4 / FUNNEL <3.5 Compression predictions are model outputs, not investment advice. All predictions will be tracked and published.