Domain Focus — May 2026
Holodeck’s deepest coverage. Twenty-one real-estate archetypes spanning operators, capital markets, renter behavior, and private-market mechanics. Every forecast is logged with a timestamp before its outcome is known — the live Brier, including misses, is on the audit page.
Cross-domain validation
| RE sub-domain | n | Brier | Δ vs baseline |
|---|---|---|---|
| Construction permits | 2 | 0.012 | − 0.177 |
| Commercial cap rates | 4 | 0.030 | − 0.197 |
| Regional submarkets | 4 | 0.051 | − 0.087 |
| Single-family housing | 9 | 0.087 | − 0.058 |
| Multifamily rent & vacancy | 4 | 0.096 | − 0.112 |
| mortgage rates treasury fed | 40 | 0.210 | — |
| REIT performance | 3 | 0.247 | − 0.042 |
| Major news events | 3 | 0.316 | + 0.074 |
Cross-domain holdout: every question is held out and the swarm must predict it cold. Lower Brier = better; 0.25 = a coin flip. n = 69 across 8 RE sub-domains. Updates live as more questions resolve.
Featured questions
What you get
Each result page renders these alongside a live calibration chip (per-domain Brier score), a P10–P90 confidence interval, the raw probability distribution, and the top quotes from agents who took the opposite side.
Sample output · Convergence across 12 archetypes × hundreds of independent samples
Sample output · Archetype cluster breakdown
Archetypes in this swarm
These are sample outputs. See real resolved predictions →
Track Record
Every prediction is logged with a timestamp before its outcome is known, then resolved against the real world — prices, releases, settlements. The audit page shows every one, including the misses. Methodology Brier 0.168 (95% CI [0.080, 0.256]) on 69 time-gated cross-domain RE forecasts across 8 sub-domains. 39% better than the phase-1 baseline (n=111, Brier 0.277).
Every prediction. Every outcome.
Including the ones we got wrong.
Access
Enterprise
Custom
Open access
Free
Professional
$49 / mo