Crene is investment thesis review infrastructure. One live investment thesis is decomposed into assumptions, independent model reads, weekly movement, and rethink conditions. Public examples include thesis maps for yes or no views, factor maps for continuous distributions, and scenario records for multiple coherent futures. Calibration results are published separately for the leakage controlled benchmark and the broader resolved corpus.

How does AI prediction calibration work at Crene?

Calibration applies at the consensus layer, not at the underlying model layer. Crene queries four frontier models in isolation, computes a cross model median consensus, and tracks resolution against a tiered source allowlist. Every resolved event becomes a permanent calibration point. Per tier accuracy, per model Brier scores, and per domain accuracy are computed live and published at https://crene.com/methodology.

What is a Brier score?

A Brier score is the proper scoring rule for probability forecasts on binary outcomes. It is the mean squared error between the forecast probability and the realized outcome (0 or 1). Lower scores indicate better calibration. The no skill baseline is 0.25 (always predicting 0.5); a perfect forecaster scores 0.0. Crene publishes the consensus Brier score and per model Brier scores across the resolved event corpus.

Crene serves investment teams that need a repeatable thesis review process: hedge funds, asset managers, macro PMs, CIO offices, family offices, and thematic investment teams. A team brings one thesis it is actively debating. Crene maps what the thesis depends on, where models disagree, what changed each week, and what would force a review. Data and API access remain available as the proof layer behind the workflow.

What is the difference between an event, a thesis map, a factor map, and a scenario at Crene?

Events are base binary outcomes with scalar probabilities. Thesis maps decompose yes or no investment views into assumptions. Factor maps decompose continuous variables into driver distributions. Scenario records map long horizon questions into coherent pathways. Public collection pages distinguish live maps from frozen thesis records and archived scenario records.

68% | Will shelter CPI remain above 4.0% year over year in any H2 2026 release?

Reasoning

Shelter CPI (roughly 40% of core CPI) has proven sticky well above 4.0% year over year throughout 2025 and into mid 2026, with no clear trend toward rapid disinflation. The Fed's pivot from cutting bias (reflected in the June 17, 2026 hold and the six month upward revision of the year end 2026 funds rate from 3.4% to 3.8% in the SEP) signals persistent inflation concerns that would only materialize if shelter inflation had substantially cooled. H2 2026 releases cover data from June through December 2026. Given the typical 12 month lag in shelter CPI momentum (lease agreements, owner equivalent rent, and contract renegotiation cycles), the June 2026 SEP's continued elevated rate projection suggests policymakers expect inflation stickiness through year end. Historical precedent shows shelter CPI declines from elevated levels are gradual, typically requiring sustained policy restrictiveness or demand destruction. The 3.50 to 3.75% target range unchanged since mid June combined with nine of eighteen participants projecting rates above this range indicates the Committee retains substantial hawkish tilt, which would only persist if shelter deflation was not yet evident.

Key uncertainty

The pace of owner equivalent rent deceleration in the months immediately ahead of H2 2026 releases. If lease growth and renegotiation rates drop sharply between August and October 2026, shelter CPI could fall below 4.0%, particularly in the December release. Conversely, wage growth and labor market resilience could keep shelter inflation elevated.

Will shelter CPI remain above 4.0% year over year in any H2 2026 release?

What changed, what remains unresolved, and why it matters.

Model disagreement

Belief trajectory

Will shelter CPI remain above 4.0% year over year in any H2 2026 release?

What changed, what remains unresolved, and why it matters.

Model disagreement

Belief trajectory