Loading…
Loading…
Every number elsewhere on MacroLens traces to a primary public source. This is where we hold ourselves to the same bar: an auditable, continuously-scored record of how our calls have actually done — wins and losses counted the same way. It is the corpus a skeptical metals PM or DORA officer reads before trusting anything else.
In grown-up terms: calibration is measured against free public prediction markets (a benchmark, kept in its own track); the lead-time ledger counts every primary-source-vs-press pair including the ones we lost; and every empirical claim ships with its effect size, its p-value, and the Python that reproduces it. Nothing here is a demo.
We score our own axis — tariff, sanctions, export-control — against free public prediction markets, then check whether what we said lines up with what happened.
A measured track record, not an assertion. The market price is the benchmark, kept in its own alternative-signal track and never merged into the official register score.
Open →Each forward call is logged on a dated, auditable line while the measure is still not law — so the flagged→enacted window becomes a permanent, checkable lead time.
Counted honestly, wins and losses: the retroactive primary-source-vs-press ledger currently reads 5 led · 43 tied · 8 lagged across 56 pairs.
Open →Most trace a single register action end-to-end; a few follow one policy regime through three or more — the qualitative transmission map from a statecraft action to the named companies it touches.
Plus 20 method pressure-tests on commodities outside the scored universe. Every action links to its primary public source; no case rests on an internal or subscription analyst report.
Open →Event studies on policy-shock cohorts, each with an effect size, a p-value, and a re-runnable Python audit — the empirical layer behind the register.
The honest finding is reported whether or not it favours the thesis; nulls are kept, not dropped.
Open →Free public sources only. These surfaces accrue over calendar time — the calibration board and the lead-time ledger earn their record forward, never backfilled to flatter the score.