Observatory
Hand-picked highlights from the Observatory corpus, organized by domain. A curated subset, not the full programme.
A Standing Validation Apparatus
The Observatory is not a forecast and not a portfolio. It is an apparatus. It maintains roughly 1594 candidate signals across 69 mechanism clusters and subjects every one of them to a fixed gauntlet: the four-battery validation framework — Associative (does the relationship exist), Confound (does it survive controls), Stability (does it hold across eras and regimes), Mechanism (is there a plausible causal substrate) — the Whewell consilience rubric, and a separate devil's-advocate review whose explicit job is to kill the signal if it can. A signal is retained only while it stays on the confirmed side of that process — and it is re-tested, not grandfathered.
The aggregate state of the corpus — verdict distribution, battery completion, validation-depth tiers — is published live on the Observatory Dashboard. Counts above are drawn from the per-signal dossiers as of the most recent site build. For where the apparatus came from — the first week of the validation programme running against live data — see the Observatory Week One Retrospective.
What the Apparatus Has Found
A consilience is a finding where several independent lines of evidence — different data sources, different research traditions, different centuries — converge on the same structural claim. These are a selection of the strongest cross-disciplinary findings to clear the full protocol. None of them is investment guidance; each is a structural claim about a mechanism. The full curated tour is on the Featured Signals page.
Cosmic ray flux and agricultural yield
Galactic cosmic ray flux modulates cloud nucleation rates — a mechanism demonstrated at the CERN CLOUD experiment — with downstream effects on regional cloud cover and crop yield. The fingerprint is visible across three independent long records: 1,200 years of Kyoto cherry-blossom phenology, 177 years of Broadbalk wheat, and 1,300 years of Nile nilometer data. One honest caveat: the upstream physics is consilience-grade, but the specific cosmic-ray-to-corn-price link is underpowered — it failed our April 2026 retest and sits on the Killed Signals page.
Electromagnetic frequency biology
Across at least seven independent research traditions — orthopaedic PEMF, photobiomodulation, far-UVC pathogen control, focused-ultrasound lithotripsy, GHz membrane resonance and others — specific electromagnetic frequencies activate specific molecular targets. As an umbrella principle this is the Observatory's tightest single consilience; individual frequency-target pairs are reviewed signal by signal.
Financial stress cycle hierarchy
Across five financial-stress series, the Juglar band (~7-11 years) carries roughly three times the spectral power of the Kitchin band (~3-5 years) — Bonferroni-significant in four to five of the five series. Medium-term credit cycles dominate short-term inventory cycles in financial-stress dynamics.
Housing starts Juglar cycle
US housing starts show a dominant ~7-year spectral peak in the Juglar band — not the 18-year Kuznets cycle popularly associated with real estate. False-alarm probability 0.007 against phase-randomised surrogates.
Bank of England 800-year CPI cycle
The Bank of England's millennium-scale CPI series carries a long cycle near 55.75 years, with phase-randomised surrogate FAP = 0.006. It is the narrow survivor of a battery that killed the broader Kondratiev wave on the same dataset — a tighter version of the same intuition.
Traditional ecological knowledge
Six indigenous knowledge traditions with no shared citation graph converge on specific, testable claims — land-equivalent ratios in intercropping, frost-buffer effects of canopy retention, Snell's-law-consistent wave navigation. Four confirmed, one weak, one killed. The batch survives as a consilience; the population-level "ancient knowledge outperforms" claim does not, and is published in the kill list.
AMOC structural weakening
Atlantic Meridional Overturning Circulation weakening is confirmed through GRACE-FO ocean-bottom-pressure trends and corroborated by multiple independent measurement chains. The question is rate, not direction. Tracked live as a causal chain below.
Live Causal Chain Monitor
Two structural chains with validator-confirmed underlying signals are tracked live against public observable thresholds. Each moves on its own dynamics.
The BIS credit-to-GDP gap is the most reliable medium-term banking-crisis precursor in the literature. At -12.1% the corpus is in the deleveraging aftermath of the GFC; crisis-threshold risk is structurally unavailable until the gap turns positive — est. 11 yrs away at current recovery rate.
Validator-backed by the confirmed Minsky / BIS credit-gap signal. Structural research only — not investment advice.
GRACE-FO ocean bottom pressure declining. Physical AMOC weakening confirmed. GRACE-FO data period: 2018-06 to 2026-01. Physical weakening is the confirmed precursor; the sea-level insurance trigger is not yet reached.
Benchmark Manipulation Clock
An independent mechanism chain: the regulatory-enforcement cycle by which captured financial benchmarks move from suspicion to prosecution. LIBOR (2012), the FX 4pm fix (2013) and the precious-metals Gold Fix (2015) all traced the same arc.
COLD — no current enforcement signals, transaction-based design provides genuine resistance
WARM — NM SIC lawsuit standing established, ESRB regulatory concern building
HOT — Operation Token Mirrors indictments March 2026, CFTC Q1 2026 priority
Next monitoring: 2026-07-01. Trigger: Any DOJ/CFTC filing using word 'benchmark' in crypto context — immediate upgrade to ACTIVE_PROSECUTION
Tracked as a documentary/historical pattern: the enforcement cadence is the claim, not a trading signal. The Clock describes when regulatory enforcement has historically arrived — not how any instrument will trade around it.
The Discipline That Makes the Findings Credible
The Observatory has retired 289 of 1594 signals — a corpus-wide kill rate of 18.1%. The trading-signal battery applies additional criteria and produces a higher rate. Neither number is a problem; both are the point. A corpus that publishes only its confirmations is producing the replication crisis, not escaping it. The kills are published in full.
Moon-phase equity returns, the Super Bowl Indicator, Schumann-resonance-to-anything, the Mayan tzolkin equity cycle, COT 10-year Treasury positioning — all tested, all killed under phase-randomised surrogate testing or era-split replication, all documented openly. Even an internally favoured framework was retired: the four-chain convergence score was withdrawn in April 2026 once two of its four chains lost their underlying validated signals to noise verdicts. A score where half the legs are null is not a score. The two chains whose causal signals still stand are the ones monitored above.
This is the trade the apparatus makes: the kill rate buys credibility for the confirmations. The confirmed signals are credible because 289 others did not make it.
For a different register on the same corpus, the Phronopolis Lab publishes persona reflections on Observatory findings — historical voices thinking aloud about individual validated signals.
This is structural research, not a timing signal. The mechanism, not the outcome, is the claim. Why the per-signal portfolio is not public →