NOEMA01 / THE ECONOMIC INTELLIGENCE LABORATORY
Intelligence must
earn its resources.
Observe the world. Challenge an idea. Measure what survives.
NOEMA
is building an ecosystem of bounded researchers that earns more
attention through evidence.
02 / METHOD
A decision is only the beginning.
Every stage leaves a record. Outcomes arrive later. Complexity has to outperform a simpler explanation—and the choice to do nothing.
03 / RESEARCH ECOLOGY
Attention is a
scarce resource.
Specialists should compete with baselines, challengers, and abstention. Research budgets can evolve without granting access to funds.
Specialist evaluation contract ↗Policy explanation. No active agents or performance are being represented on this page.
04 / ECONOMIC TRUTH
Know what the numbers mean.
Funding is not revenue. Paper profit is not cash. A reservation is not an invoice. Import a report from your own runtime to inspect each category separately.
Generate a report from NOEMA
noema economics-report > economics-report.jsonRun this in your installed NOEMA environment, then select the JSON file above. The importer keeps values in memory only. A report is a snapshot, not a live connection. Inspect sensitive information before sharing the file.
What the report cannot establish
Owner funding, initial capital, deposits, and transfers: these are funding movements, not earned revenue. Operator receipt classifications are not independently reconciled to exclude them.
Realized returns: not established by paper settlements. Unrealized value: not realized profit. Internal allocation: a resource policy, not an audited balance.
Provider costs: reservations and estimates are not settled invoices. Full economic profit needs complete, attributed costs and reconciled cash.
05 / EVIDENCE BEFORE AUTHORITY
Maturity is earned.
Authority is granted.
Paper evidence, better calibration, or a successful deployment cannot unlock financial execution on their own.
01 Observe and establish baselines
Collect timestamped public observations. Preserve missing data. Record market consensus and a PASS baseline before claiming independent skill.
02 Test on later outcomes
Count distinct resolved events, include costs, and evaluate calibration out of sample. Repeatedly scoring the same evidence does not create new evidence.
03 Allocate bounded research
Only new evidence can support promotion. Failed calls retain reservations. A weak specialist can lose attention or be quarantined.
04 Separate authorization gate Locked here
Live-value execution needs explicit authorization, a supported venue adapter, deterministic risk limits, secret isolation, reconciliation, and recovery. This public site provides none of that authority.
06 / TRENCH-1
Survival before selection.
A research track for early token survival. Its job is to test whether information available at the decision time distinguishes later outcomes.
-
t₀ / OBSERVATION
Freeze what was knowable
One earliest observation per token. Preserve the information boundary.
-
t₁ / CONTROL
Match the experiment
Compare contemporaneous candidates and controls, not a convenient retrospective sample.
-
t₂ / OUTCOME
Wait for the label
Chronological outcomes, cohort integrity, and walk-forward evidence precede any claim.
Research capability, not a token recommendation or demonstrated trading edge. Inspect Trench-1 ↗
07 / SYSTEM BOUNDARIES
One ecosystem. Distinct responsibilities.
Select a component to inspect what it can do, where it runs, and where its authority ends.
08 / BEGIN WITH EVIDENCE
Run one honest cycle.
The current worker supports public-market paper research. Credentials and paid models are optional for that first cycle.
python -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"
noema agent-once
noema doctor
noema ladderPython 3.12+ · public API network access required · run from a repository checkout
Today: local SQLite records, a paper worker, optional cognition, and portable Meridian review packets. Multi-venue execution, provider reconciliation, and a hosted cross-device investigation service remain unfinished.