AR
Institutional Market Intelligence

The research desk
hedge funds run.
Now yours.

A proprietary real-time event detection engine and AI research agents — continuously reading filings, transcripts, and regulatory language across global markets. You see the view before it becomes price.

Engine Active · Monitoring 41 Lanes
Last Signal: AI Infrastructure · 14 Jun 2026
Scoreboard · 23 Theses Tracked
The Engine

We do not publish on a schedule. We publish when the engine surfaces a signal that clears validation.

01 — DETECTION
Continuous event detection

A proprietary algorithm monitors SEC filings, earnings transcripts, regulatory language, central bank communications, and global news flow without interruption. It is built to identify the pattern forming, not report the headline that follows.

EDGAR Transcripts FERC GDELT
02 — VALIDATION
Multi-agent adversarial review

Every detected signal passes through three independent agent layers. A skeptic argues the signal is noise. A corroboration layer verifies it across independent sources. A consistency layer checks it against contradicting evidence. Only signals that survive all three reach a subscriber.

Skeptic Corroboration Consistency
03 — PUBLICATION
Structured narrative with falsification

The output is a market narrative with its evidence trail, the conditions that would confirm it, and an explicit pre-committed statement of what would prove it wrong. Every thesis is graded publicly on the scoreboard when it resolves.

Thesis Evidence Scoreboard
Sample Brief

This is what a research brief looks like.

Every brief follows the same structure: sector tag, signal strength, thesis, evidence trail with sources, and pre-committed conditions for confirmation and falsification. The format is the product.

Aegean Research
Issue No. 014 · 14 Jun 2026
AI Infrastructure & Energy — Market Narrative
Signal High Corroborated across 4 independent sources
i.Thesis

The equity market has not yet priced the regulatory language shift accumulating in FERC filings around datacenter power interconnection. Language appearing in Q4 transcripts across hyperscalers and regulated utilities suggests a 12–18 month timeline compression not reflected in current utility equity valuations.

ii.Detected Signals — Evidence Trail
FERC filing frequency of "priority interconnection" language increased 187% versus the Q3 2024 baseline across 14 monitored dockets. Source: FERC EDGAR docket monitor, Jan–Mar 2026
Hyperscaler capex guidance has shifted from "planning" to "committed" across MSFT, AMZN, and GOOGL transcripts — a pattern historically preceding capital deployment acceleration. Source: Q4 2025 earnings transcripts, Aegean corpus
Sentiment divergence detected: news coverage of datacenter power constraints is net-negative, yet VST, CEG, and NRG have not re-rated to reflect implied demand — a 2.4σ divergence from historical correlation. Source: GDELT sentiment index vs. Bloomberg price, 90-day window
↑ Thesis Confirmed If

FERC issues formal priority queue policy within six months; VST or CEG guidance explicitly references accelerated interconnection in next earnings cycle.

↓ Thesis Falsified If

FERC filing language reverts toward neutral across two consecutive quarters; hyperscaler capex guidance is revised downward or deferred.

How We Differ

Aegean Research is not a newsletter.

01 / 04

Event-triggered, not scheduled

Newsletters publish on a fixed cadence regardless of whether they have something worth saying. Aegean Research publishes when the detection engine surfaces a signal that clears validation. If nothing clears the bar in a given week, nothing goes out. The silence is the product working correctly.

02 / 04

Adversarial validation, not opinion

No single analyst's confidence reaches you. Every published view has survived three layers of independent agent scrutiny — skeptic, corroboration, consistency. The architecture is the moat. The output is not an opinion. It is a thesis with an evidence trail.

03 / 04

Pre-committed falsification

Every thesis is published with an explicit statement of what would prove it wrong — written before the view goes out. The conditions are not adjusted afterward. This is standard practice on an institutional research desk. It is almost entirely absent from retail research.

04 / 04

Public scoreboard, honestly graded

Every published thesis is tracked. When it resolves — confirmed or falsified — it is graded against the criteria committed to at publication. Wrong calls are not buried. The track record is the trust mechanism.