What this is. Decision support, not a signal service. Nine pre-registered tests
found no automated prediction edge in news — and one of them measured why:
~60% of the price move around conflict news happens before the news exists.
So this hands you the measured base rate, whether the market has already moved, the
individual historical cases, and the questions only you can answer.
The default answer is no trade.
Loading…
What this ranks. Stories alive in one information sphere and silent in another,
corroborated by independent sources or physical sensors, whose mapped assets have
not yet moved. Rejected stories are shown too, with the reason — the rejections are
how you judge whether to trust the scorer.
Loading…
Everything that needs your attention, in one place. New signals since you last
looked are marked. No Telegram, no email — the dashboard is the alert channel.
Loading…
Founders' Circle access. Members sign in at /member with the email and
password you give them. They see the monthly portfolio plan, the markets, the world news that matters and
how to read it — never sources, controls, signals or probabilities.
Open the member view ↗
·
Add member
Members
Loading…
This is where the probability scale comes from. Every event class × asset × horizon is
measured against ten years of history and — crucially — against its own base rate.
Overlapping windows are removed, same-day duplicates collapsed, and any result concentrated
in a single year is rejected. Most pairings show no edge. That is the honest answer.
Loading…
Who gets told, and how. Signals broadcast to WhatsApp and email. Each person has
their own conviction threshold, so someone can receive only the 85+ calls while you take
everything. Every message carries the full execution plan.
Add recipient
Recipients
Loading…
Delivery log
Every send is recorded per person per channel — a re-run can never double-message.
This is where the system is tested against reality, including when it fails.
Every case runs on data that existed at the time, and every backtest is paired with a
CONTROL — the identical rule applied to non-event dates. A backtest without a control is
marketing, not evidence.
Loading…
The reverse pipeline. Everything else here runs news → asset. This runs
asset → why. Every mapped asset is checked for statistically unusual price and
volume, and each anomaly is handed to a search across all 180+ sources in 17 languages.
An anomaly nothing explains is the strongest signal the system can produce — price
is carrying information the public record does not.
Loading…
Capital movement
Insider filings, activist stakes and federal awards — who is putting money at risk.
Evidence, not commentary. This records what a market-moving official
said or signed, from a primary source with a timestamp, and measures what the
mapped assets did afterwards. It does not assert motive. Where an asset moved abnormally
before publication, that is shown as a timing observation with dates attached —
anticipation has many innocent explanations, and the data cannot say who knew what.
Loading…
Loading…
Forward record only. Every row begins the moment a signal fires and is marked to market
thereafter. No backtest, no simulation, no backfilled entries. Small samples are labelled as
descriptive, not as evidence of edge.
Loading…
Loading…
Loading…
Prediction-market odds that moved sharply. When these reprice before matching news
volume appears, money is moving on information that has not been published yet.
Loading…
Blind spots matter more than averages. A whole language going dark is the failure mode
that silently kills the edge — "we saw nothing from Iran" must be distinguishable from
"the Iranian feed has been failing for a week".
Loading…
Loading…