StradionSignal matrix
FreeFor every symbol and horizon: the probability of a positive return, the expected return, and a 90% range — with the model and data-date on each row. Percentages only; no redistributed prices.
Open Signals →
A terminal that keeps its own scorecard
Probabilistic forecasts, scored out of sample.
Stradion publishes the probability of an up move, the expected return, and a 90% range for a public set of large-caps, then scores every call once its window closes — against only the data available at the time, misses included. Free, no account.
— forecasts scored out of sample · updated daily
Directional hit rate
—
90% interval coverage
—
Forecasts scored
—
Latest call we missed
—
Scored walk-forward, against only the data available at the time —
including the windows with no edge.
See the full record →
What's here
free unless notedFree, no account. Start anywhere.
Daily news brief
FreeA plain-language read on each watch list every morning, with a stance and price-behaviour note per symbol. Subscribe by RSS — no account.
Read the brief →Track record
FreeEvery matured call, scored. Hit rate and coverage per model and horizon — and “where we were wrong”, ranked. Export the full record as CSV.
See the record →Oikos terminal
By invitationThe full research desk behind the public surfaces: screener, model scorecards, insider/options/sentiment signals, and shared watchlists.
Request access →Pro
$29/mo · soonYour own symbols — a custom matrix and brief — plus signal-change alerts and the full exportable record. Launching soon.
See pricing →Scoring
out of sample, published in fullEvery forecast is scored out of sample.
Graded after its horizon elapses, against only the data available at the time.
Each forecast is graded after its window closes, on the data the model could have used at the time — a walk-forward, out-of-sample replay that includes the stretches where the models do poorly. Our own analysis finds short-horizon directional confidence doesn't yet rank outcomes well enough to trade on; it begins to separate around the 20-day horizon, and that result is stated on the record. A model that is right 57% of the time is wrong the other 43%, and the published numbers reflect it.
Pro — your own symbols, alerts, and the full exportable record — is launching soon.