Signals

Volume changed for two hours running, in markets deep enough to measure

LeagueAllflame

What these are worth knowingWhat these are worth knowingAcross three leagues the detector never saw while it was being built, it calls direction right 53% of the time against a 50% control on random hours — real (z ≈ 2.5–2.8) and small. This page used to headline 66–72%, and that was one league: the same league the direction rule had been chosen on. The rule holds on all three unseen leagues, so the direction is genuine; only the size was an artefact of measuring it where it was strongest. Three points does not pay for the roughly twenty percent it costs to cross a band twice, at any horizon measured, so nothing on this site acts on these — the portfolio page deliberately excludes them from its suggestions.

Direction called right
53%
on 3 leagues the detector never saw
Control
50%
the same measurement on random hours
Signals measured
6,142
out of sample, z ≈ 2.5–2.8

Three points better than a coin. Crossing a band costs about twenty, so no single row here is a trade — this is information about which way a market is leaning, and nothing else on the site acts on it.

Allflame — last 24 hours

One row per market, strongest reading first.

No signals in the last 24 hours.

This is the normal state. The volume detector fires about eleven times a day across roughly 550 markets, and a quiet day is a quiet market rather than a broken pipeline.