Logged before launch, scored after.

Anyone can publish a forecast. What makes one worth reading is that it was written down before the result was known, with a metric and a time window that decide whether it was right. These are our Wind Tunnel calls: public events, called with the same method we use for clients, sealed before and scored after. Misses stay here next to the hits.

Client work is scored privately, with the client, and nobody outside their team sees it. The card shows how a private call looks in their portal.

Bring us a decision

Sample prediction, logged before launchScored against your real numbers after 14 December.

Pending

MetricHomepage clicks

PopulationNew visitors, Romania

Predicted+8% to +15%

AgainstHero A, today’s homepage

Hero A+8%+15%

Logged 4 Nov, 09:04, before launchWindow 1–14 December · confidence medium-high

Wind Tunnel calls

Calls on public events, posted when sealed and scored when the window closes. The record starts at WT-02: WT-01 was made with an earlier version of the method, so its score would say nothing about the calls we make now.
Call	Event	Prediction	Sealed	Window	Status
WT-02 goes up here once it is sealed, before the event it calls.

Why every score feeds Thymos

Scored outcomes go back into Thymos as calibration for that market: every Wind Tunnel call, and client calls where the contract allows it. We will say the scores improve our predictions once the scores themselves show it.

Delivery time gets the same treatment. Median time to a recommendation, by scope, will be published here as engagements are timed.