Methodology
Read everything. Rank everyone. Price the question.
D is not a sentiment reader and not a chart. It is a pipeline from evidence to a probability, with every step scored against what actually happened.
6,000+
global sources read every day
1,000+
ranked experts scored on past calls
4
domains: geopolitics, tech, macro, AI
The pipeline
Five stages, in order.
This one really is a sequence, so it is numbered. Nothing skips a stage.
Collect
6,000+ sources are read continuously and deduplicated. Every claim carries a date, a source and a first-appearance timestamp, so a story that is repeated eighty times still counts once.
Weigh
1,000+ named experts are scored on how their past calls resolved, by domain. An expert who is well calibrated on Fed policy carries no extra weight on Gulf shipping.
Price
Evidence and weighted judgment resolve into a single probability per question. That number is D's price, and it is published whether or not it differs from the market.
Qualify
A price becomes a Signal only when the gap to the venue clears the edge threshold and a dated, near-term reason exists for the market to close that gap.
Mark
The paper position is marked at the stated exit window, or at resolution if that comes first. The result enters the track record either way.
The loop
Every outcome is evidence about the method that produced it.
- 01Evidence6,000+ sources, deduplicated and dated.
- 02Judgment1,000+ ranked experts, weighted by past accuracy.
- 03ForecastA price on every market D tracks.
- 04OutcomeResolution recorded against the forecast.
- 05Research LabOutcomes studied for what moved and what missed.
- 06FindingsWeights updated. The next forecast starts sharper.