Forecasting Methodology — How Nowpattern Builds Forecasts
Nowpattern
Forecasting Methodology — How Nowpattern Builds Forecasts
A reader-facing methodology page that separates scoring, resolution, and integrity.
1121Registered forecasts58Resolved54Publicly scored32.1%Brier Index
Why a forecasting platform?
Nowpattern is designed to keep the question, probability, resolution rule, and public result in one visible system. The public tracker currently carries 1121 registered forecasts, of which 58 are resolved.
The honest KPI here is not total predictions. It is the size and quality of the publicly scored sample.
The five requirements for a valid forecast
- A resolution question that an outside reader can judge later
- A locked initial probability that is not swapped after the fact
- A visible deadline and external resolution source
- Hits and misses both remain on the public surface
- Non-scorable cases are excluded with an explicit reason
Methodology by category
Geopolitics emphasizes regime incentives and market reaction, economics emphasizes policy transmission, and technology emphasizes shipping, regulation, and adoption speed. Across categories, the judgment question comes first and probability comes second.
How the loop improves
The public score is an EvolutionLoop input. We watch where low-probability hits and high-confidence misses cluster, then adjust calibration by category. The current public Brier Index is 32.1%, from a raw Brier average of 0.4608.
How to compare us externally
Compared with mature forecasting communities, Nowpattern is still in the corpus-building phase. That is why the important moat number is not total forecasts, but the 54 publicly scored cases and how that sample grows.