Methodology & sources

What the measures mean

Prediction markets express expectations through contract prices. Predpulse makes their movement, policy outcomes and trading activity easier to inspect. Each measure answers a different question.

Explore the indices →

Three ways to read the market

01

Belief Shift

How much have expectations moved?

The average absolute 24-hour change in YES midpoints, measured in percentage points across a pinned sample of Economics or Politics events.

A 24-hour score needs comparable saved quotes. It is a measure of movement, not a category probability.

02

Fed policy balance

Which way does policy lean?

The normalized share of rate-hike outcomes minus the share of rate-cut outcomes for one supported Federal Reserve meeting. The outcome distribution shows the balance behind the score.

Meeting-specific, with the meeting date displayed. The score is not an expected interest rate change.

03

Market Attention

Where is trading activity concentrated?

A map of seven categories, sized by their share of reported 24-hour dollar volume in the bounded Polymarket event sample. Event families are counted once.

Sampled trading activity, not the whole market, investor sentiment or capital inflows. The map can display on its first capture.

Captured, not live

Publication is scheduled hourly and may be delayed. Every product displays its capture time. Missing quotes stay missing, and comparisons are withheld when contract identity, rules or timing are not comparable.

Belief Shift and Fed policy balance use two-sided YES bid/ask midpoints from Polymarket. A move from 40% to 43% is 3 percentage points. These indices are analytical measurements, not investable funds.

Sources & evidence

The dashboard draws on Polymarket, Kalshi and Manifold. The three index products currently use Polymarket. Venue units differ: dollar volume and Kalshi contract counts are not combined into one volume measure.

Open an index to inspect its captured evidence. The research view provides saved contract observations and history; it does not establish forecast accuracy.

Inspect evidence & history →

Predpulse is a beta prototype. Coverage is sampled and data may be incomplete or delayed. Not financial advice. Not affiliated with the source venues.