Skip to main content
Relevance is the core unit of measurement on Noise. Every market has a Relevance Index — a composite metric that quantifies how relevant a trend is right now. It is the number displayed on each market’s trade page.
Relevance Index values are displayed with the ° brand symbol throughout the product — for example, 581.50°. This is a Noise convention to denote relevance units.

​
What Relevance Measures

Relevance captures the real-time intensity of activity around a trend by aggregating signals from across the internet. It reflects how much people are posting about, engaging with, and actively interested in a given subject at any moment. The Relevance Index rises when a trend gains momentum — more posts, more engagement, more activity across sources. It falls when activity subsides. Relevance is continuous and always updating.

​
Data Sources

Each market is configured with data sources across two categories.

​
Content Sources

Content sources track social and media activity across platforms including X (Twitter), Reddit, YouTube, Instagram, Substack, and RSS news feeds. From each source, three metrics are measured:
  • Engagement velocity — a weighted sum of interactions (views, likes, shares, replies, quotes, bookmarks), tuned per platform to reflect signal quality
  • Post count — total volume of content mentioning the trend
  • Unique authors — number of distinct accounts posting, which helps distinguish genuine interest from concentrated or automated activity

​
Signal Sources

Signal sources capture activity from prediction markets including Polymarket and Kalshi:
  • Volume — dollar volume of prediction market activity related to the trend
  • Market count — number of active prediction markets, indicating breadth of informed interest

​
How Relevance is Computed

Raw metrics from each source are smoothed using exponential moving averages (EMAs) to filter out transient noise and prevent short-lived spikes from distorting the index. The smoothing is configured per market with a half-life parameter — the time it takes for a past observation to decay to half its influence. The smoothed values across all sources and metrics are then combined into a single composite value through a weighted aggregation. Each source and metric type has a configurable weight that reflects its relative signal quality. The result is the Relevance Index — the absolute value displayed on the trade page.

​
From Relevance to Oracle Price

The Relevance Index is not the oracle price. The oracle price is derived from relevance data but also incorporates trading activity — see below.
Two transformations occur between the Relevance Index and the oracle price:
  1. Normalization — the Relevance Index is normalized against its own historical average to produce the Attention Factor, centered around 1.0. This makes the signal comparable across markets regardless of their absolute activity levels. An Attention Factor above 1.0 means the trend is receiving above-average relevance; below 1.0, below-average.
  2. Oracle composition — the oracle price combines the Attention Factor with normalized Noise trading volume. This means the oracle reflects both external relevance data and actual trading activity on the platform. The oracle price is updated every 1–3 seconds.
The oracle price is the reference used for funding rate calculations. Market prices themselves are always determined by traders on the order book — the oracle creates a gravitational pull via the funding mechanism, not a hard peg.