Skip to content
Worldview
Get started

News Feed

Article feed with AI relevance scores, sentiment icons, and news routing tiers.

On this page (6)

The News tab on each instrument page shows a curated article feed for that instrument. Articles are ranked by an AI relevance model that scores them against the specific instrument's entity context — not just keyword matching.

Article card anatomy

Each article card shows:

  • Headline — linked to the original source
  • Source and publication time (relative, e.g. "3 hours ago")
  • Relevance score — 0–100, shown as a colored bar
  • Sentiment icon — positive, negative, or neutral
  • Routing tier badge
  • Entity tags — up to 3 extracted named entities

Relevance scoring

Relevance scores are computed by a two-stage model:

  1. Market routing score — based on entity co-occurrence and relationship distance in the knowledge graph (e.g., a direct mention scores higher than a supplier mention)
  2. LLM relevance scoremeta-llama/Meta-Llama-3.1-8B-Instruct scores the article title against the entity's description (0–1 scale)
  3. Display score — a weighted blend: 0.5 × market + 0.4 × llm + 0.1 × routing_tier_weight

Score interpretation

A score of 80+ means the article is highly specific to this instrument. Scores 50–80 indicate relevant context (sector news, related company). Scores below 50 are broad market articles where this instrument was a passing mention.

Sentiment icons

IconMeaning
Green upward arrowPositive sentiment — bullish framing, beat expectations, upgrade
Red downward arrowNegative sentiment — bearish framing, miss, downgrade, legal issue
Gray dashNeutral — factual update, routine announcement

Sentiment is assigned by the NLP pipeline's named entity sentiment model, scoped to mentions of this specific entity within the article — not the overall article tone.

Routing tiers

Articles are assigned to one of four tiers based on source quality and entity directness:

TierDescription
TIER_1Primary source (SEC filing, Fed statement, press release)
TIER_2Tier-1 wire service (Reuters, Bloomberg, AP)
TIER_3Financial media (WSJ, FT, MarketWatch, Seeking Alpha)
TIER_4Aggregated / social signals (Reddit, Twitter-firehose)

Tier is used as a signal weight in the screener's "news velocity" filter and in the morning brief's source selection.

Article impact windows

For articles published during or just after a trading session, Worldview tracks the instrument's price return in four windows:

  • T+0 — same-day close vs. open
  • T+1 — next-day close vs. publish-day close
  • T+2 — two-day close
  • T+5 — five-day close

These windows appear as a small sparkline on article cards that are at least 5 trading days old, showing how the stock moved after the article was published.

Filtering

Use the filter bar above the article list to narrow by:

  • Sentiment — All / Positive / Negative / Neutral
  • Tier — All / Tier 1 / Tier 1+2 / All tiers
  • Date range — Last 24h / 7d / 30d / Custom
  • Minimum relevance — slide to set a score floor

Was this page helpful?

Last updated