About this preview

What this is

An advance preview of MarketBrief with stock-ticker tagging and market-impact analysis on every story. Each story gets the affected stocks with a direction, an expected move size, and a why-it-matters read — and then gets checked 1–2 weeks later against follow-up coverage to see whether the call played out.

The data

Scores are model reads on a synthetic news corpus — not validated against real prices. Aftermath verdicts come from corpus follow-ups, not market data.

5,599 stories across 16 weeks (Apr 13 – Jul 30, 2026) from a synthetic news corpus, tagging 711 distinct stocks. 1,647 calls were checked against later coverage. The eventual production version runs the same analysis on MarketBrief's live news feed.

How the model reads the news

The playbook

The who-benefits/who-suffers heuristic, with worked examples per event type.

The full study

Method and findings across all 5,599 calls — including where coverage-based ranking misses movers.

Track record

calls that played out

52.6%

680 of 1293 checked calls with follow-up signal

played out vs fizzled

2.7:1

680 played out · 255 fizzled across checked calls

stories with likely impact

2,206

of 5,599 covered stories

catchable at publish time

95%

115 of 121 big hindsight movers were flaggable before the market reaction

direction agreement

95%

18 of 19 calls where posts quoted a real move for the same instrument

exact size agreement

85%

publish-time size matched the hindsight read across the study sample

What to distrust

The model deliberately over-tags two patterns: war rhetoric and headline funding-round sizes. Missile-order headlines and billion-dollar raises read scarier than they trade — when a story's only claim to impact is a big number or a threat, expect the fade.

Outcomes of the 1,647 checked calls

All 5,599 stories by expected move size

Movers per week — click a column to open that week