Media

Press & media kit

For journalists, researchers, and media partners. Questions? Email press@theprobablenews.com

What Probable is

Probable is a daily forecasting brief that answers one question — “will this happen?” — with a clear probability number and the work shown.

Each weekday we publish a read on the most decision-relevant story in news, politics, or markets: our own probability estimate, synthesized from four input classes — prediction markets (Kalshi, Polymarket, Manifold), professional analysts (Wall Street research, Fed surveys, Bloomberg consensus), public opinion polls (FiveThirtyEight, Gallup, Pew), and official government data. We run the math ourselves. AI drafts the synthesis. A human reviews every piece before it publishes. We maintain a public scoreboard so readers can see when we are right and when we miss.

The editorial positioning: NPR if it ran on prediction markets. Calm, evidence-first, transparent about uncertainty. Not opinion, not partisan spin. Not financial advice.

Key facts

Founded
2026
Publishing cadence
Every weekday — website, newsletter, YouTube Shorts, Instagram Reels + Feed
Domain
theprobablenews.com
Newsletter platform
Beehiiv (free to subscribe)
YouTube
@theprobablenews
Instagram
@theprobablenews
Primary prediction market cited
Kalshi (CFTC-regulated, US-legal)
Editorial disclosure
AI-drafted from cited sources; human-reviewed before every publish
Cost to reader
Free at launch

The editorial model

How the forecast number is produced

Probable does not borrow a number from any single source. Every forecast is computed by a deterministic algorithm (“forecast engine”) that runs before any AI writing begins. The engine classifies the story type, applies source-class weights appropriate for that type, discounts thin-liquidity markets, and uses log-odds aggregation to combine inputs correctly.

The AI’s job is to explain the number in plain English and cite every source used. It cannot invent a different number. A human reviewer reads the output before it publishes.

Full methodology →

Four source classes

  • Prediction markets — Kalshi, Polymarket, Manifold
  • Professional analysts — Fed, Goldman, Bloomberg surveys
  • Public polls — FiveThirtyEight, Gallup, Pew
  • Official data — BLS, Treasury, Fed releases

Public accountability

Every forecast is tracked on the public scoreboard. When we are wrong, we say so by name. A correction log is published automatically after human review.

Brand assets

Probable.

The Probable wordmark is set in Lora (serif), with the terminal period in ember (#b54213). Page background cream (#fbfaf7), body ink (#1a1a1a), navy (#1e3a5f) for high-probability labels.

High-resolution logo files and brand guidelines are available on request. Email press@theprobablenews.com

Contact

Press inquiries

press@theprobablenews.com

Interviews, fact-checks, media requests

Sponsorships

sponsor@theprobablenews.com

Advertising and partnership inquiries

General

hello@theprobablenews.com

Everything else

Probable is an independent media brand. All forecasts are for informational purposes only. Not financial advice. Methodology · Privacy policy · Terms