−42%
LENS 01
Fallacies & Distortions
Catches loaded wording, cherry-picked numbers and leaps in logic — then discounts that evidence.
Pre-launch · Now tracking selected 2026 midterm races
- 1Paste a market link.
- 2We gather balanced news and run five lenses.
- 3Compare our read with the market.
Before you call it,
test the hype.
Paste any Polymarket or Kalshi question. Five separate lenses read the news on both sides, show you where they agree — and how far the market price has drifted from the evidence.
Tests the YES and the NO, every timeWe gather the news — you don't cherry-pick itEvery source shown
● SAMPLE TEST SET
Senate control · 2026 +6 ptsHouse control · 2026 −3 ptsFed cut · December +11 ptsShutdown before deadline −2 pts
The five lenses
One question. Five ways of seeing it. One honest reading.
Run one lens or all five. When they agree, you can examine the shared reasoning. When they split, that's exactly what you need to know before you commit.
YES WORLDNO WORLD
Which story needs fewer miracles?
LENS 02
Backward Story
Tells the story backward from a YES world and a NO world. Which story needs fewer miracles?
BUZZFACTS
LENS 03
Hype vs. Function
Weighs the buzz against what has actually been done, said on record, or delivered.
LENS 04
Trajectory & Sensitivity
Reads the price path and how easily it could tip before the end date — like a near-term weather forecast.
“expected” ↑“on track” ↑“stalled” ↓tone: calmer
LENS 05
Language Pre-Signals
Watches how key words and tone shift week to week — often before the price moves.
Separate readings, not independent studies. The lenses share evidence and can share blind spots.
Why five lenses, not one answer
Ask an AI once and you get a guess. Ask it five different ways and you get a picture.
Forecasting research keeps finding the same thing: a single AI prompt tends to be noisy and overconfident, while combining several different ways of reasoning comes out steadier — in recent tests, close to strong human forecasters on sites like Metaculus.
Hype Test is built on that idea. It never asks "what will happen?" once. It runs five separate lenses, each with its own job, then shows you where they agree and where they split.
Read: "The AI Superforecasters Are Here" (Astral Codex Ten) ↗- ONE PROMPT
One answer, often too sure of itself, easy to nudge.
- FIVE LENSES
Five separate readings, each checking a different weak spot.
- THE SPREAD
Disagreement is shown, not hidden — it tells you how settled the question really is.
- An information tool, not a prediction guarantee or betting advice.
How a Hype Test works
- 01
Paste a market
A Polymarket or Kalshi link, or any yes/no question you're curious about.
- 02
We gather both sides
Recent news for YES and NO, plus the official rules and price history where available.
- 03
Five lenses weigh in
Each lens gives its own estimate and shows its reasons, so nothing is a black box.
- 04
See the hype gap
Where the lenses agree, where they split, and how far the price is from the evidence.
Enter a question or review test details
Both sides, every time
No lens sees only the YES.
Sources you can check
News excerpts link to where they came from.
Independent
Not affiliated with Polymarket or Kalshi.
Built for learning
You come out sharper even when a call misses.
Pre-launch · Midterms 2026
Put the contested midterm races through Hype Test before Election Day.
See where the five lenses agree, examine the evidence, and keep your readouts in My Tests.
Midterm readout updates
Email sign-up coming soon
Try a midterm market No account needed to run your own test.