Junctiq
Log InSign Up
Accountable Market Detection

See the market move before consensus.

Don't take our word for it — pressure-test your own bet. Type an idea and watch the engine find every relevant piece of evidence, challenge the premise, and name the one weakness most likely to kill it. No signup. Nothing saved to the record.

Junctiq // Hypothesis Sandbox
Live engine preview · illustrative
0GRADE F

Falsifiability Score

Start typing a bet. The gauge scores how falsifiable it is — live.

  • Name what concretely happens — built, funded, replaced, displaced

    ·25

  • Add a date or window (e.g. “by Q4 2026”) — without one it can never be falsified

    ·25

  • Add a number, % or threshold — otherwise no one can score it right or wrong

    ·25

  • State your confidence (e.g. “68% likely”) so the call can be calibrated

    ·15

  • Anchor to a base rate or analogue (“like X did”) to discipline the estimate

    ·10

  • Highest-leverage fix

    Add a date or window (e.g. “by Q4 2026”) — without one it can never be falsified

The honest numbers

Every number here is live. Every call is on the record.

The Transparency Panel
Predictions published
2

Locked in — can't be changed

Waiting for outcomes
2

Still inside their time window

Confirmed outcomes
0

First results coming soon

Accuracy score unlocks in
—

First prediction coming due

No fake data. No borrowed logos. No testimonials we can't back up. When the first predictions come due, every outcome — right or wrong — goes public. Until then, this panel just tells you the truth.

Live discovery

Markets forming right now.

ai_ml_infrastructure6/13/2026

unproven

Confidence

14.84%

Market pressure

14.84

developer_tooling6/13/2026

unproven

Confidence

8.83%

Market pressure

8.83

Why this is different from AI assistants

An AI assistant only knows this moment. Our engine remembers how markets actually form.

4 months · continuous evidence

MarAprMayJun

vs. ChatGPT / Claude

We remember how markets form

An AI assistant answers from a frozen moment. Our record tracks months of signals building up — evidence that was there long before anyone thought to search for it.

Committing prediction

API market consolidation

Confidence: 74% · Window: 6–9 months

vs. Analyst reports

Every call is locked in permanently

Every call is locked with a cryptographic hash (SHA-256) before we publish it. We can't quietly change what we said later. An analyst PDF has no such proof.

Prediction committed

DevOps fatigue cluster68%

vs. Static models

We get sharper as we go

Every outcome feeds back into the engine. Wrong calls make the next ones sharper — and we publish the wrong calls too.

The Founder Program

Be on the record before the track record is proven.

The record is empty today, and that's a one-time opening. The founding group does the real work of confirming outcomes — proving the engine works — and earns standing that can't be bought later.

Apply to the Founder Program

Run your ideas through the engine, then save and track them.

Submit real proof of what happened to the markets we called.

Earn a founding-member badge you can show off.

Help shape how accountable forecasting works.

The engine is running right now.

4,190 pieces of evidence read. 38 markets being watched. 2 calls on the record.

Browse opportunities View the ledger