Check your investment thesis against $1B of real money.

ThesisGap decomposes your thesis into atomic claims, matches each against a Polymarket prediction market, and quantifies the gap between crowd consensus and your assumptions.

Sign in to analyze your first thesis. Starter pack is $9 for 20 analyses. Credits never expire. Failed analyses are automatically refunded.

What is ThesisGap?

ThesisGap is a SaaS tool that reconciles investment theses against prediction markets. You paste a natural-language thesis — e.g. “the Fed will not cut more than 50bps in 2026” or “Bitcoin breaks $120,000 by year-end” — and the system uses an AI to decompose it into verifiable sub-claims, then queries Polymarket (roughly $1B of open interest) for markets that match each sub-claim, and finally uses a dual-model ensemble to classify whether each matched market supports, contradicts, or is neutral to the claim.

The final output is a weighted Gap score in percentage points: positive means you are more bullish than the market, negative means you are more bearish. When a sub-claim has no matching market, a separate LLM provides a standalone analysis labeled “AI-only signal, not market-backed” so you can separate market-grounded conclusions from pure AI inference.

How does it work?

The pipeline has four steps, all transparent and independently verifiable:

1. Decompose. The primary LLM splits the free-text thesis into 3–7 atomic sub-claims, each with an explicit time window and a verifiable predicate (the “X reaches Z by date Y” form).
2. Retrieve. Keywords extracted from each sub-claim query the Polymarket CLOB API's /markets endpoint. Candidates are ranked by semantic relevance and the top-N per claim are passed on.
3. Classify. Two independent LLMs label each candidate market as supports / contradicts / neutral. Ties break in favor of the primary model.
4. Score. You drag a “subjective probability” slider for each sub-claim. The system weights by claim importance and market relevance, yielding a Gap score from −100 to +100, live-recomputed as you slide.

Full math and worked example: methodology page.

Who is it for?

Active retail investors, macro and crypto researchers, prediction-market enthusiasts. Anyone who already writes investment theses can use ThesisGap to cross-check the probability assumptions in their head against real-money crowd consensus. Especially useful for:

Long-term holders who want to periodically stress-test their positions
Writers and analysts on X / Substack who want data to back up their takes
Researchers in topics Polymarket covers densely (crypto, macro, elections)
Rational investors who want to see “where do I disagree with the crowd” rather than “just give me an answer”

Who isn't it for? Short-term traders looking for buy/sell signals. ThesisGap outputs probability gaps, not trade instructions. Polymarket itself is CFTC-restricted in the US and inaccessible from mainland China; the product serves Polymarket-accessible geographies.

What does it cost?

Pay per credit. No subscription. Starter is $9 for 20 analyses (~$0.45 each), Pro is $29 for 80 (~$0.36 each), Power is $79 for 250 (~$0.32 each). One-time Stripe checkout. No recurring charges, no cancellation flow.

Credits never expire. Buy and let them sit. If pricing rules change, existing credits are grandfathered under the old terms.

Failed analyses are refunded. If the AI decomposition fails or produces no valid sub-claims, the credit is returned to your balance. You only pay for analyses that actually ship results.

See all three tiers →