Dashboard

AI research needspoint-in-timebeliefdata.BeliefStatesuppliesit.

Thesis

Why BeliefState exists

Founded by Ant.

AI should see the evidence, not another market summary.

Why I built it
AI trading systems cannot evaluate human judgment without source, timing, revision, and outcome history.
Accountability
A useful belief should leave a record: who said it, what supported it, how it changed, and what happened next.
The goal
Make public investor judgment queryable as point-in-time data for AI trading systems and quantitative research.

For AI and quantitative research

Human judgment becomes useful when its history is structured.

BeliefState records who believed what, when, based on which evidence, how conviction changed, and what happened afterward as one point-in-time record.

Market discovery

Query who believed what, when, and why.

BeliefState connects identities, beliefs, companies, sectors, and themes. Every record keeps its evidence, timing, revisions, uncertainty, and measured market outcome.

A thesis should show its work.

Claims can change. Previous versions, losing periods, and reasons stay visible.

  1. 01 / Source captured

    Original publication, author, and timestamp stay attached.

  2. 02 / Thesis formed

    Claim, supporting evidence, time horizon, and catalysts become one record.

  3. 03 / Claim updated

    Authors can change their view. Every earlier claim, timestamp, and reason stays in the record.

  4. 04 / Risk changes

    Counterevidence, uncertainty, and invalidation conditions remain visible.

  5. 05 / Investor decides

    Changes to attention, conviction, or capital are recorded with their reason.

  6. 06 / Outcome measured

    Performance is measured across every version, including losing periods before a claim changed.

Founder and product record

Built independently by Ant.

BeliefState is an independent product. Current milestones are inspectable product capabilities, not customer-count or fundraising claims.

Public record
Source-linked investor calls and measured outcomes are publicly inspectable.
Method
Identity, entry rules, benchmarks, revisions, and limitations are documented.
Delivery
REST and MCP deliver evidence-linked belief and outcome data to AI trading systems and quantitative research workflows.

Investor belief data

A market model should know why people believe what they believe.

BeliefState is a point-in-time investor belief data service for AI trading systems, hedge funds, and quantitative researchers. It links identities, beliefs, evidence, counterevidence, catalysts, risks, invalidation conditions, revisions, outcomes, and methodology versions over time.

If this resonates, follow @beliefstate on X.