The planned architecture

AUR System

AUR is being designed as a chain of responsibilities, not a black box. Each component has a job, a boundary, and a record of what it received and produced.

01 / The map

A high-level path through AUR

Market information moves through data, intelligence, signal, risk, and execution. Safety and audit cross the system, while the Research Brain sits above it to ask and revise questions.

MARKET → DATA → INTELLIGENCE → SIGNAL → RISK → EXECUTION

02 / The components

Twelve roles, one accountable story.

This page describes planned roles in plain language. It does not publish active signal parameters, protected outcomes, credentials, private infrastructure, or reconstructable edge mechanics.

Market Data

What it does: Collects time-stamped observations from the public market environment.
Why it exists: A system cannot make a timing claim from data it did not actually have.
What it receives: Publicly available market observations and provenance notes.
What it produces: A bounded, time-ordered input record for later stages.

Public statusPLANNED FOUNDATION

Data Quality

What it does: Checks whether inputs are complete, coherent, and usable for the question at hand.
Why it exists: Clean-looking data can still be late, revised, duplicated, or misaligned.
What it receives: Market Data records, timestamps, gaps, and quality flags.
What it produces: Quality warnings, exclusions, and an explanation of what can be trusted.

Public statusPLANNED GUARDRAIL

Market Intelligence

What it does: Turns observations into carefully described context about market structure.
Why it exists: Context helps a person understand what a signal could mean without pretending context is proof.
What it receives: Quality-screened observations and public research definitions.
What it produces: Human-readable features and context, with uncertainty carried forward.

Public statusPLANNED RESEARCH LAYER

Market Regime

What it does: Describes broad conditions that may change how an observation should be interpreted.
Why it exists: A relationship that looks stable in one environment may not mean the same thing in another.
What it receives: Market Intelligence, time context, and documented regime questions.
What it produces: A cautious context label and the reasons it should remain uncertain.

Public statusPLANNED RESEARCH LAYER

Signal Engine

What it does: Tests whether a defined research idea produces a timely, reviewable event.
Why it exists: Separating an event definition from a trading action makes the research easier to falsify.
What it receives: Approved research inputs and a versioned hypothesis definition.
What it produces: A candidate event record, not a promise, recommendation, or public signal.

Public statusPROTECTED DESIGN AREA

Confidence Engine

What it does: Keeps uncertainty visible when the evidence around an event is incomplete.
Why it exists: A fluent model response should not become false precision.
What it receives: Candidate events, quality context, and independent checks.
What it produces: A bounded confidence description for research review, not a probability of profit.

Public statusPROTECTED DESIGN AREA

Risk Engine

What it does: Sets the public safety question around exposure, failure modes, and limits.
Why it exists: Risk has to be considered before an attractive result becomes an action.
What it receives: Candidate events, uncertainty, constraints, and safety policy.
What it produces: A risk decision or refusal, with sensitive mechanics kept private.

Public statusPUBLIC SAFETY INTENT; MECHANICS PRIVATE

Execution Engine

What it does: Represents how an approved action would meet real-world constraints.
Why it exists: Execution assumptions can change the meaning of a research result.
What it receives: A permitted decision, market conditions, and execution constraints.
What it produces: A recorded execution intent or a documented no-action decision.

Public statusCONCEPTUAL ONLY

Position Manager

What it does: Tracks the state and lifecycle of any permitted exposure.
Why it exists: A system needs memory of what is already open before it can reason about what comes next.
What it receives: Audited decisions, state records, and safety limits.
What it produces: A state transition record and a reason for any change.

Public statusCONCEPTUAL ONLY

Safety Layer

What it does: Places hard public-safety and operational boundaries across the whole chain.
Why it exists: A boundary is useful only when it can stop a downstream action.
What it receives: Policy checks, data quality, risk decisions, and system state.
What it produces: Allow, pause, or refuse decisions with a traceable reason.

Public statusPUBLIC SAFETY REQUIREMENT

Audit Trail

What it does: Preserves a chronological account of inputs, versions, decisions, and exceptions.
Why it exists: Memory makes it possible to distinguish a new finding from a polished retelling of an old one.
What it receives: Events and decisions from every permitted stage.
What it produces: A reviewable record without exposing credentials or protected implementation.

Public statusPLANNED EVIDENCE LAYER

Research Brain

What it does: Coordinates questions, versions, independent checks, and what should be learned next.
Why it exists: The research process needs durable memory so it does not endlessly repeat attractive failures.
What it receives: Public questions, prior records, validation plans, and audit summaries.
What it produces: A next research question, a documented change, or a reason to stop.

Public statusRESEARCH DIRECTION

03 / Public boundary

Specific about standards. Careful about protected mechanics.

This page describes planned roles in plain language. It does not publish active signal parameters, protected outcomes, credentials, private infrastructure, or reconstructable edge mechanics.

AUR is not a fund, broker, investment adviser, or signal service. Nothing here is a promise of returns or a claim that a durable edge has been proven.