Educational purposes only. Not financial advice. Data may be outdated — no freshness guarantee. Logic & metrics are verified accurate.

Internal BetaTeaser of SuperIntel

Architecture overview

HedgeVision is structured as a five-layer pipeline — from raw data ingestion through signal generation, risk review, and operator-controlled execution. Each layer is independently auditable and designed to be observable.

Data ingestion

Market data, news feeds, and macro indicators are ingested and normalized. Multiple data providers supported via pluggable adapters.

OHLCV feedsNews API integrationMacro indicator pipelineData quality checks

Pairs discovery engine

Systematic scanning of instrument universes for cointegrated pairs. ADF and Engle-Granger tests with rolling validation.

Cointegration tests (ADF/EG)Rolling validation windowsCorrelation stability scoringPair ranking and selection

Signal generation

Ensemble signal scoring across statistical, ML, and sentiment dimensions. Signals are weighted and passed to the risk chain for review.

Z-score mean reversion signalsML model ensembleLLM sentiment overlayConfidence scoring

Risk chain

Multi-stage veto system. Signals pass through exposure checks, drawdown guards, and position sizing before reaching execution.

Exposure capsDrawdown veto triggersPosition size constraintsCorrelation-adjusted risk

Execution path

Modular broker and exchange adapter layer. Orders are staged for operator review before submission.

Modular broker adaptersPre-trade compliance checksOperator review stageExecution audit log

HedgeVision vs SuperIntel

HedgeVision exposes the statistical arbitrage layer as a public, legible surface. SuperIntel adds multi-model orchestration, persistent research memory, cross-domain intelligence routing, and private risk infrastructure that goes well beyond pair trading. SuperIntel is the full system. HedgeVision is a window into it.

SuperIntel overview

Technology stack Request invite