Skip to content

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

22 Commits

Folders and files

Repository files navigation

Digital Analysis

A stronger, production-oriented evolution of the Digital Oracle idea.

This repository is designed as an agent-native market intelligence system:

  • preserve the original project's strengths: low-dependency providers, structured market signals, transparent reasoning
  • add missing layers: autonomous planning, signal scoring, contradiction analysis, LLM/model routing, report generation, and productization support

Goals

  1. Replicate the quality bar of the original market-signal collection approach
  2. Extend it into a true autonomous analyst architecture
  3. Support multiple model backends: cloud LLM APIs and local models
  4. Make it deployable as a product: API service, analyst workspace, scheduled monitoring, and report delivery

Architecture

User Question
  -> Task Understanding
  -> Signal Planning
  -> Provider Execution
  -> Signal Normalization
  -> Autonomous Analysis Engine
  -> Model-Assisted Reasoning
  -> Report Generation
  -> Product/API/UI Delivery

See:

Repository Structure

src/digital_analysis/
  config/             Runtime settings and model/provider config
  contracts/          Shared domain models
  providers/          External data source adapters
  planner/            Task understanding and signal planning
  execution/          Parallel execution, retries, caching, snapshots
  analysis/           Signal scoring, contradiction analysis, horizon grouping
  models/             LLM/API/local model adapters and routing
  reports/            Structured report builders and renderers
  product/            Product-facing service contracts and orchestration
  cli/                Command-line entrypoints
  tests/              Unit and integration tests

Current Status

This repo is now a working enhanced foundation that preserves and extends the original market-signal approach.

Implemented so far:

  1. package layout and project configuration
  2. architecture / upgrade / product strategy docs
  3. task contracts + evidence contracts
  4. task classifier + priceability checker + simple planner
  5. execution layer:
    • HTTP GET/POST
    • concurrent gather
    • snapshot record/replay
  6. provider coverage now includes:
    • Polymarket
    • Kalshi
    • U.S. Treasury
    • CNN Fear & Greed
    • CME FedWatch
    • Web Search
    • SEC EDGAR
    • CoinGecko
    • BIS
    • CFTC
    • Deribit
    • World Bank
    • Yahoo Price
    • Stooq
    • YFinance Options
  7. baseline autonomous analysis:
    • contradiction detection
    • horizon grouping
    • confidence scoring
    • scenario composition
  8. report and synthesis layer:
    • markdown renderer
    • synthesis prompt builder
    • model-backed report synthesizer
  9. model layer:
    • router
    • OpenAI-compatible adapter
    • Ollama adapter
  10. orchestration and product-facing surface:
  • end-to-end orchestrator
  • analysis service
  • FastAPI app scaffold
  • CLI entrypoint
  1. standard-library test suite passing

Near-term next priorities:

  1. provider-driven evidence extraction from real fetched data
  2. richer signal scoring and weighting
  3. API deployment instructions and runtime wiring
  4. scheduled monitoring / watchlists / persistence
  5. stronger end-to-end examples using real providers and models

Running the API

Install API dependencies:

python3 -m pip install fastapi uvicorn

Start the API:

cd /var/minis/workspace/digital-analysis
PYTHONPATH=src python3 -m uvicorn examples.api_app:app --host 127.0.0.1 --port 8787

Recommended validation method in development/tests:

  • use FastAPI TestClient for in-process verification
  • local loopback HTTP can be environment-sensitive on iSH

Example request body:

{
  "question": "Will there be a recession next year?",
  "synthesize": false
}

Example Scripts

  • Recession Demo
  • Geopolitical Demo
  • API App

License

MIT

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages