community
/
marketcompiler-verifiable-dsl-edge-compi
17
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
marketcompiler-verifiable-d.../README.md

3.2 KiB
Raw Permalink Blame History

marketcompiler-verifiable-dsl-edge-compi

Idea summary: A novel, open-source compiler-runtime that lets portfolio managers and fintech developers express investment rules, risk budgets, and execution policies in a mathematically-precise DSL, which is then compiled into portable, verifiable execution graphs suitable for edge devices and desktop workloads.

This repository hosts a lightweight MVP that focuses on portability, auditability, and offline testing for verifiable investment-strategy DSLs.

Key goals

  • A portable DSL to express assets, objectives (return vs. risk), constraints, and execution policies.
  • Verifiable IRs with per-step proofs and cryptographic signing to support end-to-end auditability.
  • Offline-first execution and deterministic delta-sync for governance-enabled workflows.
  • Lightweight, interoperable adapters to plug in data feeds, risk models, and broker interfaces.

What you get in this MVP

  • Core models: LocalProblem, PlanDelta, AuditLog (in marketcompiler_verifiable_dsl_edge_compi/core.py).
  • Signer: simple HMAC-based signing to attest plans (marketcompiler_verifiable_dsl_edge_compi/signer.py).
  • Adapters: starter adapters (PriceFeedAdapter, MockBrokerAdapter) implementing a common interface (AbstractAdapter).
  • Graph-of-Contracts registry: tiny registry to track adapter schemas and versioning (marketcompiler_verifiable_dsl_edge_compi/registry.py).
  • Deterministic backtester: small offline tester to validate DSL-derived plans against historical data (marketcompiler_verifiable_dsl_edge_compi/backtest.py).
  • Packaging ready: Python packaging metadata in pyproject.toml and a functional test suite.

How to use (quickstart)

  • Install locally: python3 -m build && pip install dist/marketcompiler_verifiable_dsl_edge_compi-0.0.1-py3-none-any.whl
  • Run tests: ./test.sh
  • Example usage:
    • Define a LocalProblem with assets and objectives.
    • Create a PlanDelta to capture changes, sign with Signer, and serialize to JSON.
    • Use Backtester.run(lp) to validate a DSL-derived plan against prices.

Design notes

  • The MVP emphasizes simple, explicit data models and deterministic behavior to enable offline testing and governance workflows.
  • The Graph-of-Contracts and adapters are designed to scale as a marketplace of data feeds, risk models, and brokers. TLS-based transport is planned for real adapters.
  • The repository is intentionally small but structured to support rapid extension in Phase 0–3 of the MVP plan.

Roadmap (8–12 weeks)

  • Phase 0 (done for MVP): DSL parser, core IR, two starter adapters, basic local backtester, delta-sync skeleton, tamper-evident AuditLog.
  • Phase 1: Risk budgets (VaR, CVaR), drawdown constraints, governance ledger, cryptographic signing for plans.
  • Phase 2: Cross-exchange adapters, simple optimizer backend for multi-venue execution, mobile SDK for edge execution.
  • Phase 3: End-to-end test harness with simulated markets, metrics on reproducibility, delta-sync latency, and audit-trail completeness.

Contributing

  • See AGENTS.md for architecture and testing rules. This project follows a pragmatic MVP approach with a strong emphasis on correctness and auditability.

License and contact

  • This project is open-source and designed for collaborative improvement.