community
/
deltaforge-real-time-cross-asset-strateg
34
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

build(agent): new-agents-4#58ba63 iteration

This commit is contained in:
agent-58ba63c88b4c9625 2026-04-23 23:39:55 +02:00
parent 78dc59df4c
commit 33cf9e010d
1 changed files with 20 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
README.md
View File

@ -28,3 +28,23 @@ Packaging and publishing
For contributors For contributors
- See AGENTS.md for architectural guidelines and testing commands. - See AGENTS.md for architectural guidelines and testing commands.
- Open issues for API changes; keep DSL changes backward compatible where feasible. - Open issues for API changes; keep DSL changes backward compatible where feasible.
Technical Overview (MVP scope)
- Assets and venues: The MVP models two assets across two venues as a minimal cross-venue coordination example.
- DSL surface: The DSL exposes Asset, MarketSignal, SharedSignals, PlanDelta and StrategyDelta to declare objectives and plan hedges.
- Coordination: A lightweight ADMM-inspired coordinator (and a compatibility shim) aggregates per-venue plans into a globally coherent plan.
- Adapters: Two starter adapters (equity_feed, options_feed) provide feed-like signals used in demos.
- Execution: A toy execution engine scaffolds latency-aware routing across venues.
- Backtesting: Deterministic replay engine validates end-to-end flows.
- Tests: A comprehensive test suite exercises MVP components and end-to-end flows.
How to extend (high level)
- Implement a new asset class or venue by extending the Asset model and providing a simple adapter.
- Extend the coordinator with more refined dual-variable state and additional feasibility checks for multi-venue plans.
- Add new primitives (e.g., calendar spreads, dispersion plays) in dsl.py and wiring in the planner.
- Build more realistic adapters (data feeds and brokers) and a real execution adapter.
- Expand the backtester to support Monte Carlo risk/PnL simulations and deterministic replay with more scenarios.
Publishing readiness
- The package is configured to read README.md into the distribution metadata via pyproject.toml.
- A READY_TO_PUBLISH file will be created once all requirements are satisfied and tests pass in CI.