Upload V5_DESIGN.md with huggingface_hub
Browse files- V5_DESIGN.md +34 -0
V5_DESIGN.md
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
V5 Trading System — Design and Roadmap
|
| 2 |
+
|
| 3 |
+
Goal
|
| 4 |
+
----
|
| 5 |
+
Build a production-ready V5 trading system for XAUUSD that is robust to market costs and suitable for live paper/live trading. Focus on risk controls, walk-forward validation, realistic fills, and a live inference API.
|
| 6 |
+
|
| 7 |
+
Key Requirements
|
| 8 |
+
----------------
|
| 9 |
+
- Realistic backtests with slippage, commissions, partial fills, and order types
|
| 10 |
+
- Walk-forward training & validation to reduce overfitting
|
| 11 |
+
- Modular model artifacts: tree-only + Keras artifacts (same as V4) but with versioning
|
| 12 |
+
- Live inference API with strict risk manager, position sizing, and circuit-breakers
|
| 13 |
+
- Paper-trade adapter to test on real exchange simulator/broker sandbox
|
| 14 |
+
- Monitoring & alerting (P&L, drawdown, latency, model drift)
|
| 15 |
+
|
| 16 |
+
Milestones
|
| 17 |
+
----------
|
| 18 |
+
1. V5 spec + scaffolding (this doc + FastAPI scaffold)
|
| 19 |
+
2. Realistic backtester improvements
|
| 20 |
+
3. Walk-forward validation pipeline
|
| 21 |
+
4. Ensemble retrain with robust techniques
|
| 22 |
+
5. Live inference + paper-trade adapter
|
| 23 |
+
6. Monitoring, HF model card, and deployment
|
| 24 |
+
|
| 25 |
+
Acceptance Criteria
|
| 26 |
+
-------------------
|
| 27 |
+
- Backtest Sharpe > target (configurable) on OOS sets with realistic costs
|
| 28 |
+
- Max drawdown within risk budget
|
| 29 |
+
- Live inference latency < 100ms per request (p95) on target hardware
|
| 30 |
+
- Automated rollback on model drift or large drawdown
|
| 31 |
+
|
| 32 |
+
Notes
|
| 33 |
+
-----
|
| 34 |
+
Start from existing V4 codebase for feature engineering and model saving. Reuse safe serialization approach. Incrementally add safety features and validate each milestone with tests.
|