86f7f1fa76
Codebase audit identified 23 findings across 4 dimensions (production safety, code health, deployment readiness, test quality). This commit fixes all of them. Broker execution layer (was entirely stubbed): - Real IBKR TWS client via ibapi crate (950+ lines, feature-gated) - ICMarkets ctrader-openapi now always-on (removed feature flag) - Real broker routing with health monitoring and exponential backoff reconnect - Validated against live IB Gateway Docker (6/6 connectivity tests pass) Deployment blockers: - Fixed 6 broken Dockerfiles (removed COPY foxhunt-deploy) - Created foxhunt K8s namespace, secret templates, migration job - Added liveness probes to all 7 K8s services - IB Gateway manifest (ghcr.io/gnzsnz/ib-gateway:stable) - IBKR credentials in Scaleway Secret Manager via Terragrunt - Fixed port collisions and mismatches across services Production safety (9 critical + 6 high/medium fixes): - Asset-class-specific VaR volatility (not flat 2%) - Real parametric VaR with z-score 95th percentile - Kyle's lambda regression (100-bar rolling window) - Per-feature running statistics from historical data - VWAP-based slippage reference, regime duration tracking - Real Databento JSON parsing for OHLCV/Trade/Quote Code health: - Removed #![allow(dead_code)] from ml, data, config - Fixed log:: → tracing:: in 4 production files - Removed dead workspace deps (ratatui, crossterm) Verified: cargo check --workspace (0 errors), trading_engine 330 tests pass. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
ml
Machine learning models for Foxhunt.
Models
- DQN (Rainbow) -- Deep Q-Network with prioritized experience replay, dueling heads, noisy nets, double Q-learning
- PPO -- Proximal Policy Optimization with GAE, LSTM policies, clip-higher option
- TFT -- Temporal Fusion Transformer for multi-horizon time series forecasting
- Mamba2 -- State space model for efficient sequence prediction
- Liquid Networks -- Biologically inspired neural networks for non-stationary data
- TLOB -- Transformer-based Limit Order Book analysis
- Flash Attention -- Optimized attention implementation
Training
Two paths per model:
- Standalone trainer -- direct training loop (e.g.,
DQN::train,PpoTrainer) - UnifiedTrainable adapter -- wraps models for the hyperopt pipeline (e.g.,
DQNTrainableAdapter,UnifiedTrainablePPO)
Inference
InferenceAdapterBridge connects models to the ensemble coordinator in adaptive-strategy. Each model exposes an InferenceAdapter trait for prediction.
Backend
- Candle v0.9.1 --
VarMap,AdamW,loss.backward(),GradStore,opt.step(&grads) - CUDA required for training -- tested on RTX 3050 Ti 4GB, max batch size 230
- CPU inference supported
Hyperopt
ArgminOptimizer (Particle Swarm Optimization) with per-model adapters:
DQN, PPO, ContinuousPPO, TFT, Mamba2. Uses ParameterSpace trait for continuous parameter mapping.
ModelType Enum
15 variants: CompactDQN, DistilledMicroNet, DQN, RainbowDQN, MAMBA, TFT, TGGN, LNN, TLOB, PPO, Transformer, Mamba, LiquidNet, TGNN, Ensemble.
Key Modules
dqn, ppo, tft, mamba, liquid, tlob, flash_attention, ensemble, evaluation, inference, trainers, hyperopt, checkpoint, preprocessing, data_loaders, features, model_factory, training_pipeline, regime_detection, stress_testing, validation, bridge, common, metrics.
Testing
SQLX_OFFLINE=true cargo test -p ml --lib # ~2009 tests