825db90f23
Major fixes: - C51 v_range calibrated for reward v4 (±2.0, was ±25/±0.5) - Wrong Flat index in Q-gap filter (qe[4]→qe[2] in branching_action_select) - hold_time tracks total position duration (was only losing bars) - Entropy coefficient wired to C51 backward kernel (0.001, was unwired) - Count bonus wired to GPU action selection (per-branch UCB) - Q-gap warmup ramp (0→threshold over 5 epochs, was static) - IQN lambda gradient scaling (max_grad_norm × (1+lambda)) - PER beta annealing 4x faster (500 steps, was 2000) - Reward normalization disabled (scrambled per-bar returns) - Capital floor uses natural return (was hardcoded -1.0) - Financial metrics pipeline: real per-trade GPU stats (was Trades=1) New features: - MSE loss CUDA kernel for C51 warmup phase - Blended MSE→C51 loss with linear alpha ramp - GPU trade_stats_reduce kernel for per-trade financial metrics - TradeStats struct with real win/loss/PF from portfolio states - Behavioral smoke test (Q-values, action entropy, trades) - 50-epoch convergence test with anomaly detection - c51_warmup_epochs in hyperopt search space (41D) Dead code removed: - portfolio_sim_kernel (150 lines CUDA) - DSR/PnL/drawdown reward v2 computations - 7 dead kernel params from env_step signature - GpuPortfolioSimulator (never called) - Reward normalization block + state fields 0 warnings, 0 errors, 1241 unit tests + 8 smoke tests pass. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Foxhunt
Production HFT trading system in Rust.
Architecture
The workspace contains 32 crates organized as follows:
Core Libraries (16)
| Crate | Purpose |
|---|---|
trading_engine |
Order processing, FIX 4.4, IB TWS, SIMD, RDTSC timing |
risk |
VaR, Kelly, circuit breakers, kill switches, compliance |
risk-data |
Risk data types and shared structures |
trading-data |
Trading data types |
ml |
DQN Rainbow, PPO, TFT, Mamba2, ensemble inference |
ml-data |
ML data types and feature definitions |
data |
Market data ingestion and storage |
backtesting |
Replay engine, strategy tester |
adaptive-strategy |
Ensemble execution, microstructure analysis |
common |
Shared types, resilience, error handling |
storage |
S3 and local model storage |
model_loader |
Model serialization and loading |
market-data |
Market data feed handlers |
database |
PostgreSQL access layer (SQLx) |
config |
Configuration management |
tli |
CLI commands and tooling |
Services (8)
| Service | Purpose |
|---|---|
backtesting_service |
gRPC backtesting service |
broker_gateway_service |
FIX routing, broker connectivity |
trading_service |
Core trading operations |
ml_training_service |
Model training orchestration |
data_acquisition_service |
Market data acquisition |
trading_agent_service |
Autonomous trading agents |
api_gateway |
gRPC API gateway with auth |
web-gateway |
Axum REST + WebSocket gateway |
Frontend
web-dashboard/ -- React 19 + TypeScript + Vite + TradingView charts.
Building
# Check compilation (no PostgreSQL required)
SQLX_OFFLINE=true cargo check --workspace
# Run tests for a specific crate
SQLX_OFFLINE=true cargo test -p <crate> --lib
# Clippy
SQLX_OFFLINE=true cargo clippy --workspace
ML Models
Four production model architectures on Candle v0.9.1 with CUDA:
- DQN Rainbow -- Deep Q-Network with prioritized replay, dueling heads, noisy nets
- PPO -- Proximal Policy Optimization with GAE and LSTM policies
- TFT -- Temporal Fusion Transformer for multi-horizon forecasting
- Mamba2 -- State space model for sequence prediction
Each model has a standalone trainer and a UnifiedTrainable adapter for the hyperopt pipeline.
Infrastructure
- Git: Gitea at
git.fxhnt.ai(Tailscale-only), Scaleway DEV1-S - Observability: OpenTelemetry OTLP (env
OTEL_EXPORTER_OTLP_ENDPOINT) - Database: PostgreSQL with SQLx offline mode for CI
License
Proprietary. All rights reserved.