0178a53ab0
Implementation plan for SP15 spec at docs/superpowers/specs/2026-05-06-trader-
discipline-and-recovery-design.md (commit 5417e2756).
Six phases ~4900 LOC, Approach B parallel-where-independent (~10-15 days realistic):
- P.1 Branch + sub-worktrees scaffolding
- 0.0 sp15_isv_slots.rs lands FIRST on sp15 (slots 397-442, ISV_TOTAL_DIM 396→443)
- Phase 0 (3 sub-tasks): EGF diagnostic + ISV-driven retune + anchor test 2.21
- Phase 1 (7 sub-tasks): unified sharpe, cost-net (commission+spread+OFI), DD,
8 baselines fused trunk, dd_pct trunk concat (LAYOUT BREAK), CLI flags,
test slice consumption
- Phase 2 (2A scaffold + 2B 17 tests + 2C 5 tests paired with 3.5)
- Phase 3 (5 teachings sequential): r_quality/r_discipline split, explicit cost,
quadratic DD, regret signal, confidence-aware Hold floor (sigmoid)
- Phase 3.5 (4 mechanisms): asymmetric DD reward, cooldown gate, plasticity
injection (TWO-STEP: Flat + cooldown, Kaiming-He init), recovery PER curriculum
- Phase 4 (4 sub-tasks): pre-flight, L40S Q1-Q7+Q8 walk-forward, sealed Q9 eval-
only workflow, production-track gate
Each task: TDD-discipline (write failing test → run-fail → implement → run-pass
→ commit). Per-commit discipline rule: Phase 2 behavioral test + wire-up audit
+ pearl candidate. Sub-worktree merge model: each phase merges back to sp15
atomically; sp15 merges to main only after Phase 4 production-track gate passes.
Self-review: spec coverage table maps every section to its implementing task.
Some derivative tasks (3.2-3.5) summarized as templated bullets per writing-
plans pattern for skilled-implementer derivative work.
Co-Authored-By: Claude Opus 4.7 (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.