Add BROKER_GATEWAY_SERVICE_URL, DATA_ACQUISITION_SERVICE_URL, and
ML_SERVICE_URL env vars so the gateway can connect to all backend services.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace 3 stub RPCs with real implementations:
- get_system_status: fan-out gRPC health checks to all backends, reports SystemStatus
- get_health_check: fan-out checks with response time measurement, returns HealthCheck entries
- get_metrics: queries Prometheus for requested metric names
MonitoringServiceHandler now accepts backend URLs via with_backends() builder.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Read new env vars: BROKER_GATEWAY_SERVICE_URL, DATA_ACQUISITION_SERVICE_URL,
ML_SERVICE_URL, TRADING_AGENT_SERVICE_URL
- Create lazy channels: trading channel shared by trading/risk/config proxies,
separate channels for ML, broker, data_acq, trading_agent
- Register 7 new services on router with auth interceptor
- Update reflection service with all 6 new proto descriptors
- Add broker/data_acq/trading_agent to health check backend list
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
fxt 2.0 overhaul: consolidated proto/, absorbed monitoring into API Gateway,
new 15-command CLI with gRPC, MCP server mode, TUI cockpit framework.
159 files changed, net -11,835 lines.
Replace `let _ = sync_tensor.to_vec0::<f32>()` with `drop()` to
satisfy clippy's let_underscore_must_use lint on the Result return.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add quantile_huber_loss_per_sample() that preserves the batch dimension
(mean over quantiles only) so PER importance-sampling weights can be
applied per-sample before final batch reduction. Mirrors the existing
quantile_huber_loss() but returns [batch] instead of scalar.
Two tests verify shape correctness, non-negativity, consistency with
the scalar variant (mean of per-sample == scalar loss), and IS weight
application behavior.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Embed DQNConfig architecture hash (SHA-256 of state_dim, num_actions,
hidden_dims, dueling/distributional/noisy/IQN flags) in safetensors
file header on save. Validate hash on load to catch shape mismatches
before touching the VarMap - prevents silent corruption from loading
checkpoints trained with different network architectures.
All 5 save paths (trainable_adapter, trainer serialize_model,
DQNAgentType::save_checkpoint, RegimeConditionalDQN per-head) now
embed metadata. All 3 load paths (DQN::load_from_safetensors,
trainable_adapter::load_checkpoint, ensemble adapter) validate.
No backward compatibility: checkpoints without metadata are rejected
with a clear error message to re-train.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Wire position-delta cost model: costs only apply on position changes,
proportional to delta (not flat per-bar)
- Differentiate order types: Market=15bps, LimitMaker=5bps, IoC=10bps
- Apply urgency weight: Patient=0.5x, Normal=1.0x, Aggressive=1.5x
- Append 3 portfolio features (position, unrealized PnL, drawdown) to match
training's 54-dim state vector (was 51, causing shape mismatch on load)
- Fix default num_actions 3->45 to match training's factored action space
- Add --bars-per-year CLI flag for per-symbol annualization
(ES/NQ=347760, 6E=345000, ZN=105840)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Adds accessor methods for all 11 service clients (trading, ml, ml_training,
broker_gateway, trading_agent, monitoring, config, data_acquisition, risk,
health, config_service). All clients share a single gRPC channel.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Atomic checkpoint writes: write to .tmp then fs::rename() (POSIX atomic)
- Remove redundant thread::sleep(50ms) after CUDA tensor readback sync
- NormStats: bail on missing file instead of computing from test data (lookahead bias)
- q_value_std: warn when missing from training metrics instead of silent 0.0 fallback
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Remove from workspace members, delete K8s deployment, remove nginx routing.
Training metrics and system health now served directly by API Gateway.
Changes:
- Cargo.toml: remove services/monitoring_service from workspace members
- services/monitoring_service/: deleted entirely (6 files)
- infra/k8s/services/monitoring-service.yaml: deleted
- infra/k8s/network-policies/monitoring-service.yaml: deleted
- infra/k8s/network-policies/api-gateway.yaml: remove egress rule to monitoring-service
- infra/k8s/network-policies/web-gateway.yaml: remove egress rule to monitoring-service
- infra/k8s/services/api-gateway.yaml: remove stale MONITORING_SERVICE_URL env var
- infra/k8s/gitlab/tailscale-proxy.yaml: redirect monitor.fxhnt.ai to api-gateway:50051
- .gitlab-ci.yml: remove monitoring_service from compile, copy, and deploy loops
- services/trading_service/src/services/monitoring.rs: update stale comments to
reference api_gateway (not monitoring_service) as training metrics host
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The DQNConfig construction in trainer.rs hardcoded weight_decay to 1e-4,
ignoring the hyperopt-tuned value in DQNHyperparameters. This meant PSO
optimization of weight_decay (search space index 17, range [1e-5, 1e-3])
had no effect on actual training runs.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
DQNConfig.weight_decay (default 1e-4, hyperopt range [1e-5, 1e-3]) was
silently ignored — ParamsAdam always received weight_decay: None. Now
conditionally passes Decay::DecoupledWeightDecay when weight_decay > 0,
enabling L2 regularization that hyperopt has been tuning for.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Port Prometheus scraping and training metrics RPCs directly into
api_gateway, eliminating the monitoring-service backend proxy.
MonitoringServiceProxy replaced with MonitoringServiceHandler that
queries Prometheus directly for training metrics, GPU stats, and
active job counts.
Changes:
- monitoring_proxy.rs -> monitoring_handler.rs with embedded PrometheusClient
- Remove MonitoringBackendConfig and setup_monitoring_proxy from server.rs
- main.rs reads PROMETHEUS_URL + MONITORING_STREAM_INTERVAL instead of
MONITORING_SERVICE_URL
- Remove monitoring-service from backend health check list
- Monitoring is always available (no graceful degradation needed)
- Fix fxt build.rs compile_protos type mismatch
- All 9 ported unit tests + 0 clippy warnings
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
fxt now compiles from the same service protos as all backend services.
Deleted 8 fat-client protos (3,286 lines) from bin/fxt/proto/.
Updated include_proto! macros to use new package names (trading, ml,
config instead of foxhunt.tli, foxhunt.ml, foxhunt.config).
Added risk, data_acquisition, and config_service proto modules.
Fixed duplicate include_proto in agent.rs to use crate::proto.
Existing command implementations will be rewritten to use new proto
types in Task 6.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
4-phase plan from deep audit: critical bug fixes (weight_decay, IQN PER,
q_value_std, 54/51 dim mismatch), train/eval parity (TradeExecutor,
checkpoint validation), SOTA improvements (differential Sharpe, PQN
LayerNorm, temporal PER, dormant neuron resets), and search space cleanup.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Delete per-service proto/ directories. All 7 services now compile
from the single canonical proto/ at workspace root.
- trading_service: 5 protos + ml_training client -> ../../proto/
- ml_training_service: ml_training server + fxt_trading client -> ../../proto/
- broker_gateway_service: broker_gateway -> ../../proto/
- trading_agent_service: trading_agent + ml client -> ../../proto/
- data_acquisition_service: data_acquisition -> ../../proto/
- api_gateway: 8 proto compilations -> ../../proto/
- monitoring_service: keeps local proto (3 training RPCs only, deleted in Task 5)
Added proto/fxt_trading.proto (fat-client proto, package foxhunt.tli)
separate from proto/trading.proto (backend, package trading) since the
API gateway needs both for protocol translation.
Added unimplemented stubs for merged monitoring.proto RPCs:
- trading_service: 3 training RPCs (served by monitoring_service)
- api_gateway monitoring_proxy: 13 system health RPCs (Task 4)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Integrate three standalone PPO modules into the hyperopt training loop:
- PPORewardShaper: hold penalty + rolling Sharpe + diversity bonus per step
- CompositeReward: risk-adjusted reward (downside dev + differential return)
- TrajectoryReplayBuffer: ExO-PPO M=4 rollouts with IS-weighted replay
Also fixes GAE hardcoded gamma=0.99/lambda=0.95 — now uses hyperopt params.
2726 tests pass, 0 clippy errors.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Merge monitoring_service training metrics into trading_service system health
monitoring.proto. All 11 proto files now in one canonical location.
Merged monitoring.proto has 13 RPCs (10 system + 3 training) with all
message types from both source protos preserved. Added cpu_percent,
memory_used_mb, memory_total_mb fields to GetLiveTrainingMetricsResponse.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Integrate all PPO improvement modules into the core training paths:
- Symlog value predictions in compute_value_loss (MLP + LSTM)
- Adaptive entropy auto-tuning replaces fixed entropy_coeff
- Percentile P5/P95 advantage scaling for heavy-tailed returns
- DAPO clip_epsilon_high wired in all 7 PPOConfig construction sites
- Shape mismatch fix in adaptive_entropy (unsqueeze scalar)
2726 tests pass, 0 clippy errors.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Clean rewrite into full-scale operations platform.
Single proto/ root, monitoring_service removed,
CLI + MCP + cockpit TUI with purple/cyan theme.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Change --optimizer default from "pso" to "tpe" since TPE has better
sample efficiency in 25D spaces. Fix clippy let_underscore_must_use
on CUDA sync tensor readback.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Fix 20 exotic parameters to validated defaults, keeping only the 25
that genuinely affect trading performance. This makes both PSO and
TPE dramatically more effective at exploring the parameter space.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add optimize_with_tpe() function that uses Tree-Parzen Estimator for
Bayesian hyperparameter optimization. Add --optimizer=tpe CLI flag
to hyperopt_baseline_rl binary (default: pso for backward compat).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace DType::F32 with training_dtype(&device) for action one-hot
encoding and state embedding conversion. On BF16 devices, model
weights are BF16 but inputs were F32, causing dtype mismatch errors.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Force CUDA synchronization between trials by creating a tiny tensor and
reading it back (forces cudaDeviceSynchronize). This ensures GPU memory
from dropped models is actually freed before the next trial starts.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Tighten max_position_absolute search range from [4,8] to [1,4] to
prevent catastrophic leverage during hyperopt. Add drawdown circuit
breaker to PortfolioTracker that force-closes positions at >20%
drawdown and refuses new trades while flat.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add a standalone TPE optimizer for Bayesian hyperparameter optimization that
replaces PSO by modeling good/bad trial distributions with per-dimension
kernel density estimates. Includes Latin Hypercube Sampling for initial
exploration, Silverman bandwidth selection, log-sum-exp numerical stability,
and JSON-based trial persistence. 15 tests covering splits, bounds, KDE
density, convergence on 1D/2D objectives, and history serialization.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
sample_noise(), sigma init, disable_noise(), and epsilon buffers all
used hardcoded DType::F32. When weights are BF16 on Ampere+ CUDA,
this caused dtype mismatch in mul/add operations during forward pass.
Fix: use training_dtype(device) for all noise-related tensors.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
A+B approach: smooth penalties + position limits + CUDA cleanup +
search space reduction (45D→25D) + TPE optimizer replacing PSO.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
After converting all VarBuilder sites from F32 to training_dtype (BF16 on
Ampere+ GPUs), input tensors from callers remain F32, causing dtype
mismatches in matmul/add/mul operations. This commit adds systematic
boundary casts across all 10 model architectures:
- Input boundary: ensure_training_dtype() at each model forward() entry
- Output boundary: to_dtype(F32) at each model forward() exit
- Internal intermediates: hidden state init, gradient extraction,
positional encodings, causal masks, SSM state matrices, B-spline
basis values all cast to match computation dtype
- Ensemble adapters: ensure_training_dtype after Tensor::from_vec
36 files, +293/-72 lines. 2640 tests pass, 0 failures, 0 clippy warnings.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace DType::F32 with training_dtype(&device) in VarBuilder::from_varmap
calls across TFT, Mamba2, Liquid/CfC, KAN, xLSTM, Diffusion, TGGN, TLOB.
61 sites changed across 25 files (production constructors + test helpers).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace DType::F32 with training_dtype(&device) in all VarBuilder::from_varmap
calls across 16 DQN files (~55 call sites). This enables automatic BF16 weight
initialization on Ampere+ GPUs while keeping F32 on CPU and older hardware.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- gpu_replay_buffer: allocate states/next_states with training_dtype(),
cast incoming batches at ingestion boundary
- gpu_weights: cast BF16 model weights to F32 before extraction for
CUDA f32 kernels in both extract_one() and sync_one()
- mod.rs: cast DqnGpuData, GpuBufferPool, PpoGpuData uploads to
training_dtype(); cast portfolio tensors to match features dtype;
cast bar_target_values readback to F32
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Cast state tensors to BF16 (via training_dtype()) at all DQN network
input sites: compute_loss_internal states/next_states, select_actions_batch,
curiosity module state/next_state, validation batch, and Q-value logging
batch. Non-network tensors (actions, rewards, dones, weights) are left
as F32 since they participate in loss math, not forward passes.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Previously scalar_tensor() rejected BF16 and F16 with an error,
blocking BF16 training for Mamba2. Now BF16/F16 scalars are created
as F32 then cast to the target dtype, matching Candle's internal
precision requirements while supporting half-precision training.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
