jgrusewski 9c249e2a27 fix(dqn): symmetric distributional support + near-zero output init to unfreeze Q-values ...

Root cause: C51 distributional DQN Q-values frozen at ~7.0 for all 30 epochs.
Asymmetric hyperopt support (v_min=-6.7, v_max=17.6) caused init Q ≈ midpoint
= 5.48, creating a self-reinforcing equilibrium that gradient signal couldn't
escape. Additionally, log_q_values() queried the wrong (untrained) network.

Three fixes:
1. Enforce symmetric support in hyperopt: v_min=-v_range, v_max=+v_range
   (midpoint always 0, Q starts unbiased)
2. Near-zero init for distributional output layers (value_out, advantage_out):
   0.01× Xavier scale → softmax ≈ uniform → Q ≈ midpoint regardless of support
3. Fix log_q_values() to use self.forward() (unified dispatch through
   dist_dueling → dueling → standard) instead of self.q_network.forward()

Tests: 350 ml-dqn + 841 ml + 3 ml-core = 1194 passed, 0 failed

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

2026-03-08 18:11:22 +01:00

..

backtesting

docs: create/update README.md for all 17 crates

2026-03-01 22:47:39 +01:00

common

feat(infra): QuestDB metrics sink + monitoring network policies

2026-03-08 02:42:53 +01:00

config

refactor: rename JWT issuer foxhunt-api-gateway → foxhunt-api, drop compat aliases

2026-03-05 00:22:04 +01:00

ctrader-openapi

docs: create/update README.md for all 17 crates

2026-03-01 22:47:39 +01:00

data

fix(ml): streaming parallel OFI + fix hardcoded state_dim=54 in backtest

2026-03-07 23:28:18 +01:00

database

docs: create/update README.md for all 17 crates

2026-03-01 22:47:39 +01:00

market-data

docs: create/update README.md for all 17 crates

2026-03-01 22:47:39 +01:00

ml

fix(dqn): symmetric distributional support + near-zero output init to unfreeze Q-values

2026-03-08 18:11:22 +01:00

ml-asset-selection

refactor(ml): extract universe, backtesting, asset-selection into sub-crates

2026-03-08 15:17:22 +01:00

ml-backtesting

refactor(ml): extract universe, backtesting, asset-selection into sub-crates

2026-03-08 15:17:22 +01:00

ml-checkpoint

refactor(ml): remove dead code across 8 sub-crates — delete 742 lines

2026-03-08 15:17:22 +01:00

ml-core

fix(dqn): symmetric distributional support + near-zero output init to unfreeze Q-values

2026-03-08 18:11:22 +01:00

ml-data

docs: create/update README.md for all 17 crates

2026-03-01 22:47:39 +01:00

ml-data-validation

refactor(ml): split monolith into 9 sub-crates + delete dead code

2026-03-08 15:17:22 +01:00

ml-dqn

fix(dqn): symmetric distributional support + near-zero output init to unfreeze Q-values

2026-03-08 18:11:22 +01:00

ml-ensemble

refactor(ml): split monolith into 9 sub-crates + delete dead code

2026-03-08 15:17:22 +01:00

ml-explainability

refactor(ml): extract regime-detection, explainability, paper-trading

2026-03-08 15:17:22 +01:00

ml-features

fix: resolve rebase artifacts — import paths and MLError variant

2026-03-08 15:21:46 +01:00

ml-hyperopt

refactor(ml): split monolith into 9 sub-crates + delete dead code

2026-03-08 15:17:22 +01:00

ml-labeling

refactor(ml): remove dead code across 8 sub-crates — delete 742 lines

2026-03-08 15:17:22 +01:00

ml-observability

refactor(ml): remove dead code across 8 sub-crates — delete 742 lines

2026-03-08 15:17:22 +01:00

ml-paper-trading

refactor(ml): extract regime-detection, explainability, paper-trading

2026-03-08 15:17:22 +01:00

ml-ppo

refactor(ml): remove dead code across 8 sub-crates — delete 742 lines

2026-03-08 15:17:22 +01:00

ml-regime

refactor(ml): split monolith into 9 sub-crates + delete dead code

2026-03-08 15:17:22 +01:00

ml-regime-detection

refactor(ml): extract regime-detection, explainability, paper-trading

2026-03-08 15:17:22 +01:00

ml-risk

refactor(ml): split monolith into 9 sub-crates + delete dead code

2026-03-08 15:17:22 +01:00

ml-security

refactor(ml): extract observability, stress-testing, security into sub-crates

2026-03-08 15:17:22 +01:00

ml-stress-testing

refactor(ml): extract observability, stress-testing, security into sub-crates

2026-03-08 15:17:22 +01:00

ml-supervised

refactor(ml): remove dead code across 8 sub-crates — delete 742 lines

2026-03-08 15:17:22 +01:00

ml-universe

refactor(ml): remove dead code across 8 sub-crates — delete 742 lines

2026-03-08 15:17:22 +01:00

ml-validation

refactor(ml): split monolith into 9 sub-crates + delete dead code

2026-03-08 15:17:22 +01:00

model_loader

docs: create/update README.md for all 17 crates

2026-03-01 22:47:39 +01:00

risk

refactor: remove trading_engine dep from ml and risk crates

2026-03-05 23:19:38 +01:00

risk-data

docs: create/update README.md for all 17 crates

2026-03-01 22:47:39 +01:00

storage

docs: create/update README.md for all 17 crates

2026-03-01 22:47:39 +01:00

trading_engine

refactor: replace all api_gateway/web-gateway references across codebase

2026-03-05 00:18:58 +01:00

trading-data

docs: create/update README.md for all 17 crates

2026-03-01 22:47:39 +01:00

training_uploader

feat: implement all 15 fxt CLI commands with real gRPC calls

2026-03-03 22:16:35 +01:00