From 9c9c7c6d8c435eaf5bf101c9a0cca0d25cf76ba7 Mon Sep 17 00:00:00 2001
From: jgrusewski <jgrusewski@localhost>
Date: Mon, 20 Apr 2026 22:21:09 +0200
Subject: [PATCH] =?UTF-8?q?cleanup(fflag):=20delete=20dead=20use=5Fregime?=
 =?UTF-8?q?=5Fconditioning=20flag=20=E2=80=94=20[FFLAG-009]?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

DQNConfig.use_regime_conditioning was declared, defaulted true at 3
construction sites, but never read anywhere in the workspace. Classic
write-only flag (like use_different_seeds in DEAD-001). Deleted field +
docstring + 3 assignment sites. Regime conditioning is already
unconditional per the trainer — the flag was vestigial.
---
 crates/ml-dqn/src/dqn.rs                          | 10 ----------
 crates/ml/src/trainers/dqn/trainer/constructor.rs |  1 -
 2 files changed, 11 deletions(-)

diff --git a/crates/ml-dqn/src/dqn.rs b/crates/ml-dqn/src/dqn.rs
index 317de060a..371a0f3b7 100644
--- a/crates/ml-dqn/src/dqn.rs
+++ b/crates/ml-dqn/src/dqn.rs
@@ -187,13 +187,6 @@ pub struct DQNConfig {
     pub num_urgency_levels: usize,
 
     /// Regime-conditional importance sampling for the branching loss path.
-    /// When true, per-sample losses are scaled by regime-specific factors
-    /// (Trending=1.2, Ranging=0.8, Volatile=0.6) derived from ADX/CUSUM
-    /// features at state indices 40-41.
-    /// Requires `state_dim` >= 42 to have effect; silently no-ops on smaller vectors.
-    /// Default: true.
-    pub use_regime_conditioning: bool,
-
     /// Feature index for ADX(14) in the market feature vector.
     /// Default: 40 (position in the 42-dim feature layout).
     /// Must match the feature extraction pipeline.
@@ -304,8 +297,6 @@ impl Default for DQNConfig {
             num_order_types: 3,
             num_urgency_levels: 3,
 
-            // Regime conditioning: Enabled by default (applies per-regime IS weights to loss)
-            use_regime_conditioning: true,
             regime_adx_idx: 40,
             regime_cusum_idx: 41,
             regime_adx_threshold: 0.25,
@@ -756,7 +747,6 @@ impl DQNConfig {
             branch_hidden_dim: 128,
             num_order_types: 3,
             num_urgency_levels: 3,
-            use_regime_conditioning: true,
             regime_adx_idx: 40,
             regime_cusum_idx: 41,
             regime_adx_threshold: 0.25,
diff --git a/crates/ml/src/trainers/dqn/trainer/constructor.rs b/crates/ml/src/trainers/dqn/trainer/constructor.rs
index 96840169a..6a11c5769 100644
--- a/crates/ml/src/trainers/dqn/trainer/constructor.rs
+++ b/crates/ml/src/trainers/dqn/trainer/constructor.rs
@@ -290,7 +290,6 @@ impl DQNTrainer {
             iqn_embedding_dim: 64,                              // Fixed (not in search space)
             iqn_lambda: hyperparams.iqn_lambda as f32,          // IQN dual-head loss weight (f64->f32)
             branch_hidden_dim: hyperparams.branch_hidden_dim,
-            use_regime_conditioning: true, // Always enable per-regime IS weights for branching loss
             cvar_alpha: hyperparams.cvar_alpha,
 
             #[allow(clippy::cast_possible_truncation)]