diff --git a/crates/ml/src/trainers/dqn/trainer/constructor.rs b/crates/ml/src/trainers/dqn/trainer/constructor.rs
index 540bb3eb4..e22681523 100644
--- a/crates/ml/src/trainers/dqn/trainer/constructor.rs
+++ b/crates/ml/src/trainers/dqn/trainer/constructor.rs
@@ -30,6 +30,13 @@ use super::DQNTrainer;
 
 impl DQNTrainer {
     pub(crate) fn new_internal(mut hyperparams: DQNHyperparameters, debug_logging: bool, override_device: Option<MlDevice>) -> Result<Self> {
+        // Validate batch_size > 0
+        if hyperparams.batch_size == 0 {
+            return Err(anyhow::anyhow!(
+                "batch_size must be greater than 0"
+            ));
+        }
+
         // WAVE 26 P2.2: Validate gradient_accumulation_steps > 0
         if hyperparams.gradient_accumulation_steps == 0 {
             return Err(anyhow::anyhow!(
diff --git a/crates/ml/src/trainers/dqn/trainer/tests.rs b/crates/ml/src/trainers/dqn/trainer/tests.rs
index 28f5ac9ef..388d8fc94 100644
--- a/crates/ml/src/trainers/dqn/trainer/tests.rs
+++ b/crates/ml/src/trainers/dqn/trainer/tests.rs
@@ -115,13 +115,16 @@ async fn test_feature_vector_to_state() {
     );
 
     let state = state.unwrap();
-    // State dimension is 45 (42 market + 3 portfolio)
-    // - Market features: 0-41 (42 features)
-    // - Portfolio features: 42-44 (3 features, populated by PortfolioTracker)
+    // State dimension depends on whether OFI (MBP-10) data dirs are configured:
+    // - Without OFI: 45 = 42 market + 3 portfolio
+    // - With OFI:    53 = 42 market + 8 OFI + 3 portfolio
+    // Smoketest TOML sets mbp10_data_dir, so OFI is enabled.
+    let expected_dim = if trainer.hyperparams.mbp10_data_dir.is_empty() { 45 } else { 53 };
     assert_eq!(
         state.dimension(),
-        45,
-        "State dimension should be 45 (42 market + 3 portfolio features)"
+        expected_dim,
+        "State dimension mismatch (OFI enabled={})",
+        !trainer.hyperparams.mbp10_data_dir.is_empty()
     );
 }
 
diff --git a/crates/ml/src/training_profile.rs b/crates/ml/src/training_profile.rs
index 3298e8843..901d2f661 100644
--- a/crates/ml/src/training_profile.rs
+++ b/crates/ml/src/training_profile.rs
@@ -1088,13 +1088,13 @@ mod tests {
 
         // [training] values must have been applied
         assert_eq!(hp.epochs, 200);
-        assert_eq!(hp.batch_size, 1024);
+        assert_eq!(hp.batch_size, 8192);
         assert!((hp.learning_rate - 0.0001).abs() < 1e-10);
 
         // OFI MBP-10 data dir should be set in production profile
         assert_eq!(
             hp.mbp10_data_dir.as_str(),
-            "/mnt/training-data/futures-baseline-mbp10",
+            "/data/futures-baseline-mbp10",
             "dqn-production profile must wire mbp10_data_dir"
         );
     }
@@ -1350,7 +1350,7 @@ mod tests {
 
         // [training] section
         assert_eq!(hp.epochs, 200);
-        assert_eq!(hp.batch_size, 1024);
+        assert_eq!(hp.batch_size, 8192);
         assert!((hp.learning_rate - 0.0001).abs() < 1e-10);
         assert!((hp.gamma - 0.99).abs() < 0.001);
         assert!((hp.reward_scale - 1.0).abs() < 0.01);