fix(trading_service, ml): replace silently-mock allocation data with logged defaults

- allocation.rs: get_asset_volatilities now uses flat 0.20 (20% annual vol)
  instead of index-scaled 0.15+i*0.05; get_covariance_matrix diagonal is
  0.04 (0.20^2) instead of 0.0225; get_ml_predictions uses 0.0 (neutral)
  instead of 0.05+i*0.02. All three emit tracing::warn so mock state is
  visible in production logs.
- training.rs: train_all emits tracing::warn that it is using a placeholder
  loop and directs callers to use model-specific trainers for production.

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

ml/src/training.rs

@@ -333,6 +333,9 @@ impl TrainingPipeline {
 
             // SIMPLIFIED: Basic training loop - full implementation requires actual model training
             // TODO: Implement proper gradient descent, backpropagation, and loss calculation
+            tracing::warn!(
+                "train_all: using placeholder training loop. Use model-specific trainers for production."
+            );
             for epoch in 0..self.config.epochs.min(3) {
                 // Placeholder metrics - real implementation needs actual training
                 let train_loss = 1.0 / (epoch as f64 + 1.0);

services/trading_service/src/allocation.rs

@@ -663,30 +663,33 @@ impl PortfolioAllocator {
         &self,
         assets: &[String],
     ) -> Result<HashMap<String, f64>, CommonError> {
-        // Mock data for now - in production, calculate from historical prices
+        tracing::warn!(
+            "get_asset_volatilities: using default values (not yet wired to market data)"
+        );
         let mut volatilities = HashMap::new();
-        for (i, symbol) in assets.iter().enumerate() {
-            // Simulate different volatilities
-            let vol = 0.15 + (i as f64 * 0.05);
-            volatilities.insert(symbol.clone(), vol);
+        for symbol in assets.iter() {
+            // 20% annual volatility default — replace with historical calculation
+            volatilities.insert(symbol.clone(), 0.20);
         }
         Ok(volatilities)
     }
 
     /// Get covariance matrix for assets
     async fn get_covariance_matrix(&self, assets: &[String]) -> Result<Vec<Vec<f64>>, CommonError> {
-        // Mock data - in production, calculate from historical returns
+        tracing::warn!(
+            "get_covariance_matrix: using default values (not yet wired to market data)"
+        );
         let n = assets.len();
         let mut matrix = vec![vec![0.0; n]; n];
 
         for i in 0..n {
             for j in 0..n {
                 if i == j {
-                    // Variance on diagonal
-                    matrix[i][j] = 0.0225; // 15% vol squared
+                    // Variance on diagonal: 0.20^2 = 0.04
+                    matrix[i][j] = 0.04;
                 } else {
-                    // Correlation off-diagonal
-                    matrix[i][j] = 0.01; // Low correlation
+                    // Low correlation off-diagonal
+                    matrix[i][j] = 0.01;
                 }
             }
         }
@@ -699,11 +702,13 @@ impl PortfolioAllocator {
         &self,
         assets: &[String],
     ) -> Result<HashMap<String, f64>, CommonError> {
-        // Mock data - in production, call ML service
+        tracing::warn!(
+            "get_ml_predictions: using default values (not yet wired to market data)"
+        );
         let mut predictions = HashMap::new();
-        for (i, symbol) in assets.iter().enumerate() {
-            let pred = 0.05 + (i as f64 * 0.02);
-            predictions.insert(symbol.clone(), pred);
+        for symbol in assets.iter() {
+            // Neutral prediction — replace with actual ML inference
+            predictions.insert(symbol.clone(), 0.0);
         }
         Ok(predictions)
     }