ok

frontend/backend/backtest_service.py

@@ -136,7 +136,8 @@ def calculate_and_save_backtest(date: str, symbol: str, decision: str,
             return_1w=result['return_1w'],
             return_1m=result['return_1m'],
             prediction_correct=result['prediction_correct'],
-            hold_days=result.get('hold_days')
+            hold_days=result.get('hold_days'),
+            return_at_hold=result.get('return_at_hold'),
         )
 
     return result
@@ -231,6 +232,8 @@ def get_backtest_data_for_frontend(date: str, symbol: str) -> dict:
         'actual_return_1d': result['return_1d'],
         'actual_return_1w': result['return_1w'],
         'actual_return_1m': result['return_1m'],
+        'return_at_hold': result.get('return_at_hold'),
+        'hold_days': result.get('hold_days'),
         'price_at_prediction': result['price_at_prediction'],
         'current_price': result.get('price_1m_later') or result.get('price_1w_later'),
         'price_history': price_history

frontend/backend/database.py

@@ -172,6 +172,13 @@ def init_db():
         cursor.execute("ALTER TABLE backtest_results ADD COLUMN hold_days INTEGER")
     except sqlite3.OperationalError:
         pass  # Column already exists
+    try:
+        cursor.execute("ALTER TABLE backtest_results ADD COLUMN return_at_hold REAL")
+        # New column added — delete stale backtest data so it gets recalculated with return_at_hold
+        cursor.execute("DELETE FROM backtest_results")
+        print("Migration: Added return_at_hold column, cleared stale backtest data for recalculation")
+    except sqlite3.OperationalError:
+        pass  # Column already exists
 
     # Create indexes for new tables
     cursor.execute("""
@@ -193,6 +200,80 @@ def init_db():
     conn.commit()
     conn.close()
 
+    # Re-extract hold_days from raw_analysis for rows that have the default value (5)
+    # This fixes data where the signal processor LLM failed to extract the actual hold period
+    _fix_default_hold_days()
+
+
+def _fix_default_hold_days():
+    """Re-extract hold_days from raw_analysis for rows where hold_days is NULL or 5 (defaults).
+
+    The signal processor sometimes defaults to 5 or leaves hold_days NULL when the
+    LLM fails to extract the actual hold period. This function uses regex on the
+    raw_analysis text to find the correct value.
+    """
+    import re
+
+    patterns = [
+        r'(\d+)[\s-]*(?:day|trading[\s-]*day)[\s-]*(?:hold|horizon|period|timeframe)',
+        r'(?:hold|holding)[\s\w]*?(?:for|of|period\s+of)[\s]*(\d+)[\s]*(?:trading\s+)?days?',
+        r'setting\s+(\d+)\s+(?:trading\s+)?days',
+        r'(?:over|within|next)\s+(\d+)\s+(?:trading\s+)?days',
+        r'(\d+)\s+trading\s+days?\s*\(',
+    ]
+
+    def extract_days(text):
+        if not text:
+            return None
+        # Search the conclusion/rationale section first (last 500 chars)
+        conclusion = text[-500:]
+        for pattern in patterns:
+            for match in re.finditer(pattern, conclusion, re.IGNORECASE):
+                days = int(match.group(1))
+                if 1 <= days <= 90:
+                    return days
+        # Fall back to full text
+        for pattern in patterns:
+            for match in re.finditer(pattern, text, re.IGNORECASE):
+                days = int(match.group(1))
+                if 1 <= days <= 90:
+                    return days
+        return None
+
+    conn = get_connection()
+    conn.row_factory = sqlite3.Row
+    cursor = conn.cursor()
+
+    try:
+        # Fix rows where hold_days is NULL or the default 5
+        cursor.execute(
+            "SELECT id, symbol, date, raw_analysis, hold_days, decision FROM stock_analysis "
+            "WHERE (hold_days IS NULL OR hold_days = 5) "
+            "AND decision != 'SELL' "
+            "AND raw_analysis IS NOT NULL AND raw_analysis != ''"
+        )
+        rows = cursor.fetchall()
+        fixed = 0
+        for row in rows:
+            extracted = extract_days(row['raw_analysis'])
+            old_val = row['hold_days']
+            if extracted is not None and extracted != old_val:
+                cursor.execute(
+                    "UPDATE stock_analysis SET hold_days = ? WHERE id = ?",
+                    (extracted, row['id'])
+                )
+                fixed += 1
+                print(f"  Fixed hold_days for {row['symbol']} ({row['date']}): {old_val} -> {extracted}")
+
+        if fixed > 0:
+            conn.commit()
+            # Also clear backtest results so they recalculate with correct hold_days
+            cursor.execute("DELETE FROM backtest_results")
+            conn.commit()
+            print(f"Fixed {fixed} stock(s) with missing/default hold_days. Cleared backtest cache.")
+    finally:
+        conn.close()
+
 
 def save_recommendation(date: str, analysis_data: dict, summary: dict,
                         top_picks: list, stocks_to_avoid: list):
@@ -884,7 +965,7 @@ def save_backtest_result(date: str, symbol: str, decision: str,
                          price_1w_later: float = None, price_1m_later: float = None,
                          return_1d: float = None, return_1w: float = None,
                          return_1m: float = None, prediction_correct: bool = None,
-                         hold_days: int = None):
+                         hold_days: int = None, return_at_hold: float = None):
     """Save a backtest result for a stock recommendation."""
     conn = get_connection()
     cursor = conn.cursor()
@@ -894,14 +975,14 @@ def save_backtest_result(date: str, symbol: str, decision: str,
             INSERT OR REPLACE INTO backtest_results
             (date, symbol, decision, price_at_prediction,
              price_1d_later, price_1w_later, price_1m_later,
-             return_1d, return_1w, return_1m, prediction_correct, hold_days)
-            VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+             return_1d, return_1w, return_1m, prediction_correct, hold_days, return_at_hold)
+            VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
         """, (
             date, symbol, decision, price_at_prediction,
             price_1d_later, price_1w_later, price_1m_later,
             return_1d, return_1w, return_1m,
             1 if prediction_correct else 0 if prediction_correct is not None else None,
-            hold_days
+            hold_days, return_at_hold
         ))
         conn.commit()
     finally:
@@ -933,6 +1014,7 @@ def get_backtest_result(date: str, symbol: str) -> Optional[dict]:
                 'return_1m': row['return_1m'],
                 'prediction_correct': bool(row['prediction_correct']) if row['prediction_correct'] is not None else None,
                 'hold_days': row['hold_days'] if 'hold_days' in row.keys() else None,
+                'return_at_hold': row['return_at_hold'] if 'return_at_hold' in row.keys() else None,
                 'calculated_at': row['calculated_at']
             }
         return None
@@ -962,7 +1044,8 @@ def get_backtest_results_by_date(date: str) -> list:
                 'return_1w': row['return_1w'],
                 'return_1m': row['return_1m'],
                 'prediction_correct': bool(row['prediction_correct']) if row['prediction_correct'] is not None else None,
-                'hold_days': row['hold_days'] if 'hold_days' in row.keys() else None
+                'hold_days': row['hold_days'] if 'hold_days' in row.keys() else None,
+                'return_at_hold': row['return_at_hold'] if 'return_at_hold' in row.keys() else None,
             }
             for row in cursor.fetchall()
         ]
@@ -995,7 +1078,9 @@ def get_all_backtest_results() -> list:
                 'return_1d': row['return_1d'],
                 'return_1w': row['return_1w'],
                 'return_1m': row['return_1m'],
-                'prediction_correct': bool(row['prediction_correct'])
+                'prediction_correct': bool(row['prediction_correct']),
+                'hold_days': row['hold_days'] if 'hold_days' in row.keys() else None,
+                'return_at_hold': row['return_at_hold'] if 'return_at_hold' in row.keys() else None,
             }
             for row in cursor.fetchall()
         ]
@@ -1013,7 +1098,7 @@ def calculate_accuracy_metrics() -> dict:
             'total_predictions': 0,
             'correct_predictions': 0,
             'by_decision': {'BUY': {'accuracy': 0, 'total': 0}, 'SELL': {'accuracy': 0, 'total': 0}, 'HOLD': {'accuracy': 0, 'total': 0}},
-            'by_confidence': {'High': {'accuracy': 0, 'total': 0}, 'Medium': {'accuracy': 0, 'total': 0}, 'Low': {'accuracy': 0, 'total': 0}}
+            'by_confidence': {'HIGH': {'accuracy': 0, 'total': 0}, 'MEDIUM': {'accuracy': 0, 'total': 0}, 'LOW': {'accuracy': 0, 'total': 0}}
         }
 
     total = len(results)
@@ -1035,7 +1120,7 @@ def calculate_accuracy_metrics() -> dict:
 
     # By confidence level
     by_confidence = {}
-    for conf in ['High', 'Medium', 'Low']:
+    for conf in ['HIGH', 'MEDIUM', 'LOW']:
         conf_results = [r for r in results if r.get('confidence') == conf]
         if conf_results:
             conf_correct = sum(1 for r in conf_results if r['prediction_correct'])

frontend/backend/server.py

@@ -228,7 +228,7 @@ def run_analysis_task(symbol: str, date: str, analysis_config: dict = None):
         add_log("agent", "system", f"Starting propagation for {symbol}...")
         add_log("data", "data_fetch", f"Fetching market data for {symbol}...")
 
-        final_state, decision, hold_days = ta.propagate(symbol, date)
+        final_state, decision, hold_days, confidence, risk = ta.propagate(symbol, date)
 
         # Check cancellation after graph execution (skip saving results)
         if _is_cancelled(symbol):
@@ -251,8 +251,8 @@ def run_analysis_task(symbol: str, date: str, analysis_config: dict = None):
         analysis_data = {
             "company_name": symbol,
             "decision": decision.upper() if decision else "HOLD",
-            "confidence": "MEDIUM",
-            "risk": "MEDIUM",
+            "confidence": confidence or "MEDIUM",
+            "risk": risk or "MEDIUM",
             "raw_analysis": raw_analysis,
             "hold_days": hold_days
         }
@@ -947,9 +947,10 @@ async def get_backtest_result(date: str, symbol: str):
         'actual_return_1d': result['return_1d'],
         'actual_return_1w': result['return_1w'],
         'actual_return_1m': result['return_1m'],
+        'return_at_hold': result.get('return_at_hold'),
+        'hold_days': result.get('hold_days'),
         'price_at_prediction': result['price_at_prediction'],
         'current_price': result.get('price_1m_later') or result.get('price_1w_later'),
-        'hold_days': result.get('hold_days'),
     }
 
 

frontend/src/components/pipeline/FlowchartNode.tsx

@@ -1,10 +1,11 @@
 import {
   TrendingUp, TrendingDown, Users, Newspaper, FileText,
   Scale, Target, Zap, Shield, ShieldCheck,
-  Clock, Loader2, CheckCircle, AlertCircle, ChevronDown, ChevronUp
+  Clock, Loader2, CheckCircle, AlertCircle, ChevronDown, ChevronUp, GitBranch
 } from 'lucide-react';
 import { useState } from 'react';
 import type { FlowchartNodeData, PipelineStepStatus } from '../../types/pipeline';
+import { STEP_INPUT_SOURCES } from '../../types/pipeline';
 
 interface FlowchartNodeProps {
   node: FlowchartNodeData;
@@ -100,6 +101,7 @@ export function FlowchartNode({ node, isSelected, onClick }: FlowchartNodeProps)
 
   const hasPreview = !!(node.output_summary || node.agentReport?.report_content || node.debateContent);
   const previewText = node.output_summary || node.agentReport?.report_content || node.debateContent || '';
+  const inputCount = (STEP_INPUT_SOURCES[node.id] || []).length;
 
   return (
     <div className="w-full">
@@ -145,6 +147,12 @@ export function FlowchartNode({ node, isSelected, onClick }: FlowchartNodeProps)
 
         {/* Status + Duration */}
         <div className="flex items-center gap-1.5 flex-shrink-0">
+          {inputCount > 0 && node.status === 'completed' && (
+            <span className="hidden sm:inline-flex items-center gap-0.5 text-[10px] px-1.5 py-0.5 rounded-full bg-blue-50 dark:bg-blue-900/20 text-blue-500 dark:text-blue-400 border border-blue-200 dark:border-blue-800/50">
+              <GitBranch className="w-2.5 h-2.5" />
+              {inputCount}
+            </span>
+          )}
           {node.duration_ms != null && node.status === 'completed' && (
             <span className="text-[10px] sm:text-xs font-mono font-semibold text-gray-500 dark:text-gray-400">
               {formatDuration(node.duration_ms)}

frontend/src/components/pipeline/NodeDetailDrawer.tsx

@@ -1,6 +1,7 @@
-import { useState } from 'react';
-import { X, Clock, Timer, CheckCircle, AlertCircle, FileText, MessageSquare, ChevronDown, ChevronRight, Terminal, Bot, User, Wrench } from 'lucide-react';
+import { useState, useMemo } from 'react';
+import { X, Clock, Timer, CheckCircle, AlertCircle, FileText, MessageSquare, ChevronDown, ChevronRight, Terminal, Bot, User, GitBranch, ArrowRight, Info, Wrench } from 'lucide-react';
 import type { FlowchartNodeData, FullPipelineData, StepDetails } from '../../types/pipeline';
+import { STEP_INPUT_SOURCES, FLOWCHART_STEPS, mapPipelineToFlowchart } from '../../types/pipeline';
 
 interface NodeDetailDrawerProps {
   node: FlowchartNodeData;
@@ -26,24 +27,33 @@ function formatDuration(ms: number): string {
   return `${mins}m ${remSecs}s`;
 }
 
-function getFallbackContent(node: FlowchartNodeData, data: FullPipelineData | null): string {
+/**
+ * Extract the ACTUAL meaningful output for a step.
+ * Prefers agent_reports and debate content over raw step_details.response,
+ * since the latter may contain "TOOL_CALL:" noise or be empty for debate steps.
+ */
+function getStepOutput(node: FlowchartNodeData, data: FullPipelineData | null): string {
+  // 1. Agent report content (for analyst steps)
   if (node.agentReport?.report_content) return node.agentReport.report_content;
+  // 2. Debate content (for debate participants)
   if (node.debateContent) return node.debateContent;
+  // 3. Output summary from pipeline step
   if (node.output_summary) return node.output_summary;
-  if (!data) return '';
-
-  if (node.debateType === 'investment' && data.debates?.investment) {
-    const d = data.debates.investment;
-    if (node.debateRole === 'bull') return d.bull_arguments || '';
-    if (node.debateRole === 'bear') return d.bear_arguments || '';
-    if (node.debateRole === 'judge') return d.judge_decision || '';
-  }
-  if (node.debateType === 'risk' && data.debates?.risk) {
-    const d = data.debates.risk;
-    if (node.debateRole === 'risky') return d.risky_arguments || '';
-    if (node.debateRole === 'safe') return d.safe_arguments || '';
-    if (node.debateRole === 'neutral') return d.neutral_arguments || '';
-    if (node.debateRole === 'judge') return d.judge_decision || '';
+  // 4. Try debate tables directly
+  if (data) {
+    if (node.debateType === 'investment' && data.debates?.investment) {
+      const d = data.debates.investment;
+      if (node.debateRole === 'bull') return d.bull_arguments || '';
+      if (node.debateRole === 'bear') return d.bear_arguments || '';
+      if (node.debateRole === 'judge') return d.judge_decision || '';
+    }
+    if (node.debateType === 'risk' && data.debates?.risk) {
+      const d = data.debates.risk;
+      if (node.debateRole === 'risky') return d.risky_arguments || '';
+      if (node.debateRole === 'safe') return d.safe_arguments || '';
+      if (node.debateRole === 'neutral') return d.neutral_arguments || '';
+      if (node.debateRole === 'judge') return d.judge_decision || '';
+    }
   }
   return '';
 }
@@ -94,12 +104,243 @@ function CodeBlock({ content, maxHeight = 'max-h-64' }: { content: string; maxHe
   );
 }
 
+/** Tool call with optional result */
+interface ToolCallEntry {
+  name: string;
+  args: string;
+  result_preview?: string;
+}
+
+/** Get tool calls: prefer stored step_details.tool_calls, fall back to text parsing */
+function getToolCalls(details: StepDetails | undefined, rawResponse: string): ToolCallEntry[] {
+  // Prefer stored tool_calls with actual results from the backend
+  if (details?.tool_calls && details.tool_calls.length > 0) {
+    return details.tool_calls.map(tc => ({
+      name: tc.name,
+      args: tc.args || '',
+      result_preview: tc.result_preview,
+    }));
+  }
+  // Fallback: parse TOOL_CALL: patterns from response text (legacy data)
+  const regex = /TOOL_CALL:\s*(\w+)\(([^)]*)\)/g;
+  const calls: ToolCallEntry[] = [];
+  let match;
+  while ((match = regex.exec(rawResponse)) !== null) {
+    calls.push({ name: match[1], args: match[2] });
+  }
+  return calls;
+}
+
+/** Strip TOOL_CALL lines from response to get only the prose */
+function stripToolCalls(text: string): string {
+  return text.replace(/TOOL_CALL:\s*\w+\([^)]*\)\s*/g, '').trim();
+}
+
+/** Collapsible result preview for a single tool call */
+function ToolResultPreview({ result }: { result: string }) {
+  const [expanded, setExpanded] = useState(false);
+  const isLong = result.length > 200;
+  const display = expanded ? result : result.slice(0, 200);
+
+  return (
+    <div className="mt-1.5">
+      <div className="text-[10px] font-semibold text-green-600 dark:text-green-400 mb-0.5 flex items-center gap-1">
+        <CheckCircle className="w-2.5 h-2.5" />
+        Result:
+      </div>
+      <div className="max-h-40 overflow-y-auto rounded bg-slate-900 dark:bg-black/40 px-2 py-1.5">
+        <pre className="text-[10px] text-green-300 dark:text-green-400 font-mono whitespace-pre-wrap leading-relaxed break-all">
+          {display}{isLong && !expanded ? '...' : ''}
+        </pre>
+      </div>
+      {isLong && (
+        <button
+          onClick={() => setExpanded(!expanded)}
+          className="text-[10px] text-blue-500 hover:text-blue-400 mt-0.5"
+        >
+          {expanded ? 'Show less' : `Show all (${result.length} chars)`}
+        </button>
+      )}
+    </div>
+  );
+}
+
+/** Tool calls panel showing invocations and their results */
+function ToolCallsPanel({ calls }: { calls: ToolCallEntry[] }) {
+  if (calls.length === 0) return null;
+
+  return (
+    <div className="divide-y divide-slate-200 dark:divide-slate-700">
+      {calls.map((call, i) => {
+        // Distinguish: undefined/null = no data stored, empty string = tool returned nothing, error = starts with [
+        const resultStored = call.result_preview !== undefined && call.result_preview !== null;
+        const isError = resultStored && call.result_preview!.startsWith('[');
+        const hasData = resultStored && call.result_preview!.length > 0 && !isError;
+        const isEmpty = resultStored && call.result_preview!.length === 0;
+        return (
+          <div key={i} className="px-3 py-2 flex items-start gap-2">
+            <div className={`flex-shrink-0 mt-0.5 w-5 h-5 rounded flex items-center justify-center ${
+              hasData
+                ? 'bg-green-100 dark:bg-green-900/30'
+                : isError
+                  ? 'bg-red-100 dark:bg-red-900/30'
+                  : resultStored
+                    ? 'bg-slate-100 dark:bg-slate-800'
+                    : 'bg-amber-100 dark:bg-amber-900/30'
+            }`}>
+              <Wrench className={`w-3 h-3 ${
+                hasData ? 'text-green-600 dark:text-green-400'
+                  : isError ? 'text-red-600 dark:text-red-400'
+                  : resultStored ? 'text-slate-500 dark:text-slate-400'
+                  : 'text-amber-600 dark:text-amber-400'
+              }`} />
+            </div>
+            <div className="flex-1 min-w-0">
+              <div className="flex items-center gap-1.5">
+                <span className="text-xs font-semibold text-gray-800 dark:text-gray-200 font-mono">{call.name}</span>
+                <span className="text-[10px] text-gray-400">(#{i + 1})</span>
+                {resultStored && (
+                  <span className={`text-[9px] px-1 py-0.5 rounded font-medium ${
+                    hasData
+                      ? 'bg-green-100 dark:bg-green-900/30 text-green-600 dark:text-green-400'
+                      : isError
+                        ? 'bg-red-100 dark:bg-red-900/30 text-red-600 dark:text-red-400'
+                        : 'bg-slate-100 dark:bg-slate-800 text-slate-500 dark:text-slate-400'
+                  }`}>
+                    {hasData ? 'executed' : isError ? 'error' : 'empty result'}
+                  </span>
+                )}
+              </div>
+              <div className="mt-0.5 text-[11px] font-mono text-gray-500 dark:text-gray-400 break-all">
+                ({call.args})
+              </div>
+              {hasData && (
+                <ToolResultPreview result={call.result_preview!} />
+              )}
+              {isError && (
+                <div className="mt-1 text-[10px] text-red-600 dark:text-red-400">
+                  {call.result_preview}
+                </div>
+              )}
+              {isEmpty && (
+                <div className="mt-1 text-[10px] text-slate-500 dark:text-slate-400 italic">
+                  Tool executed but returned no data
+                </div>
+              )}
+              {!resultStored && (
+                <div className="mt-1 text-[10px] text-amber-600 dark:text-amber-400 italic">
+                  No result data available (legacy run)
+                </div>
+              )}
+            </div>
+          </div>
+        );
+      })}
+    </div>
+  );
+}
+
+/** Context received panel showing data lineage */
+function ContextReceivedPanel({ node, pipelineData }: { node: FlowchartNodeData; pipelineData: FullPipelineData | null }) {
+  const inputSources = STEP_INPUT_SOURCES[node.id] || [];
+  const allNodes = useMemo(() => mapPipelineToFlowchart(pipelineData), [pipelineData]);
+
+  if (inputSources.length === 0) {
+    return (
+      <div className="flex items-center gap-2 px-3 py-2.5 rounded-lg bg-slate-50 dark:bg-slate-900/40 border border-slate-200 dark:border-slate-700">
+        <Info className="w-3.5 h-3.5 text-slate-400 flex-shrink-0" />
+        <span className="text-xs text-slate-500 dark:text-slate-400">
+          Independent step — no forwarded context from previous steps
+        </span>
+      </div>
+    );
+  }
+
+  return (
+    <div className="rounded-lg border border-blue-200 dark:border-blue-800/60 bg-blue-50/50 dark:bg-blue-900/10 overflow-hidden">
+      <div className="flex items-center gap-2 px-3 py-2 bg-blue-50 dark:bg-blue-900/20 border-b border-blue-200 dark:border-blue-800/60">
+        <GitBranch className="w-3.5 h-3.5 text-blue-500 flex-shrink-0" />
+        <span className="text-xs font-semibold text-blue-700 dark:text-blue-300">
+          Context Received from {inputSources.length} Previous Steps
+        </span>
+      </div>
+      <div className="px-3 py-2.5 flex flex-wrap gap-1.5">
+        {inputSources.map(sourceId => {
+          const stepDef = FLOWCHART_STEPS.find(s => s.id === sourceId);
+          const sourceNode = allNodes.find(n => n.id === sourceId);
+          const isCompleted = sourceNode?.status === 'completed';
+          const label = stepDef?.label || sourceId;
+
+          return (
+            <span
+              key={sourceId}
+              className={`inline-flex items-center gap-1 text-[11px] px-2 py-1 rounded-md border ${
+                isCompleted
+                  ? 'bg-green-50 dark:bg-green-900/20 text-green-700 dark:text-green-400 border-green-200 dark:border-green-800/60'
+                  : 'bg-slate-50 dark:bg-slate-800 text-slate-500 dark:text-slate-400 border-slate-200 dark:border-slate-700'
+              }`}
+            >
+              {isCompleted ? (
+                <CheckCircle className="w-3 h-3 text-green-500 flex-shrink-0" />
+              ) : (
+                <Clock className="w-3 h-3 text-slate-400 flex-shrink-0" />
+              )}
+              {label}
+              <ArrowRight className="w-2.5 h-2.5 opacity-40" />
+            </span>
+          );
+        })}
+      </div>
+    </div>
+  );
+}
+
+/** Empty state for steps with no data */
+function EmptyState({ status }: { status: string }) {
+  if (status === 'pending') {
+    return (
+      <div className="text-center py-6 text-gray-400 dark:text-gray-500">
+        <Clock className="w-8 h-8 mx-auto mb-2 opacity-50" />
+        <p className="text-sm">This step hasn't run yet</p>
+        <p className="text-xs mt-1">Run an analysis to see results here</p>
+      </div>
+    );
+  }
+  if (status === 'running') {
+    return (
+      <div className="text-center py-6 text-blue-500 dark:text-blue-400">
+        <div className="w-8 h-8 mx-auto mb-2 border-2 border-blue-500 border-t-transparent rounded-full animate-spin" />
+        <p className="text-sm">Processing...</p>
+      </div>
+    );
+  }
+  return (
+    <div className="text-center py-6 text-gray-400 dark:text-gray-500">
+      <FileText className="w-8 h-8 mx-auto mb-2 opacity-50" />
+      <p className="text-sm">No output data available</p>
+    </div>
+  );
+}
+
 export function NodeDetailDrawer({ node, pipelineData, onClose }: NodeDetailDrawerProps) {
   const details: StepDetails | undefined = node.step_details;
-  const fallbackContent = getFallbackContent(node, pipelineData);
+  const inputSources = STEP_INPUT_SOURCES[node.id] || [];
+  const hasForwardedContext = inputSources.length > 0;
 
-  // Determine if we have structured data or just fallback
-  const hasStructuredData = details && (details.system_prompt || details.user_prompt || details.response);
+  // Step Output: the ACTUAL meaningful output (agent report / debate content first, then raw response)
+  // Always strip TOOL_CALL lines from step output since they're noise in the output view
+  const rawStepOutput = getStepOutput(node, pipelineData) || details?.response || '';
+  const stepOutput = stripToolCalls(rawStepOutput);
+  // Raw LLM Response: the raw text from step_details (may contain TOOL_CALL noise)
+  const rawResponse = details?.response || '';
+  // Get tool calls: prefers stored results from backend, falls back to text parsing
+  const parsedToolCalls = useMemo(() => getToolCalls(details, rawResponse), [details, rawResponse]);
+  // Clean response = raw response minus TOOL_CALL lines (for display without noise)
+  const cleanRawResponse = useMemo(() => parsedToolCalls.length > 0 ? stripToolCalls(rawResponse) : rawResponse, [rawResponse, parsedToolCalls]);
+  // Only show raw response section if it differs from step output (avoid duplication)
+  const showRawResponse = rawResponse && rawResponse !== stepOutput;
+
+  const hasAnyContent = stepOutput || details?.user_prompt || details?.system_prompt;
 
   return (
     <div className="mt-3 rounded-xl border border-slate-200 dark:border-slate-700 bg-white dark:bg-slate-800 shadow-lg overflow-hidden animate-in slide-in-from-top-2">
@@ -153,26 +394,30 @@ export function NodeDetailDrawer({ node, pipelineData, onClose }: NodeDetailDraw
         )}
       </div>
 
-      {/* Content sections */}
+      {/* Content sections — unified flow, no branching */}
       <div className="p-3 space-y-2">
-        {hasStructuredData ? (
+        {/* 1. Context Received — always first */}
+        <ContextReceivedPanel node={node} pipelineData={pipelineData} />
+
+        {hasAnyContent ? (
           <>
-            {/* System Prompt */}
-            {details.system_prompt && (
+            {/* 2. System Instructions — role definition (default OPEN) */}
+            {details?.system_prompt && (
               <Section
-                title="System Prompt"
+                title="System Instructions"
                 icon={Bot}
                 iconColor="text-violet-500"
+                defaultOpen={true}
                 badge={`${details.system_prompt.length} chars`}
               >
                 <CodeBlock content={details.system_prompt} maxHeight="max-h-48" />
               </Section>
             )}
 
-            {/* User Prompt / Input */}
-            {details.user_prompt && (
+            {/* 3. Input Prompt — forwarded context or initial prompt */}
+            {details?.user_prompt && (
               <Section
-                title="User Prompt / Input"
+                title="Input Prompt"
                 icon={User}
                 iconColor="text-blue-500"
                 badge={`${details.user_prompt.length} chars`}
@@ -181,80 +426,44 @@ export function NodeDetailDrawer({ node, pipelineData, onClose }: NodeDetailDraw
               </Section>
             )}
 
-            {/* Tool Calls */}
-            {details.tool_calls && details.tool_calls.length > 0 && (
+            {/* 4. Tool Calls — parsed from raw response (analyst steps) */}
+            {parsedToolCalls.length > 0 && (
               <Section
                 title="Tool Calls"
                 icon={Wrench}
                 iconColor="text-amber-500"
-                badge={`${details.tool_calls.length} calls`}
+                badge={`${parsedToolCalls.length} calls`}
               >
-                <div className="p-3 space-y-3">
-                  {details.tool_calls.map((tc, i) => (
-                    <div key={i} className="space-y-1">
-                      <div className="flex items-start gap-2 text-xs">
-                        <span className="font-mono font-semibold text-amber-600 dark:text-amber-400 whitespace-nowrap">
-                          {tc.name}()
-                        </span>
-                        {tc.args && (
-                          <span className="font-mono text-gray-500 dark:text-gray-400 truncate">
-                            {tc.args}
-                          </span>
-                        )}
-                      </div>
-                      {tc.result_preview && (
-                        <div className="ml-4 p-2 bg-slate-900 dark:bg-black/40 rounded text-[11px] font-mono text-green-300 dark:text-green-400 max-h-32 overflow-auto whitespace-pre-wrap">
-                          {tc.result_preview}
-                        </div>
-                      )}
-                    </div>
-                  ))}
-                </div>
+                <ToolCallsPanel calls={parsedToolCalls} />
               </Section>
             )}
 
-            {/* LLM Response */}
-            {details.response && (
+            {/* 5. Raw LLM Response — only when different from Step Output */}
+            {showRawResponse && (
               <Section
-                title="LLM Response"
+                title="Raw LLM Response"
                 icon={MessageSquare}
-                iconColor="text-green-500"
-                defaultOpen={true}
-                badge={`${details.response.length} chars`}
+                iconColor="text-gray-500"
+                badge={`${rawResponse.length} chars`}
               >
-                <CodeBlock content={details.response} maxHeight="max-h-80" />
+                <CodeBlock content={cleanRawResponse || rawResponse} maxHeight="max-h-48" />
+              </Section>
+            )}
+
+            {/* 6. Step Output — the actual meaningful output */}
+            {stepOutput && (
+              <Section
+                title="Step Output"
+                icon={CheckCircle}
+                iconColor="text-green-500"
+                badge={`${stepOutput.length} chars`}
+              >
+                <CodeBlock content={stepOutput} maxHeight="max-h-80" />
               </Section>
             )}
           </>
-        ) : fallbackContent ? (
-          /* Fallback: show the old-style content */
-          <>
-            <Section
-              title={node.agentType ? 'Agent Report' : node.debateRole === 'judge' ? 'Decision' : node.debateType ? 'Debate Argument' : 'Output'}
-              icon={node.agentType ? FileText : node.debateType ? MessageSquare : FileText}
-              iconColor="text-gray-500"
-              defaultOpen={true}
-              badge={`${fallbackContent.length} chars`}
-            >
-              <CodeBlock content={fallbackContent} maxHeight="max-h-80" />
-            </Section>
-          </>
-        ) : node.status === 'pending' ? (
-          <div className="text-center py-6 text-gray-400 dark:text-gray-500">
-            <Clock className="w-8 h-8 mx-auto mb-2 opacity-50" />
-            <p className="text-sm">This step hasn't run yet</p>
-            <p className="text-xs mt-1">Run an analysis to see results here</p>
-          </div>
-        ) : node.status === 'running' ? (
-          <div className="text-center py-6 text-blue-500 dark:text-blue-400">
-            <div className="w-8 h-8 mx-auto mb-2 border-2 border-blue-500 border-t-transparent rounded-full animate-spin" />
-            <p className="text-sm">Processing...</p>
-          </div>
         ) : (
-          <div className="text-center py-6 text-gray-400 dark:text-gray-500">
-            <FileText className="w-8 h-8 mx-auto mb-2 opacity-50" />
-            <p className="text-sm">No output data available</p>
-          </div>
+          <EmptyState status={node.status} />
         )}
       </div>
     </div>

frontend/src/pages/Dashboard.tsx

@@ -2,7 +2,7 @@ import { useState, useMemo, useEffect, useCallback, useRef } from 'react';
 import { Link } from 'react-router-dom';
 import { Calendar, RefreshCw, Filter, ChevronRight, TrendingUp, TrendingDown, Minus, History, Search, X, Play, Loader2, Square, AlertCircle, Terminal } from 'lucide-react';
 import TopPicks, { StocksToAvoid } from '../components/TopPicks';
-import { DecisionBadge } from '../components/StockCard';
+import { DecisionBadge, HoldDaysBadge } from '../components/StockCard';
 import TerminalModal from '../components/TerminalModal';
 import HowItWorks from '../components/HowItWorks';
 import BackgroundSparkline from '../components/BackgroundSparkline';
@@ -196,7 +196,7 @@ export default function Dashboard() {
 
     try {
       // Pass settings from context to the API
-      await api.runBulkAnalysis(undefined, {
+      const result = await api.runBulkAnalysis(undefined, {
         deep_think_model: settings.deepThinkModel,
         quick_think_model: settings.quickThinkModel,
         provider: settings.provider,
@@ -204,10 +204,25 @@ export default function Dashboard() {
         max_debate_rounds: settings.maxDebateRounds,
         parallel_workers: settings.parallelWorkers
       });
+
+      // If all stocks already analyzed, exit analyzing mode
+      if (result.status === 'completed' || result.total_stocks === 0) {
+        updateAnalysisState(false, null);
+        addNotification({
+          type: 'info',
+          title: 'Already Analyzed',
+          message: result.skipped
+            ? `All ${result.skipped} stocks already analyzed for today.`
+            : 'All stocks already analyzed for today.',
+          duration: 5000,
+        });
+        return;
+      }
+
       addNotification({
         type: 'info',
         title: 'Analysis Started',
-        message: 'Running AI analysis for all 50 Nifty stocks...',
+        message: `Running AI analysis for ${result.total_stocks} stocks${result.skipped ? ` (${result.skipped} already done)` : ''}...`,
         duration: 3000,
       });
     } catch (e) {
@@ -647,6 +662,11 @@ export default function Dashboard() {
                       <DecisionBadge decision={item.analysis.decision} size="small" />
                     </div>
                     <p className="text-xs text-gray-500 dark:text-gray-400 truncate">{item.company_name}</p>
+                    {item.analysis.hold_days != null && item.analysis.hold_days > 0 && item.analysis.decision !== 'SELL' && (
+                      <div className="mt-1">
+                        <HoldDaysBadge holdDays={item.analysis.hold_days} decision={item.analysis.decision} />
+                      </div>
+                    )}
                   </div>
                 </Link>
               );

frontend/src/pages/History.tsx

@@ -2,6 +2,7 @@ import { useState, useMemo, useEffect, useCallback } from 'react';
 import { Link } from 'react-router-dom';
 import { Calendar, TrendingUp, TrendingDown, Minus, ChevronRight, BarChart3, Target, HelpCircle, Activity, Calculator, LineChart, PieChart, Shield, Filter, Loader2, AlertCircle } from 'lucide-react';
 import { sampleRecommendations, getBacktestResult as getStaticBacktestResult, calculateAccuracyMetrics as calculateStaticAccuracyMetrics, getDateStats as getStaticDateStats, getOverallStats as getStaticOverallStats, getReturnBreakdown as getStaticReturnBreakdown } from '../data/recommendations';
+import type { ReturnBreakdown } from '../data/recommendations';
 import { DecisionBadge, HoldDaysBadge } from '../components/StockCard';
 import Sparkline from '../components/Sparkline';
 import AccuracyBadge from '../components/AccuracyBadge';
@@ -23,6 +24,9 @@ interface RealBacktestData {
   decision: string;
   return1d: number | null;
   return1w: number | null;
+  returnAtHold: number | null;
+  holdDays: number | null;
+  primaryReturn: number | null;  // return_at_hold ?? return_1d
   predictionCorrect: boolean | null;
   priceHistory?: Array<{ date: string; price: number }>;
 }
@@ -74,6 +78,21 @@ function InvestmentModeToggle({
   );
 }
 
+// Pulsing skeleton bar for loading states
+function SkeletonBar({ className = '' }: { className?: string }) {
+  return <div className={`animate-pulse bg-gray-200 dark:bg-slate-700 rounded ${className}`} />;
+}
+
+// Loading overlay for chart sections
+function SectionLoader({ message = 'Calculating backtest results...' }: { message?: string }) {
+  return (
+    <div className="flex flex-col items-center justify-center py-8 gap-2">
+      <Loader2 className="w-6 h-6 text-nifty-500 animate-spin" />
+      <span className="text-xs text-gray-500 dark:text-gray-400">{message}</span>
+    </div>
+  );
+}
+
 export default function History() {
   const [selectedDate, setSelectedDate] = useState<string | null>(null);
   const [showAccuracyModal, setShowAccuracyModal] = useState(false);
@@ -163,7 +182,7 @@ export default function History() {
 
   // Batch-fetch all backtest results per date (used by both accuracy trend and chart data)
   const [batchBacktestByDate, setBatchBacktestByDate] = useState<
-    Record<string, Record<string, { return_1d?: number; return_1w?: number; return_1m?: number; prediction_correct?: boolean; decision: string }>>
+    Record<string, Record<string, { return_1d?: number; return_1w?: number; return_1m?: number; return_at_hold?: number; hold_days?: number; prediction_correct?: boolean; decision: string }>>
   >({});
   const [isBatchLoading, setIsBatchLoading] = useState(false);
 
@@ -203,6 +222,8 @@ export default function History() {
                 return_1d: r.return_1d,
                 return_1w: r.return_1w,
                 return_1m: r.return_1m,
+                return_at_hold: r.return_at_hold,
+                hold_days: r.hold_days,
                 prediction_correct: r.prediction_correct,
                 decision: r.decision,
               };
@@ -245,11 +266,12 @@ export default function History() {
       for (const symbol of Object.keys(rec.analysis)) {
         const stockAnalysis = rec.analysis[symbol];
         const bt = dateBacktest[symbol];
-        if (!stockAnalysis?.decision || !bt || bt.return_1d === undefined || bt.return_1d === null) continue;
+        const primaryRet = bt?.return_at_hold ?? bt?.return_1d;
+        if (!stockAnalysis?.decision || primaryRet === undefined || primaryRet === null) continue;
 
         const predictionCorrect = (stockAnalysis.decision === 'BUY' || stockAnalysis.decision === 'HOLD')
-          ? bt.return_1d > 0
-          : bt.return_1d < 0;
+          ? primaryRet > 0
+          : primaryRet < 0;
 
         if (stockAnalysis.decision === 'BUY') { totalBuy++; if (predictionCorrect) correctBuy++; }
         else if (stockAnalysis.decision === 'SELL') { totalSell++; if (predictionCorrect) correctSell++; }
@@ -361,32 +383,31 @@ export default function History() {
         for (const symbol of Object.keys(rec.analysis)) {
           const stockAnalysis = rec.analysis[symbol];
           const bt = dateBacktest[symbol];
-          if (!stockAnalysis?.decision || !bt || bt.return_1d === undefined || bt.return_1d === null) continue;
-
-          const return1d = bt.return_1d;
+          const primaryRet = bt?.return_at_hold ?? bt?.return_1d;
+          if (!stockAnalysis?.decision || primaryRet === undefined || primaryRet === null) continue;
 
           // Store for PortfolioSimulator
           if (!allBacktest[date]) allBacktest[date] = {};
-          allBacktest[date][symbol] = return1d;
+          allBacktest[date][symbol] = primaryRet;
 
           const predictionCorrect = (stockAnalysis.decision === 'BUY' || stockAnalysis.decision === 'HOLD')
-            ? return1d > 0
-            : return1d < 0;
+            ? primaryRet > 0
+            : primaryRet < 0;
 
           totalPredictions++;
           if (predictionCorrect) {
             totalCorrect++;
             dateCorrectCount++;
             if (stockAnalysis.decision === 'BUY' || stockAnalysis.decision === 'HOLD') {
-              dateCorrectReturn += return1d;
+              dateCorrectReturn += primaryRet;
             } else {
-              dateCorrectReturn += Math.abs(return1d);
+              dateCorrectReturn += Math.abs(primaryRet);
             }
           } else {
             if (stockAnalysis.decision === 'BUY' || stockAnalysis.decision === 'HOLD') {
-              dateIncorrectReturn += return1d;
+              dateIncorrectReturn += primaryRet;
             } else {
-              dateIncorrectReturn += -Math.abs(return1d);
+              dateIncorrectReturn += -Math.abs(primaryRet);
             }
           }
           dateTotalCount++;
@@ -428,9 +449,10 @@ export default function History() {
         if (rec && dateBacktest) {
           for (const symbol of Object.keys(rec.analysis)) {
             const bt = dateBacktest[symbol];
-            if (!bt || bt.return_1d === undefined || bt.return_1d === null) continue;
+            const retVal = bt?.return_at_hold ?? bt?.return_1d;
+            if (retVal === undefined || retVal === null) continue;
             for (const bucket of returnBuckets) {
-              if (bt.return_1d >= bucket.min && bt.return_1d < bucket.max) {
+              if (retVal >= bucket.min && retVal < bucket.max) {
                 bucket.count++;
                 bucket.stocks.push(symbol);
                 break;
@@ -535,8 +557,9 @@ export default function History() {
 
         for (const pick of rec.top_picks) {
           const bt = dateBacktest[pick.symbol];
-          if (bt && bt.return_1d !== undefined && bt.return_1d !== null) {
-            dateReturn += bt.return_1d;
+          const retVal = bt?.return_at_hold ?? bt?.return_1d;
+          if (retVal !== undefined && retVal !== null) {
+            dateReturn += retVal;
             dateCount++;
           }
         }
@@ -564,9 +587,10 @@ export default function History() {
         if (rec && dateBacktest) {
           for (const pick of rec.top_picks) {
             const bt = dateBacktest[pick.symbol];
-            if (bt && bt.return_1d !== undefined && bt.return_1d !== null) {
+            const retVal = bt?.return_at_hold ?? bt?.return_1d;
+            if (retVal !== undefined && retVal !== null) {
               for (const bucket of topPicksDistribution) {
-                if (bt.return_1d >= bucket.min && bt.return_1d < bucket.max) {
+                if (retVal >= bucket.min && retVal < bucket.max) {
                   bucket.count++;
                   bucket.stocks.push(pick.symbol);
                   break;
@@ -690,14 +714,14 @@ export default function History() {
         const backtest = await api.getBacktestResult(date, stock.symbol);
 
         if (backtest.available) {
-          // Calculate prediction correctness based on 1-day return
-          // BUY/HOLD correct if return > 0, SELL correct if return < 0
+          // Use hold-period return when available, fall back to 1-day
+          const primaryReturn = backtest.return_at_hold ?? backtest.actual_return_1d ?? null;
           let predictionCorrect: boolean | null = null;
-          if (backtest.actual_return_1d !== undefined && backtest.actual_return_1d !== null) {
+          if (primaryReturn !== null) {
             if (stock.decision === 'BUY' || stock.decision === 'HOLD') {
-              predictionCorrect = backtest.actual_return_1d > 0;
+              predictionCorrect = primaryReturn > 0;
             } else if (stock.decision === 'SELL') {
-              predictionCorrect = backtest.actual_return_1d < 0;
+              predictionCorrect = primaryReturn < 0;
             }
           }
 
@@ -706,6 +730,9 @@ export default function History() {
             decision: stock.decision,
             return1d: backtest.actual_return_1d ?? null,
             return1w: backtest.actual_return_1w ?? null,
+            returnAtHold: backtest.return_at_hold ?? null,
+            holdDays: backtest.hold_days ?? null,
+            primaryReturn,
             predictionCorrect,
             priceHistory: backtest.price_history,
           };
@@ -752,8 +779,9 @@ export default function History() {
       rec.top_picks.map(pick => {
         // Try real backtest data first
         const realData = realBacktestData[pick.symbol];
-        if (realData?.return1d !== null && realData?.return1d !== undefined) {
-          return realData.return1d;
+        const primaryRet = realData?.primaryReturn ?? realData?.return1d;
+        if (primaryRet !== null && primaryRet !== undefined) {
+          return primaryRet;
         }
         // Only fall back to mock when actually using mock data
         return isUsingMockData ? getStaticBacktestResult(pick.symbol)?.actual_return_1d : undefined;
@@ -809,6 +837,71 @@ export default function History() {
     return Object.values(rec.analysis);
   };
 
+  // Build ReturnBreakdown from real batch backtest data for the modal
+  const buildReturnBreakdown = useCallback((date: string): ReturnBreakdown | null => {
+    const rec = recommendations.find(r => r.date === date);
+    const dateBacktest = batchBacktestByDate[date];
+    if (!rec || !dateBacktest) return null;
+
+    const correctStocks: { symbol: string; decision: string; return1d: number }[] = [];
+    const incorrectStocks: { symbol: string; decision: string; return1d: number }[] = [];
+    let correctTotal = 0;
+    let incorrectTotal = 0;
+
+    for (const symbol of Object.keys(rec.analysis)) {
+      const stockAnalysis = rec.analysis[symbol];
+      const bt = dateBacktest[symbol];
+      const retVal = bt?.return_at_hold ?? bt?.return_1d;
+      if (!stockAnalysis?.decision || retVal === undefined || retVal === null) continue;
+
+      const isCorrect = (stockAnalysis.decision === 'BUY' || stockAnalysis.decision === 'HOLD')
+        ? retVal > 0
+        : retVal < 0;
+
+      const entry = { symbol, decision: stockAnalysis.decision, return1d: retVal };
+      if (isCorrect) {
+        correctStocks.push(entry);
+        correctTotal += (stockAnalysis.decision === 'BUY' || stockAnalysis.decision === 'HOLD') ? retVal : Math.abs(retVal);
+      } else {
+        incorrectStocks.push(entry);
+        incorrectTotal += (stockAnalysis.decision === 'BUY' || stockAnalysis.decision === 'HOLD') ? retVal : -Math.abs(retVal);
+      }
+    }
+
+    const totalCount = correctStocks.length + incorrectStocks.length;
+    if (totalCount === 0) return null;
+
+    const correctAvg = correctStocks.length > 0 ? correctTotal / correctStocks.length : 0;
+    const incorrectAvg = incorrectStocks.length > 0 ? incorrectTotal / incorrectStocks.length : 0;
+    const correctWeight = correctStocks.length / totalCount;
+    const incorrectWeight = incorrectStocks.length / totalCount;
+    const weightedReturn = (correctAvg * correctWeight) + (incorrectAvg * incorrectWeight);
+
+    // Sort stocks by return magnitude
+    correctStocks.sort((a, b) => Math.abs(b.return1d) - Math.abs(a.return1d));
+    incorrectStocks.sort((a, b) => Math.abs(b.return1d) - Math.abs(a.return1d));
+
+    return {
+      correctPredictions: {
+        count: correctStocks.length,
+        totalReturn: correctTotal,
+        avgReturn: correctAvg,
+        stocks: correctStocks.slice(0, 5),
+      },
+      incorrectPredictions: {
+        count: incorrectStocks.length,
+        totalReturn: incorrectTotal,
+        avgReturn: incorrectAvg,
+        stocks: incorrectStocks.slice(0, 5),
+      },
+      weightedReturn: Math.round(weightedReturn * 10) / 10,
+      formula: `(${correctAvg.toFixed(2)}% × ${correctStocks.length}/${totalCount}) + (${incorrectAvg.toFixed(2)}% × ${incorrectStocks.length}/${totalCount}) = ${weightedReturn.toFixed(2)}%`,
+    };
+  }, [recommendations, batchBacktestByDate]);
+
+  // Whether any backtest data is still loading (for skeleton states)
+  const isBacktestDataLoading = isBatchLoading || (!isUsingMockData && !isLoadingRecommendations && Object.keys(batchBacktestByDate).length === 0);
+
   // Show loading state
   if (isLoadingRecommendations) {
     return (
@@ -859,6 +952,9 @@ export default function History() {
           <div className="flex items-center gap-2">
             <Target className="w-5 h-5 text-nifty-600 dark:text-nifty-400" />
             <h2 className="font-semibold text-gray-900 dark:text-gray-100">Prediction Accuracy</h2>
+            {isBacktestDataLoading && (
+              <Loader2 className="w-3.5 h-3.5 animate-spin text-nifty-500" />
+            )}
           </div>
           <button
             onClick={() => setShowAccuracyModal(true)}
@@ -869,34 +965,48 @@ export default function History() {
             <span className="hidden sm:inline">How it's calculated</span>
           </button>
         </div>
-        <div className="grid grid-cols-2 sm:grid-cols-4 gap-3">
-          <div className="p-3 rounded-lg bg-nifty-50 dark:bg-nifty-900/20 text-center">
-            <div className="text-2xl font-bold text-nifty-600 dark:text-nifty-400">
-              {(accuracyMetrics.success_rate * 100).toFixed(0)}%
-            </div>
-            <div className="text-xs text-gray-500 dark:text-gray-400">Overall Accuracy</div>
+        {isBacktestDataLoading ? (
+          <div className="grid grid-cols-2 sm:grid-cols-4 gap-3">
+            {['nifty', 'green', 'red', 'amber'].map(color => (
+              <div key={color} className={`p-3 rounded-lg bg-${color === 'nifty' ? 'nifty-50 dark:bg-nifty-900/20' : `${color}-50 dark:bg-${color}-900/20`} text-center`}>
+                <SkeletonBar className="h-7 w-16 mx-auto mb-1" />
+                <SkeletonBar className="h-3 w-20 mx-auto" />
+              </div>
+            ))}
           </div>
-          <div className="p-3 rounded-lg bg-green-50 dark:bg-green-900/20 text-center">
-            <div className="text-2xl font-bold text-green-600 dark:text-green-400">
-              {(accuracyMetrics.buy_accuracy * 100).toFixed(0)}%
+        ) : (
+          <div className="grid grid-cols-2 sm:grid-cols-4 gap-3">
+            <div className="p-3 rounded-lg bg-nifty-50 dark:bg-nifty-900/20 text-center">
+              <div className="text-2xl font-bold text-nifty-600 dark:text-nifty-400">
+                {(accuracyMetrics.success_rate * 100).toFixed(0)}%
+              </div>
+              <div className="text-xs text-gray-500 dark:text-gray-400">Overall Accuracy</div>
             </div>
-            <div className="text-xs text-gray-500 dark:text-gray-400">Buy Accuracy</div>
-          </div>
-          <div className="p-3 rounded-lg bg-red-50 dark:bg-red-900/20 text-center">
-            <div className="text-2xl font-bold text-red-600 dark:text-red-400">
-              {(accuracyMetrics.sell_accuracy * 100).toFixed(0)}%
+            <div className="p-3 rounded-lg bg-green-50 dark:bg-green-900/20 text-center">
+              <div className="text-2xl font-bold text-green-600 dark:text-green-400">
+                {(accuracyMetrics.buy_accuracy * 100).toFixed(0)}%
+              </div>
+              <div className="text-xs text-gray-500 dark:text-gray-400">Buy Accuracy</div>
             </div>
-            <div className="text-xs text-gray-500 dark:text-gray-400">Sell Accuracy</div>
-          </div>
-          <div className="p-3 rounded-lg bg-amber-50 dark:bg-amber-900/20 text-center">
-            <div className="text-2xl font-bold text-amber-600 dark:text-amber-400">
-              {(accuracyMetrics.hold_accuracy * 100).toFixed(0)}%
+            <div className="p-3 rounded-lg bg-red-50 dark:bg-red-900/20 text-center">
+              <div className="text-2xl font-bold text-red-600 dark:text-red-400">
+                {(accuracyMetrics.sell_accuracy * 100).toFixed(0)}%
+              </div>
+              <div className="text-xs text-gray-500 dark:text-gray-400">Sell Accuracy</div>
+            </div>
+            <div className="p-3 rounded-lg bg-amber-50 dark:bg-amber-900/20 text-center">
+              <div className="text-2xl font-bold text-amber-600 dark:text-amber-400">
+                {(accuracyMetrics.hold_accuracy * 100).toFixed(0)}%
+              </div>
+              <div className="text-xs text-gray-500 dark:text-gray-400">Hold Accuracy</div>
             </div>
-            <div className="text-xs text-gray-500 dark:text-gray-400">Hold Accuracy</div>
           </div>
-        </div>
+        )}
         <p className="text-[10px] text-gray-400 dark:text-gray-500 mt-2 text-center">
-          Based on {accuracyMetrics.total_predictions} predictions tracked over time
+          {isBacktestDataLoading
+            ? 'Fetching backtest data from market...'
+            : `Based on ${accuracyMetrics.total_predictions} predictions tracked over time`
+          }
         </p>
       </section>
 
@@ -914,23 +1024,28 @@ export default function History() {
             </div>
           )}
         </div>
-        {/* Pass real data if available, use mock fallback only when in mock mode */}
-        <AccuracyTrendChart
-          height={200}
-          data={isUsingMockData
-            ? (accuracyTrendData.length > 0 ? accuracyTrendData : undefined)
-            : accuracyTrendData
-          }
-        />
-        <p className="text-[10px] text-gray-400 dark:text-gray-500 mt-2 text-center">
-          {accuracyTrendData.length > 0 ? (
-            <>Prediction accuracy from real backtest data over {accuracyTrendData.length} trading days</>
-          ) : isUsingMockData ? (
-            <>Demo data - Start backend for real accuracy tracking</>
-          ) : (
-            <>Prediction accuracy over the past {dates.length} trading days</>
-          )}
-        </p>
+        {isBacktestDataLoading && !isUsingMockData ? (
+          <SectionLoader message="Computing accuracy trend from backtest data..." />
+        ) : (
+          <>
+            <AccuracyTrendChart
+              height={200}
+              data={isUsingMockData
+                ? (accuracyTrendData.length > 0 ? accuracyTrendData : undefined)
+                : accuracyTrendData
+              }
+            />
+            <p className="text-[10px] text-gray-400 dark:text-gray-500 mt-2 text-center">
+              {accuracyTrendData.length > 0 ? (
+                <>Prediction accuracy from real backtest data over {accuracyTrendData.length} trading days</>
+              ) : isUsingMockData ? (
+                <>Demo data - Start backend for real accuracy tracking</>
+              ) : (
+                <>Prediction accuracy over the past {dates.length} trading days</>
+              )}
+            </p>
+          </>
+        )}
       </section>
 
       {/* Risk Metrics */}
@@ -947,19 +1062,25 @@ export default function History() {
             </div>
           )}
         </div>
-        <RiskMetricsCard metrics={!isUsingMockData && !realRiskMetrics ? {
-          sharpeRatio: 0, maxDrawdown: 0, winLossRatio: 0, winRate: 0,
-          volatility: 0, totalTrades: 0,
-        } : realRiskMetrics} />
-        <p className="text-[10px] text-gray-400 dark:text-gray-500 mt-2 text-center">
-          {realRiskMetrics ? (
-            <>Risk-adjusted performance from real backtest data ({realRiskMetrics.totalTrades} trades)</>
-          ) : isUsingMockData ? (
-            <>Demo data - Start backend for real risk metrics</>
-          ) : (
-            <>Risk-adjusted performance metrics for the AI trading strategy</>
-          )}
-        </p>
+        {isBacktestDataLoading && !isUsingMockData ? (
+          <SectionLoader message="Computing risk metrics from backtest data..." />
+        ) : (
+          <>
+            <RiskMetricsCard metrics={!isUsingMockData && !realRiskMetrics ? {
+              sharpeRatio: 0, maxDrawdown: 0, winLossRatio: 0, winRate: 0,
+              volatility: 0, totalTrades: 0,
+            } : realRiskMetrics} />
+            <p className="text-[10px] text-gray-400 dark:text-gray-500 mt-2 text-center">
+              {realRiskMetrics ? (
+                <>Risk-adjusted performance from real backtest data ({realRiskMetrics.totalTrades} trades)</>
+              ) : isUsingMockData ? (
+                <>Demo data - Start backend for real risk metrics</>
+              ) : (
+                <>Risk-adjusted performance metrics for the AI trading strategy</>
+              )}
+            </p>
+          </>
+        )}
       </section>
 
       {/* Portfolio Simulator */}
@@ -987,6 +1108,7 @@ export default function History() {
             const rec = getRecommendation(date);
             const stats = dateStatsMap[date];
             const avgReturn = stats?.avgReturn1d ?? 0;
+            const hasBacktestData = !isUsingMockData ? (realDateReturns[date] !== undefined) : true;
             const isPositive = avgReturn >= 0;
 
             // Calculate filtered summary for this date
@@ -1005,11 +1127,21 @@ export default function History() {
                   }`}
                 >
                   <div className="font-semibold">{new Date(date).toLocaleDateString('en-IN', { month: 'short', day: 'numeric' })}</div>
-                  <div className={`text-sm font-bold mt-0.5 ${
-                    selectedDate === date ? 'text-white' : getValueColorClass(avgReturn)
-                  }`}>
-                    {isPositive ? '+' : ''}{avgReturn.toFixed(1)}%
-                  </div>
+                  {!hasBacktestData && isBacktestDataLoading ? (
+                    <div className={`text-sm font-bold mt-0.5 ${selectedDate === date ? 'text-white/60' : 'text-gray-400 dark:text-gray-500'}`}>
+                      <span className="inline-block w-8 h-4 animate-pulse bg-gray-300 dark:bg-slate-600 rounded" />
+                    </div>
+                  ) : !hasBacktestData ? (
+                    <div className={`text-sm mt-0.5 ${selectedDate === date ? 'text-white/60' : 'text-gray-400 dark:text-gray-500'}`}>
+                      Pending
+                    </div>
+                  ) : (
+                    <div className={`text-sm font-bold mt-0.5 ${
+                      selectedDate === date ? 'text-white' : getValueColorClass(avgReturn)
+                    }`}>
+                      {isPositive ? '+' : ''}{avgReturn.toFixed(1)}%
+                    </div>
+                  )}
                   <div className={`text-[10px] mt-0.5 ${selectedDate === date ? 'text-white/80' : 'opacity-60'}`}>
                     {filteredSummary.buy}B/{filteredSummary.sell}S/{filteredSummary.hold}H
                   </div>
@@ -1118,8 +1250,8 @@ export default function History() {
                 let predictionCorrect: boolean | null = null;
 
                 if (!isUsingMockData) {
-                  // Real data mode: only use real backtest, no mock fallback
-                  nextDayReturn = realData?.return1d ?? null;
+                  // Real data mode: use hold-period return when available
+                  nextDayReturn = realData?.primaryReturn ?? realData?.return1d ?? null;
                   priceHistory = realData?.priceHistory;
                   if (realData?.predictionCorrect !== undefined) {
                     predictionCorrect = realData.predictionCorrect;
@@ -1127,7 +1259,7 @@ export default function History() {
                 } else {
                   // Mock data mode: use real if available, fall back to mock
                   const mockBacktest = getStaticBacktestResult(stock.symbol);
-                  nextDayReturn = realData?.return1d ?? mockBacktest?.actual_return_1d ?? 0;
+                  nextDayReturn = realData?.primaryReturn ?? realData?.return1d ?? mockBacktest?.actual_return_1d ?? 0;
                   priceHistory = realData?.priceHistory ?? mockBacktest?.price_history;
                   if (realData?.predictionCorrect !== undefined) {
                     predictionCorrect = realData.predictionCorrect;
@@ -1159,6 +1291,12 @@ export default function History() {
                     <div className="flex items-center gap-3">
                       <DecisionBadge decision={stock.decision} size="small" />
                       <HoldDaysBadge holdDays={stock.hold_days} decision={stock.decision} />
+                      {nextDayReturn !== null && (
+                        <span className={`text-xs font-medium tabular-nums ${getValueColorClass(nextDayReturn)}`} title={realData?.holdDays ? `${realData.holdDays}d return` : '1d return'}>
+                          {nextDayReturn >= 0 ? '+' : ''}{nextDayReturn.toFixed(1)}%
+                          {realData?.holdDays && <span className="text-[9px] opacity-60 ml-0.5">/{realData.holdDays}d</span>}
+                        </span>
+                      )}
                       {predictionCorrect !== null && (
                         <AccuracyBadge
                           correct={predictionCorrect}
@@ -1193,54 +1331,67 @@ export default function History() {
           </div>
           <InvestmentModeToggle mode={summaryMode} onChange={setSummaryMode} />
         </div>
-        <div className="grid grid-cols-2 sm:grid-cols-4 gap-3">
-          <div
-            className="p-3 rounded-lg bg-gray-50 dark:bg-slate-700/50 text-center cursor-pointer hover:bg-gray-100 dark:hover:bg-slate-700 transition-colors"
-            onClick={() => setActiveSummaryModal('daysTracked')}
-          >
-            <div className="text-xl font-bold text-nifty-600 dark:text-nifty-400">{filteredStats.totalDays}</div>
-            <div className="text-xs text-gray-500 dark:text-gray-400 flex items-center justify-center gap-1">
-              Days Tracked <HelpCircle className="w-3 h-3" />
+        {isBacktestDataLoading && !isUsingMockData ? (
+          <div className="grid grid-cols-2 sm:grid-cols-4 gap-3">
+            {[1, 2, 3, 4].map(i => (
+              <div key={i} className="p-3 rounded-lg bg-gray-50 dark:bg-slate-700/50 text-center">
+                <SkeletonBar className="h-6 w-12 mx-auto mb-1" />
+                <SkeletonBar className="h-3 w-20 mx-auto" />
+              </div>
+            ))}
+          </div>
+        ) : (
+          <div className="grid grid-cols-2 sm:grid-cols-4 gap-3">
+            <div
+              className="p-3 rounded-lg bg-gray-50 dark:bg-slate-700/50 text-center cursor-pointer hover:bg-gray-100 dark:hover:bg-slate-700 transition-colors"
+              onClick={() => setActiveSummaryModal('daysTracked')}
+            >
+              <div className="text-xl font-bold text-nifty-600 dark:text-nifty-400">{filteredStats.totalDays}</div>
+              <div className="text-xs text-gray-500 dark:text-gray-400 flex items-center justify-center gap-1">
+                Days Tracked <HelpCircle className="w-3 h-3" />
+              </div>
+            </div>
+            <div
+              className="p-3 rounded-lg bg-gray-50 dark:bg-slate-700/50 text-center cursor-pointer hover:bg-gray-100 dark:hover:bg-slate-700 transition-colors"
+              onClick={() => setActiveSummaryModal('avgReturn')}
+            >
+              <div className={`text-xl font-bold ${getValueColorClass(filteredStats.avgDailyReturn)}`}>
+                {filteredStats.avgDailyReturn >= 0 ? '+' : ''}{filteredStats.avgDailyReturn.toFixed(1)}%
+              </div>
+              <div className="text-xs text-gray-500 dark:text-gray-400 flex items-center justify-center gap-1">
+                Avg Return <HelpCircle className="w-3 h-3" />
+              </div>
+            </div>
+            <div
+              className="p-3 rounded-lg bg-gray-50 dark:bg-slate-700/50 text-center cursor-pointer hover:bg-gray-100 dark:hover:bg-slate-700 transition-colors"
+              onClick={() => setActiveSummaryModal('buySignals')}
+            >
+              <div className="text-xl font-bold text-green-600 dark:text-green-400">
+                {filteredStats.buySignals}
+              </div>
+              <div className="text-xs text-gray-500 dark:text-gray-400 flex items-center justify-center gap-1">
+                {summaryMode === 'topPicks' ? 'Top Pick Signals' : 'Buy Signals'} <HelpCircle className="w-3 h-3" />
+              </div>
+            </div>
+            <div
+              className="p-3 rounded-lg bg-gray-50 dark:bg-slate-700/50 text-center cursor-pointer hover:bg-gray-100 dark:hover:bg-slate-700 transition-colors"
+              onClick={() => setActiveSummaryModal('sellSignals')}
+            >
+              <div className="text-xl font-bold text-red-600 dark:text-red-400">
+                {filteredStats.sellSignals}
+              </div>
+              <div className="text-xs text-gray-500 dark:text-gray-400 flex items-center justify-center gap-1">
+                Sell Signals <HelpCircle className="w-3 h-3" />
+              </div>
             </div>
           </div>
-          <div
-            className="p-3 rounded-lg bg-gray-50 dark:bg-slate-700/50 text-center cursor-pointer hover:bg-gray-100 dark:hover:bg-slate-700 transition-colors"
-            onClick={() => setActiveSummaryModal('avgReturn')}
-          >
-            <div className={`text-xl font-bold ${getValueColorClass(filteredStats.avgDailyReturn)}`}>
-              {filteredStats.avgDailyReturn >= 0 ? '+' : ''}{filteredStats.avgDailyReturn.toFixed(1)}%
-            </div>
-            <div className="text-xs text-gray-500 dark:text-gray-400 flex items-center justify-center gap-1">
-              Avg Next-Day Return <HelpCircle className="w-3 h-3" />
-            </div>
-          </div>
-          <div
-            className="p-3 rounded-lg bg-gray-50 dark:bg-slate-700/50 text-center cursor-pointer hover:bg-gray-100 dark:hover:bg-slate-700 transition-colors"
-            onClick={() => setActiveSummaryModal('buySignals')}
-          >
-            <div className="text-xl font-bold text-green-600 dark:text-green-400">
-              {filteredStats.buySignals}
-            </div>
-            <div className="text-xs text-gray-500 dark:text-gray-400 flex items-center justify-center gap-1">
-              {summaryMode === 'topPicks' ? 'Top Pick Signals' : 'Buy Signals'} <HelpCircle className="w-3 h-3" />
-            </div>
-          </div>
-          <div
-            className="p-3 rounded-lg bg-gray-50 dark:bg-slate-700/50 text-center cursor-pointer hover:bg-gray-100 dark:hover:bg-slate-700 transition-colors"
-            onClick={() => setActiveSummaryModal('sellSignals')}
-          >
-            <div className="text-xl font-bold text-red-600 dark:text-red-400">
-              {filteredStats.sellSignals}
-            </div>
-            <div className="text-xs text-gray-500 dark:text-gray-400 flex items-center justify-center gap-1">
-              Sell Signals <HelpCircle className="w-3 h-3" />
-            </div>
-          </div>
-        </div>
+        )}
         <p className="text-[10px] text-gray-400 dark:text-gray-500 mt-3 text-center">
-          {summaryMode === 'topPicks'
-            ? 'Performance based on Top Picks recommendations only (3 stocks per day)'
-            : 'Next-day return = Price change on the trading day after recommendation'
+          {isBacktestDataLoading && !isUsingMockData
+            ? 'Loading performance data from market...'
+            : summaryMode === 'topPicks'
+              ? 'Performance based on Top Picks recommendations only (3 stocks per day)'
+              : 'Returns measured over hold period (or 1-day when no hold period specified)'
           }
         </p>
       </div>
@@ -1262,25 +1413,31 @@ export default function History() {
             <InvestmentModeToggle mode={indexChartMode} onChange={setIndexChartMode} />
           </div>
         </div>
-        <IndexComparisonChart
-          height={220}
-          data={isUsingMockData
-            ? undefined
-            : (indexChartMode === 'topPicks' ? topPicksCumulativeReturns : realCumulativeReturns) ?? []
-          }
-        />
-        <p className="text-[10px] text-gray-400 dark:text-gray-500 mt-2 text-center">
-          {(indexChartMode === 'topPicks' ? topPicksCumulativeReturns : realCumulativeReturns)?.length ? (
-            <>
-              Cumulative returns for {indexChartMode === 'topPicks' ? 'Top Picks' : 'All 50 stocks'} over{' '}
-              {(indexChartMode === 'topPicks' ? topPicksCumulativeReturns : realCumulativeReturns)?.length} trading days
-            </>
-          ) : isUsingMockData ? (
-            <>Demo data - Start backend for real performance comparison</>
-          ) : (
-            <>Comparison of cumulative returns between AI strategy and Nifty50 index</>
-          )}
-        </p>
+        {isBacktestDataLoading && !isUsingMockData ? (
+          <SectionLoader message="Computing cumulative returns vs Nifty50 index..." />
+        ) : (
+          <>
+            <IndexComparisonChart
+              height={220}
+              data={isUsingMockData
+                ? undefined
+                : (indexChartMode === 'topPicks' ? topPicksCumulativeReturns : realCumulativeReturns) ?? []
+              }
+            />
+            <p className="text-[10px] text-gray-400 dark:text-gray-500 mt-2 text-center">
+              {(indexChartMode === 'topPicks' ? topPicksCumulativeReturns : realCumulativeReturns)?.length ? (
+                <>
+                  Cumulative returns for {indexChartMode === 'topPicks' ? 'Top Picks' : 'All 50 stocks'} over{' '}
+                  {(indexChartMode === 'topPicks' ? topPicksCumulativeReturns : realCumulativeReturns)?.length} trading days
+                </>
+              ) : isUsingMockData ? (
+                <>Demo data - Start backend for real performance comparison</>
+              ) : (
+                <>Comparison of cumulative returns between AI strategy and Nifty50 index</>
+              )}
+            </p>
+          </>
+        )}
       </section>
 
       {/* Return Distribution */}
@@ -1300,22 +1457,28 @@ export default function History() {
             <InvestmentModeToggle mode={distributionMode} onChange={setDistributionMode} />
           </div>
         </div>
-        <ReturnDistributionChart
-          height={200}
-          data={isUsingMockData
-            ? undefined
-            : (distributionMode === 'topPicks' ? topPicksReturnDistribution : realReturnDistribution) ?? []
-          }
-        />
-        <p className="text-[10px] text-gray-400 dark:text-gray-500 mt-2 text-center">
-          {(distributionMode === 'topPicks' ? topPicksReturnDistribution : realReturnDistribution) ? (
-            <>Distribution of {distributionMode === 'topPicks' ? 'Top Picks' : 'all 50 stocks'} next-day returns. Click bars to see stocks.</>
-          ) : isUsingMockData ? (
-            <>Demo data - Start backend for real return distribution</>
-          ) : (
-            <>Distribution of next-day returns across all predictions. Click bars to see stocks.</>
-          )}
-        </p>
+        {isBacktestDataLoading && !isUsingMockData ? (
+          <SectionLoader message="Computing return distribution from backtest data..." />
+        ) : (
+          <>
+            <ReturnDistributionChart
+              height={200}
+              data={isUsingMockData
+                ? undefined
+                : (distributionMode === 'topPicks' ? topPicksReturnDistribution : realReturnDistribution) ?? []
+              }
+            />
+            <p className="text-[10px] text-gray-400 dark:text-gray-500 mt-2 text-center">
+              {(distributionMode === 'topPicks' ? topPicksReturnDistribution : realReturnDistribution) ? (
+                <>Distribution of {distributionMode === 'topPicks' ? 'Top Picks' : 'all 50 stocks'} hold-period returns. Click bars to see stocks.</>
+              ) : isUsingMockData ? (
+                <>Demo data - Start backend for real return distribution</>
+              ) : (
+                <>Distribution of hold-period returns across all predictions. Click bars to see stocks.</>
+              )}
+            </p>
+          </>
+        )}
       </section>
 
       {/* Accuracy Explanation Modal */}
@@ -1329,7 +1492,7 @@ export default function History() {
       <ReturnExplainModal
         isOpen={showReturnModal}
         onClose={() => setShowReturnModal(false)}
-        breakdown={returnModalDate ? (isUsingMockData ? getStaticReturnBreakdown(returnModalDate) : null) : null}
+        breakdown={returnModalDate ? (isUsingMockData ? getStaticReturnBreakdown(returnModalDate) : buildReturnBreakdown(returnModalDate)) : null}
         date={returnModalDate || ''}
       />
 
@@ -1361,19 +1524,20 @@ export default function History() {
       <InfoModal
         isOpen={activeSummaryModal === 'avgReturn'}
         onClose={() => setActiveSummaryModal(null)}
-        title="Average Next-Day Return"
+        title="Average Return"
         icon={<TrendingUp className="w-5 h-5 text-green-600 dark:text-green-400" />}
       >
         <div className="space-y-3 text-sm text-gray-600 dark:text-gray-300">
-          <p><strong>Average Next-Day Return</strong> measures the mean percentage price change one trading day after each recommendation.</p>
+          <p><strong>Average Return</strong> measures the mean percentage price change over each stock's recommended hold period.</p>
           <div className="p-3 bg-gray-100 dark:bg-slate-700 rounded-lg">
             <div className="font-semibold mb-1">How it's calculated:</div>
             <ol className="text-xs space-y-1 list-decimal list-inside">
               <li>Record stock price at recommendation time</li>
-              <li>Record price at next trading day close</li>
-              <li>Calculate: (Next Day Price - Rec Price) / Rec Price × 100</li>
-              <li>Average all these returns</li>
+              <li>Record price after the recommended hold period (e.g. 15 days)</li>
+              <li>Calculate: (Exit Price - Entry Price) / Entry Price × 100</li>
+              <li>Average all these returns across stocks</li>
             </ol>
+            <p className="text-xs text-gray-500 mt-2">If no hold period is specified, falls back to 1-day return.</p>
           </div>
           <div className={`p-3 ${filteredStats.avgDailyReturn >= 0 ? 'bg-green-50 dark:bg-green-900/20' : 'bg-red-50 dark:bg-red-900/20'} rounded-lg`}>
             <div className="text-xs text-gray-500 mb-1">Current Average:</div>

frontend/src/pages/StockDetail.tsx

@@ -27,6 +27,9 @@ interface BacktestResult {
   decision: string;
   return1d: number | null;
   return1w: number | null;
+  returnAtHold: number | null;
+  holdDays: number | null;
+  primaryReturn: number | null;  // return_at_hold ?? return_1d
   predictionCorrect: boolean | null;
   isLoading?: boolean;
 }
@@ -124,16 +127,14 @@ export default function StockDetail() {
         const backtest = await api.getBacktestResult(entry.date, symbol);
 
         if (backtest.available) {
-          // Calculate prediction correctness based on 1-day return
-          // BUY is correct if return > 0, HOLD is correct if return > 0, SELL is correct if return < 0
+          // Use hold-period return when available (BUY/HOLD with hold_days), else 1-day return
+          const primaryReturn = backtest.return_at_hold ?? backtest.actual_return_1d ?? null;
           let predictionCorrect: boolean | null = null;
-          if (backtest.actual_return_1d !== undefined && backtest.actual_return_1d !== null) {
+          if (primaryReturn !== undefined && primaryReturn !== null) {
             if (entry.decision === 'BUY' || entry.decision === 'HOLD') {
-              // BUY and HOLD are correct if stock price went up
-              predictionCorrect = backtest.actual_return_1d > 0;
+              predictionCorrect = primaryReturn > 0;
             } else if (entry.decision === 'SELL') {
-              // SELL is correct if stock price went down
-              predictionCorrect = backtest.actual_return_1d < 0;
+              predictionCorrect = primaryReturn < 0;
             }
           }
 
@@ -142,6 +143,9 @@ export default function StockDetail() {
             decision: entry.decision,
             return1d: backtest.actual_return_1d ?? null,
             return1w: backtest.actual_return_1w ?? null,
+            returnAtHold: backtest.return_at_hold ?? null,
+            holdDays: backtest.hold_days ?? null,
+            primaryReturn,
             predictionCorrect,
           });
         } else {
@@ -151,6 +155,9 @@ export default function StockDetail() {
             decision: entry.decision,
             return1d: null,
             return1w: null,
+            returnAtHold: null,
+            holdDays: null,
+            primaryReturn: null,
             predictionCorrect: null,
           });
         }
@@ -161,6 +168,9 @@ export default function StockDetail() {
           decision: entry.decision,
           return1d: null,
           return1w: null,
+          returnAtHold: null,
+          holdDays: null,
+          primaryReturn: null,
           predictionCorrect: null,
         });
       }
@@ -179,7 +189,7 @@ export default function StockDetail() {
   const predictionStats = useMemo((): PredictionStats | null => {
     if (backtestResults.length === 0) return null;
 
-    const resultsWithData = backtestResults.filter(r => r.return1d !== null);
+    const resultsWithData = backtestResults.filter(r => r.primaryReturn !== null);
     if (resultsWithData.length === 0) return null;
 
     let correct = 0;
@@ -189,8 +199,8 @@ export default function StockDetail() {
     let holdTotal = 0, holdCorrect = 0;
 
     for (const result of resultsWithData) {
-      if (result.return1d !== null) {
-        totalReturn += result.return1d;
+      if (result.primaryReturn !== null) {
+        totalReturn += result.primaryReturn;
       }
       if (result.predictionCorrect !== null) {
         if (result.predictionCorrect) correct++;
@@ -916,7 +926,7 @@ export default function StockDetail() {
                   )}
                 </div>
                 <span className="text-[10px] text-gray-500 dark:text-gray-400 bg-gray-100 dark:bg-slate-600 px-2 py-0.5 rounded-full">
-                  Real 1-Day Returns
+                  Actual Returns (Hold Period)
                 </span>
               </div>
             </div>
@@ -951,18 +961,20 @@ export default function StockDetail() {
                       )}
                     </div>
 
-                    {/* Outcome - 1 Day Return */}
-                    {entry.return1d !== null ? (
+                    {/* Outcome - Hold Period Return */}
+                    {entry.primaryReturn !== null ? (
                       <>
                         <div className="flex-1 flex items-center gap-2">
                           <div className={`text-sm font-semibold ${
-                            entry.return1d >= 0
+                            entry.primaryReturn >= 0
                               ? 'text-green-600 dark:text-green-400'
                               : 'text-red-600 dark:text-red-400'
                           }`}>
-                            {entry.return1d >= 0 ? '+' : ''}{entry.return1d.toFixed(1)}%
+                            {entry.primaryReturn >= 0 ? '+' : ''}{entry.primaryReturn.toFixed(1)}%
+                          </div>
+                          <div className="text-[10px] text-gray-400 dark:text-gray-500">
+                            {entry.holdDays && entry.holdDays > 0 ? `${entry.holdDays}d` : '1d'}
                           </div>
-                          <div className="text-[10px] text-gray-400 dark:text-gray-500">next day</div>
                         </div>
 
                         {/* Prediction Result Icon */}

frontend/src/services/api.ts

@@ -408,6 +408,8 @@ class ApiService {
       return_1d?: number;
       return_1w?: number;
       return_1m?: number;
+      return_at_hold?: number;
+      hold_days?: number;
       prediction_correct?: boolean;
     }>;
   }> {

frontend/src/types/pipeline.ts

@@ -211,6 +211,25 @@ export const DEBATE_ROLES = {
   }
 } as const;
 
+/**
+ * Maps each pipeline step to the IDs of steps whose output is forwarded as context.
+ * Steps 1-4 are independent (no forwarded context).
+ */
+export const STEP_INPUT_SOURCES: Record<string, string[]> = {
+  market_analyst: [],
+  social_analyst: [],
+  news_analyst: [],
+  fundamentals_analyst: [],
+  bull_researcher: ['market_analyst', 'social_analyst', 'news_analyst', 'fundamentals_analyst'],
+  bear_researcher: ['market_analyst', 'social_analyst', 'news_analyst', 'fundamentals_analyst'],
+  research_manager: ['bull_researcher', 'bear_researcher', 'market_analyst', 'social_analyst', 'news_analyst', 'fundamentals_analyst'],
+  trader: ['research_manager', 'market_analyst', 'social_analyst', 'news_analyst', 'fundamentals_analyst'],
+  aggressive_analyst: ['trader', 'market_analyst', 'social_analyst', 'news_analyst', 'fundamentals_analyst'],
+  conservative_analyst: ['trader', 'market_analyst', 'social_analyst', 'news_analyst', 'fundamentals_analyst'],
+  neutral_analyst: ['trader', 'market_analyst', 'social_analyst', 'news_analyst', 'fundamentals_analyst'],
+  risk_manager: ['aggressive_analyst', 'conservative_analyst', 'neutral_analyst', 'trader', 'market_analyst', 'social_analyst', 'news_analyst', 'fundamentals_analyst'],
+};
+
 // ============================================================
 // Flowchart types for the 12-step visual pipeline debug view
 // ============================================================
@@ -337,3 +356,14 @@ export function mapPipelineToFlowchart(data: FullPipelineData | null): Flowchart
     };
   });
 }
+
+/**
+ * Get human-readable labels for a step's input sources.
+ */
+export function getInputSourceLabels(stepId: string): { id: string; label: string }[] {
+  const sources = STEP_INPUT_SOURCES[stepId] || [];
+  return sources.map(id => {
+    const step = FLOWCHART_STEPS.find(s => s.id === id);
+    return { id, label: step?.label || id };
+  });
+}

tradingagents/agents/analysts/fundamentals_analyst.py

@@ -1,7 +1,7 @@
 from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
 import time
 import json
-from tradingagents.agents.utils.agent_utils import get_fundamentals, get_balance_sheet, get_cashflow, get_income_statement, get_insider_sentiment, get_insider_transactions
+from tradingagents.agents.utils.agent_utils import get_fundamentals, get_balance_sheet, get_cashflow, get_income_statement, get_insider_sentiment, get_insider_transactions, execute_text_tool_calls, needs_followup_call, execute_default_tools, generate_analysis_from_data
 from tradingagents.dataflows.config import get_config
 
 from tradingagents.log_utils import add_log, step_timer, symbol_progress
@@ -70,6 +70,23 @@ def create_fundamentals_analyst(llm):
         if len(result.tool_calls) == 0:
             report = result.content
             add_log("llm", "fundamentals", f"LLM responded in {elapsed:.1f}s ({len(report)} chars)")
+            tool_results = execute_text_tool_calls(report, tools)
+            if tool_results:
+                add_log("data", "fundamentals", f"Executed {len(tool_results)} tool calls: {', '.join(t['name'] for t in tool_results)}")
+            else:
+                add_log("agent", "fundamentals", f"🔄 No tool calls found, proactively fetching data for {ticker}...")
+                tool_results = execute_default_tools(tools, ticker, current_date)
+                add_log("data", "fundamentals", f"Proactively fetched {len(tool_results)} data sources")
+
+            if tool_results and needs_followup_call(report):
+                add_log("agent", "fundamentals", f"🔄 Generating analysis from {len(tool_results)} tool results...")
+                t1 = time.time()
+                followup = generate_analysis_from_data(llm, tool_results, system_message, ticker, current_date)
+                elapsed2 = time.time() - t1
+                if followup and len(followup) > 100:
+                    report = followup
+                    add_log("llm", "fundamentals", f"Follow-up analysis generated in {elapsed2:.1f}s ({len(report)} chars)")
+
             add_log("agent", "fundamentals", f"✅ Fundamentals report ready: {report[:300]}...")
             step_timer.end_step("fundamentals_analyst", "completed", report[:200])
             symbol_progress.step_done(ticker, "fundamentals_analyst")
@@ -77,6 +94,7 @@ def create_fundamentals_analyst(llm):
                 "system_prompt": system_message[:2000],
                 "user_prompt": f"Analyze fundamentals for {ticker} on {current_date}",
                 "response": report[:3000],
+                "tool_calls": tool_results if tool_results else [],
             })
         else:
             tool_call_info = [{"name": tc["name"], "args": str(tc.get("args", {}))[:200]} for tc in result.tool_calls]

tradingagents/agents/analysts/market_analyst.py

@@ -1,7 +1,7 @@
 from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
 import time
 import json
-from tradingagents.agents.utils.agent_utils import get_stock_data, get_indicators
+from tradingagents.agents.utils.agent_utils import get_stock_data, get_indicators, execute_text_tool_calls, needs_followup_call, execute_default_tools, generate_analysis_from_data
 from tradingagents.dataflows.config import get_config
 
 from tradingagents.log_utils import add_log, step_timer, symbol_progress
@@ -93,14 +93,34 @@ Volume-Based Indicators:
         if len(result.tool_calls) == 0:
             report = result.content
             add_log("llm", "market_analyst", f"LLM responded in {elapsed:.1f}s ({len(report)} chars)")
+            # Execute any text-based tool calls and capture results
+            tool_results = execute_text_tool_calls(report, tools)
+            if tool_results:
+                add_log("data", "market_analyst", f"Executed {len(tool_results)} tool calls: {', '.join(t['name'] for t in tool_results)}")
+            else:
+                # LLM didn't produce TOOL_CALL patterns — proactively fetch data
+                add_log("agent", "market_analyst", f"🔄 No tool calls found, proactively fetching data for {ticker}...")
+                tool_results = execute_default_tools(tools, ticker, current_date)
+                add_log("data", "market_analyst", f"Proactively fetched {len(tool_results)} data sources")
+
+            # If report is mostly tool calls / thin prose, make follow-up LLM call with actual data
+            if tool_results and needs_followup_call(report):
+                add_log("agent", "market_analyst", f"🔄 Generating analysis from {len(tool_results)} tool results...")
+                t1 = time.time()
+                followup = generate_analysis_from_data(llm, tool_results, system_message, ticker, current_date)
+                elapsed2 = time.time() - t1
+                if followup and len(followup) > 100:
+                    report = followup
+                    add_log("llm", "market_analyst", f"Follow-up analysis generated in {elapsed2:.1f}s ({len(report)} chars)")
+
             add_log("agent", "market_analyst", f"✅ Market report ready: {report[:300]}...")
             step_timer.end_step("market_analyst", "completed", report[:200])
             symbol_progress.step_done(ticker, "market_analyst")
-            # Use update_details to preserve tool_calls from previous invocation
             step_timer.update_details("market_analyst", {
                 "system_prompt": system_message[:2000],
                 "user_prompt": f"Analyze {ticker} on {current_date} using technical indicators",
                 "response": report[:3000],
+                "tool_calls": tool_results if tool_results else [],
             })
         else:
             tool_call_info = [{"name": tc["name"], "args": str(tc.get("args", {}))[:200]} for tc in result.tool_calls]

tradingagents/agents/analysts/news_analyst.py

@@ -1,7 +1,7 @@
 from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
 import time
 import json
-from tradingagents.agents.utils.agent_utils import get_news, get_global_news
+from tradingagents.agents.utils.agent_utils import get_news, get_global_news, execute_text_tool_calls, needs_followup_call, execute_default_tools, generate_analysis_from_data
 from tradingagents.dataflows.config import get_config
 
 from tradingagents.log_utils import add_log, step_timer, symbol_progress
@@ -66,6 +66,23 @@ def create_news_analyst(llm):
         if len(result.tool_calls) == 0:
             report = result.content
             add_log("llm", "news_analyst", f"LLM responded in {elapsed:.1f}s ({len(report)} chars)")
+            tool_results = execute_text_tool_calls(report, tools)
+            if tool_results:
+                add_log("data", "news_analyst", f"Executed {len(tool_results)} tool calls: {', '.join(t['name'] for t in tool_results)}")
+            else:
+                add_log("agent", "news_analyst", f"🔄 No tool calls found, proactively fetching data for {ticker}...")
+                tool_results = execute_default_tools(tools, ticker, current_date)
+                add_log("data", "news_analyst", f"Proactively fetched {len(tool_results)} data sources")
+
+            if tool_results and needs_followup_call(report):
+                add_log("agent", "news_analyst", f"🔄 Generating analysis from {len(tool_results)} tool results...")
+                t1 = time.time()
+                followup = generate_analysis_from_data(llm, tool_results, system_message, ticker, current_date)
+                elapsed2 = time.time() - t1
+                if followup and len(followup) > 100:
+                    report = followup
+                    add_log("llm", "news_analyst", f"Follow-up analysis generated in {elapsed2:.1f}s ({len(report)} chars)")
+
             add_log("agent", "news_analyst", f"✅ News report ready: {report[:300]}...")
             step_timer.end_step("news_analyst", "completed", report[:200])
             symbol_progress.step_done(ticker, "news_analyst")
@@ -73,6 +90,7 @@ def create_news_analyst(llm):
                 "system_prompt": system_message[:2000],
                 "user_prompt": f"Analyze news and macro trends for {ticker} on {current_date}",
                 "response": report[:3000],
+                "tool_calls": tool_results if tool_results else [],
             })
         else:
             tool_call_info = [{"name": tc["name"], "args": str(tc.get("args", {}))[:200]} for tc in result.tool_calls]

tradingagents/agents/analysts/social_media_analyst.py

@@ -1,7 +1,7 @@
 from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
 import time
 import json
-from tradingagents.agents.utils.agent_utils import get_news
+from tradingagents.agents.utils.agent_utils import get_news, execute_text_tool_calls, needs_followup_call, execute_default_tools, generate_analysis_from_data
 from tradingagents.dataflows.config import get_config
 
 from tradingagents.log_utils import add_log, step_timer, symbol_progress
@@ -66,6 +66,23 @@ def create_social_media_analyst(llm):
         if len(result.tool_calls) == 0:
             report = result.content
             add_log("llm", "social_analyst", f"LLM responded in {elapsed:.1f}s ({len(report)} chars)")
+            tool_results = execute_text_tool_calls(report, tools)
+            if tool_results:
+                add_log("data", "social_analyst", f"Executed {len(tool_results)} tool calls: {', '.join(t['name'] for t in tool_results)}")
+            else:
+                add_log("agent", "social_analyst", f"🔄 No tool calls found, proactively fetching data for {ticker}...")
+                tool_results = execute_default_tools(tools, ticker, current_date)
+                add_log("data", "social_analyst", f"Proactively fetched {len(tool_results)} data sources")
+
+            if tool_results and needs_followup_call(report):
+                add_log("agent", "social_analyst", f"🔄 Generating analysis from {len(tool_results)} tool results...")
+                t1 = time.time()
+                followup = generate_analysis_from_data(llm, tool_results, system_message, ticker, current_date)
+                elapsed2 = time.time() - t1
+                if followup and len(followup) > 100:
+                    report = followup
+                    add_log("llm", "social_analyst", f"Follow-up analysis generated in {elapsed2:.1f}s ({len(report)} chars)")
+
             add_log("agent", "social_analyst", f"✅ Sentiment report ready: {report[:300]}...")
             step_timer.end_step("social_media_analyst", "completed", report[:200])
             symbol_progress.step_done(ticker, "social_media_analyst")
@@ -73,6 +90,7 @@ def create_social_media_analyst(llm):
                 "system_prompt": system_message[:2000],
                 "user_prompt": f"Analyze social media sentiment for {ticker} on {current_date}",
                 "response": report[:3000],
+                "tool_calls": tool_results if tool_results else [],
             })
         else:
             tool_call_info = [{"name": tc["name"], "args": str(tc.get("args", {}))[:200]} for tc in result.tool_calls]

tradingagents/agents/managers/risk_manager.py

@@ -14,7 +14,7 @@ def create_risk_manager(llm, memory):
         risk_debate_state = state["risk_debate_state"]
         market_research_report = state["market_report"]
         news_report = state["news_report"]
-        fundamentals_report = state["news_report"]
+        fundamentals_report = state["fundamentals_report"]
         sentiment_report = state["sentiment_report"]
         trader_plan = state["investment_plan"]
 
@@ -39,8 +39,10 @@ Your task: Evaluate the risk debate between Aggressive, Neutral, and Conservativ
 Your response must include:
 1. FINAL DECISION: BUY, SELL, or HOLD
 2. HOLD_DAYS: Number of trading days to hold the position before exiting (for BUY/HOLD only, write N/A for SELL)
-3. RISK ASSESSMENT: Summary of key risks identified
-4. RATIONALE: Why this decision balances risk and reward appropriately
+3. CONFIDENCE: HIGH, MEDIUM, or LOW (how confident you are in this decision)
+4. RISK_LEVEL: HIGH, MEDIUM, or LOW (overall risk level of this investment)
+5. RISK ASSESSMENT: Summary of key risks identified
+6. RATIONALE: Why this decision balances risk and reward appropriately
 
 RESPONSE FORMAT:
 - Maximum 1500 characters. Lead with your decision, then key rationale.

tradingagents/agents/utils/agent_utils.py

@@ -10,6 +10,8 @@ from langchain_core.messages import HumanMessage, RemoveMessage
 
 SIMULATION_CONTEXT = """You are an AI analyst participating in a stock market analysis simulation. Your task is to analyze financial data and provide investment perspectives based on the data provided. This is an educational demonstration of financial analysis techniques.
 
+Be precise, analytical, and data-driven. Base all conclusions on quantitative evidence from the data provided. Avoid speculative language — present findings with specific numbers and clear logical reasoning. When uncertain, state the uncertainty rather than guessing.
+
 Respond ONLY with your analysis. Do not include any meta-commentary about being an AI, ethical concerns, or disclaimers. Simply provide the requested financial analysis based on the data given."""
 
 def get_simulation_prompt(role_prompt: str) -> list:
@@ -44,6 +46,171 @@ from tradingagents.agents.utils.news_data_tools import (
     get_global_news
 )
 
+def strip_tool_call_lines(text):
+    """Remove TOOL_CALL: lines from text, returning only the prose."""
+    import re
+    return re.sub(r'TOOL_CALL:\s*\w+\([^)]*\)\s*\n?', '', text).strip()
+
+
+def needs_followup_call(report):
+    """Check if the report is mostly tool calls and needs a follow-up LLM call."""
+    clean = strip_tool_call_lines(report)
+    return len(clean) < 300
+
+
+def execute_default_tools(tools, ticker, current_date):
+    """
+    Proactively call all available tools with sensible default arguments.
+    Used when the LLM fails to produce TOOL_CALL patterns.
+    """
+    from datetime import datetime, timedelta
+
+    end_dt = datetime.strptime(current_date, "%Y-%m-%d")
+    week_ago = (end_dt - timedelta(days=7)).strftime("%Y-%m-%d")
+    three_months_ago = (end_dt - timedelta(days=90)).strftime("%Y-%m-%d")
+
+    tool_map = {t.name: t for t in tools}
+    default_configs = {
+        "get_stock_data": {"symbol": ticker, "start_date": three_months_ago, "end_date": current_date},
+        "get_indicators": [
+            {"symbol": ticker, "indicator": "rsi", "curr_date": current_date, "look_back_days": 90},
+            {"symbol": ticker, "indicator": "macd", "curr_date": current_date, "look_back_days": 90},
+            {"symbol": ticker, "indicator": "close_50_sma", "curr_date": current_date, "look_back_days": 90},
+            {"symbol": ticker, "indicator": "boll_ub", "curr_date": current_date, "look_back_days": 90},
+            {"symbol": ticker, "indicator": "atr", "curr_date": current_date, "look_back_days": 90},
+        ],
+        "get_news": {"ticker": ticker, "start_date": week_ago, "end_date": current_date},
+        "get_global_news": {"curr_date": current_date, "look_back_days": 7, "limit": 5},
+        "get_fundamentals": {"ticker": ticker, "curr_date": current_date},
+        "get_balance_sheet": {"ticker": ticker, "curr_date": current_date},
+        "get_cashflow": {"ticker": ticker, "curr_date": current_date},
+        "get_income_statement": {"ticker": ticker, "curr_date": current_date},
+    }
+
+    results = []
+    for tool in tools:
+        config = default_configs.get(tool.name)
+        if config is None:
+            continue
+        # Handle tools that need multiple calls (e.g., get_indicators with different indicators)
+        calls = config if isinstance(config, list) else [config]
+        for args in calls:
+            try:
+                result = tool.invoke(args)
+                results.append({
+                    "name": tool.name,
+                    "args": str(args),
+                    "result_preview": str(result)[:1500],
+                })
+            except Exception as e:
+                results.append({
+                    "name": tool.name,
+                    "args": str(args),
+                    "result_preview": f"[Tool error: {str(e)[:200]}]",
+                })
+    return results
+
+
+def generate_analysis_from_data(llm, tool_results, system_message, ticker, current_date):
+    """
+    Make a follow-up LLM call with actual tool data to generate the analysis.
+    Called when the first LLM response was mostly tool call requests without analysis.
+    """
+    data_sections = []
+    for r in tool_results:
+        preview = r.get('result_preview', '')
+        if not preview:
+            data_sections.append(f"### {r['name']}({r['args']})\n(No data returned — tool executed but returned empty result)")
+        elif preview.startswith('[Tool error') or preview.startswith('[Could not') or preview.startswith('[Unknown'):
+            data_sections.append(f"### {r['name']}({r['args']})\nError: {preview}")
+        else:
+            data_sections.append(f"### {r['name']}({r['args']})\n```\n{preview}\n```")
+
+    if not data_sections:
+        return ""
+
+    data_text = "\n\n".join(data_sections)
+
+    message = f"""Here are the results from the data retrieval tools for {ticker} as of {current_date}:
+
+{data_text}
+
+Based on this data, write your comprehensive analysis.
+
+{system_message}
+
+IMPORTANT:
+- Write your analysis directly based on the data above
+- Do NOT request any more tool calls or use TOOL_CALL syntax
+- Provide detailed, actionable insights with specific numbers from the data
+- Include a Markdown table summarizing key findings"""
+
+    result = llm.invoke([HumanMessage(content=message)])
+    return result.content
+
+
+def execute_text_tool_calls(response_text, tools):
+    """
+    Parse TOOL_CALL: patterns from LLM response text, execute the actual
+    tool functions, and return structured results.
+
+    Args:
+        response_text: Raw LLM response that may contain TOOL_CALL: patterns
+        tools: List of @tool-decorated LangChain tool objects available for this agent
+
+    Returns:
+        List of dicts with {name, args, result_preview} for each executed tool call
+    """
+    import re
+    import ast
+
+    tool_map = {t.name: t for t in tools}
+    regex = re.compile(r'TOOL_CALL:\s*(\w+)\(([^)]*)\)')
+    results = []
+
+    for match in regex.finditer(response_text):
+        fn_name = match.group(1)
+        raw_args = match.group(2).strip()
+        tool_fn = tool_map.get(fn_name)
+
+        if not tool_fn:
+            results.append({
+                "name": fn_name,
+                "args": raw_args,
+                "result_preview": f"[Unknown tool: {fn_name}]",
+            })
+            continue
+
+        # Parse positional args and map to parameter names
+        try:
+            parsed = ast.literal_eval(f"({raw_args},)")  # tuple of values
+            param_names = list(tool_fn.args_schema.model_fields.keys())
+            invoke_args = {}
+            for i, val in enumerate(parsed):
+                if i < len(param_names):
+                    invoke_args[param_names[i]] = val
+        except Exception:
+            invoke_args = None
+
+        # Execute the tool
+        result_text = ""
+        try:
+            if invoke_args:
+                result_text = tool_fn.invoke(invoke_args)
+            else:
+                result_text = f"[Could not parse args: {raw_args}]"
+        except Exception as e:
+            result_text = f"[Tool error: {str(e)[:200]}]"
+
+        results.append({
+            "name": fn_name,
+            "args": raw_args,
+            "result_preview": str(result_text)[:1500],
+        })
+
+    return results
+
+
 def create_msg_delete():
     def delete_messages(state):
         """Clear messages and add placeholder for Anthropic compatibility"""

tradingagents/claude_max_llm.py

@@ -36,7 +36,7 @@ class ClaudeMaxLLM(BaseChatModel):
 
     model: str = "sonnet"  # Use alias for Claude Max subscription
     max_tokens: int = 4096
-    temperature: float = 0.7
+    temperature: float = 0.2
     claude_cli_path: str = "claude"
     tools: List[Any] = []  # Bound tools
 

tradingagents/dataflows/google.py

@@ -1,7 +1,7 @@
 from typing import Annotated, Union
 from datetime import datetime
 from dateutil.relativedelta import relativedelta
-from .googlenews_utils import getNewsData
+from .googlenews_utils import getNewsData, getGlobalNewsData
 from .markets import is_nifty_50_stock, get_nifty_50_company_name
 
 
@@ -65,4 +65,22 @@ def get_google_news(
 
     # Use original query (symbol) in the header for clarity
     display_query = original_query if is_nifty_50_stock(original_query) else query.replace("+", " ")
-    return f"## {display_query} Google News, from {before} to {curr_date}:\n\n{news_str}"
+    return f"## {display_query} Google News, from {before} to {curr_date}:\n\n{news_str}"
+
+
+def get_google_global_news(
+    curr_date: Annotated[str, "Current date in yyyy-mm-dd format"],
+    look_back_days: Annotated[int, "How many days to look back"] = 7,
+    limit: Annotated[int, "Maximum number of news items to return"] = 10,
+) -> str:
+    """Fetch global/macro financial news via Google News RSS feed."""
+    news_results = getGlobalNewsData(curr_date, look_back_days=look_back_days, limit=limit)
+
+    if not news_results:
+        return ""
+
+    news_str = ""
+    for news in news_results:
+        news_str += f"### {news['title']} (source: {news['source']})\n\n{news['snippet']}\n\n"
+
+    return f"## Global Market News (past {look_back_days} days as of {curr_date}):\n\n{news_str}"

tradingagents/dataflows/googlenews_utils.py

@@ -1,108 +1,157 @@
-import json
 import requests
 from bs4 import BeautifulSoup
 from datetime import datetime
-import time
-import random
-from tenacity import (
-    retry,
-    stop_after_attempt,
-    wait_exponential,
-    retry_if_exception_type,
-    retry_if_result,
-)
-
-
-def is_rate_limited(response):
-    """Check if the response indicates rate limiting (status code 429)"""
-    return response.status_code == 429
-
-
-@retry(
-    retry=(retry_if_result(is_rate_limited)),
-    wait=wait_exponential(multiplier=1, min=4, max=60),
-    stop=stop_after_attempt(5),
-)
-def make_request(url, headers):
-    """Make a request with retry logic for rate limiting"""
-    # Random delay before each request to avoid detection
-    time.sleep(random.uniform(2, 6))
-    response = requests.get(url, headers=headers)
-    return response
+import urllib.parse
 
 
 def getNewsData(query, start_date, end_date):
     """
-    Scrape Google News search results for a given query and date range.
-    query: str - search query
-    start_date: str - start date in the format yyyy-mm-dd or mm/dd/yyyy
-    end_date: str - end date in the format yyyy-mm-dd or mm/dd/yyyy
-    """
-    if "-" in start_date:
-        start_date = datetime.strptime(start_date, "%Y-%m-%d")
-        start_date = start_date.strftime("%m/%d/%Y")
-    if "-" in end_date:
-        end_date = datetime.strptime(end_date, "%Y-%m-%d")
-        end_date = end_date.strftime("%m/%d/%Y")
+    Fetch Google News via RSS feed for a given query and date range.
 
-    headers = {
-        "User-Agent": (
-            "Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
-            "AppleWebKit/537.36 (KHTML, like Gecko) "
-            "Chrome/101.0.4951.54 Safari/537.36"
-        )
-    }
+    Uses Google News RSS which is reliable (no JS rendering or CSS selectors needed).
+    Results are filtered to only include articles within the date range.
+
+    query: str - search query (spaces or '+' separated)
+    start_date: str - start date in yyyy-mm-dd or mm/dd/yyyy format
+    end_date: str - end date in yyyy-mm-dd or mm/dd/yyyy format
+    """
+    # Normalize dates to datetime objects for filtering
+    if "/" in str(start_date):
+        start_dt = datetime.strptime(start_date, "%m/%d/%Y")
+    else:
+        start_dt = datetime.strptime(start_date, "%Y-%m-%d")
+
+    if "/" in str(end_date):
+        end_dt = datetime.strptime(end_date, "%m/%d/%Y")
+    else:
+        end_dt = datetime.strptime(end_date, "%Y-%m-%d")
+
+    # Clean up query (replace + with spaces for URL encoding)
+    clean_query = query.replace("+", " ")
+    encoded_query = urllib.parse.quote(clean_query)
+
+    # Use Google News RSS feed — reliable, no scraping issues
+    url = f"https://news.google.com/rss/search?q={encoded_query}+after:{start_dt.strftime('%Y-%m-%d')}+before:{end_dt.strftime('%Y-%m-%d')}&hl=en-IN&gl=IN&ceid=IN:en"
 
     news_results = []
-    page = 0
-    while True:
-        offset = page * 10
-        url = (
-            f"https://www.google.com/search?q={query}"
-            f"&tbs=cdr:1,cd_min:{start_date},cd_max:{end_date}"
-            f"&tbm=nws&start={offset}"
-        )
+    try:
+        resp = requests.get(url, timeout=15)
+        if resp.status_code != 200:
+            return news_results
 
-        try:
-            response = make_request(url, headers)
-            soup = BeautifulSoup(response.content, "html.parser")
-            results_on_page = soup.select("div.SoaBEf")
+        soup = BeautifulSoup(resp.content, "xml")
+        items = soup.find_all("item")
 
-            if not results_on_page:
-                break  # No more results found
+        for item in items[:20]:  # Limit to 20 articles
+            try:
+                title = item.find("title").text if item.find("title") else ""
+                pub_date_str = item.find("pubDate").text if item.find("pubDate") else ""
+                source = item.find("source").text if item.find("source") else ""
+                link = item.find("link").text if item.find("link") else ""
+                # Description often contains HTML snippet
+                desc_tag = item.find("description")
+                snippet = ""
+                if desc_tag:
+                    desc_soup = BeautifulSoup(desc_tag.text, "html.parser")
+                    snippet = desc_soup.get_text()[:300]
 
-            for el in results_on_page:
-                try:
-                    link = el.find("a")["href"]
-                    title = el.select_one("div.MBeuO").get_text()
-                    snippet = el.select_one(".GI74Re").get_text()
-                    date = el.select_one(".LfVVr").get_text()
-                    source = el.select_one(".NUnG9d span").get_text()
-                    news_results.append(
-                        {
-                            "link": link,
-                            "title": title,
-                            "snippet": snippet,
-                            "date": date,
-                            "source": source,
-                        }
-                    )
-                except Exception as e:
-                    print(f"Error processing result: {e}")
-                    # If one of the fields is not found, skip this result
-                    continue
+                # Parse and filter by date
+                if pub_date_str:
+                    try:
+                        pub_dt = datetime.strptime(pub_date_str, "%a, %d %b %Y %H:%M:%S %Z")
+                        if pub_dt.date() < start_dt.date() or pub_dt.date() > end_dt.date():
+                            continue
+                        date_display = pub_dt.strftime("%Y-%m-%d")
+                    except ValueError:
+                        date_display = pub_date_str
+                else:
+                    date_display = ""
 
-            # Update the progress bar with the current count of results scraped
+                news_results.append({
+                    "link": link,
+                    "title": title,
+                    "snippet": snippet if snippet else title,
+                    "date": date_display,
+                    "source": source,
+                })
+            except Exception:
+                continue
 
-            # Check for the "Next" link (pagination)
-            next_link = soup.find("a", id="pnnext")
-            if not next_link:
-                break
-
-            page += 1
-
-        except Exception as e:
-            print(f"Failed after multiple retries: {e}")
-            break
+    except Exception as e:
+        print(f"Google News RSS fetch failed: {e}")
 
     return news_results
+
+
+def getGlobalNewsData(curr_date, look_back_days=7, limit=10):
+    """
+    Fetch global/macro news via Google News RSS feed.
+
+    Uses broad financial/market queries to get macroeconomic news.
+    """
+    if isinstance(curr_date, str):
+        end_dt = datetime.strptime(curr_date, "%Y-%m-%d")
+    else:
+        end_dt = curr_date
+
+    from dateutil.relativedelta import relativedelta
+    start_dt = end_dt - relativedelta(days=look_back_days)
+
+    queries = [
+        "stock market India NSE Nifty",
+        "global economy markets finance",
+    ]
+
+    all_results = []
+    seen_titles = set()
+
+    for query in queries:
+        encoded = urllib.parse.quote(query)
+        url = f"https://news.google.com/rss/search?q={encoded}+after:{start_dt.strftime('%Y-%m-%d')}+before:{end_dt.strftime('%Y-%m-%d')}&hl=en-IN&gl=IN&ceid=IN:en"
+
+        try:
+            resp = requests.get(url, timeout=15)
+            if resp.status_code != 200:
+                continue
+
+            soup = BeautifulSoup(resp.content, "xml")
+            items = soup.find_all("item")
+
+            for item in items:
+                try:
+                    title = item.find("title").text if item.find("title") else ""
+                    if title in seen_titles:
+                        continue
+                    seen_titles.add(title)
+
+                    pub_date_str = item.find("pubDate").text if item.find("pubDate") else ""
+                    source = item.find("source").text if item.find("source") else ""
+                    desc_tag = item.find("description")
+                    snippet = ""
+                    if desc_tag:
+                        desc_soup = BeautifulSoup(desc_tag.text, "html.parser")
+                        snippet = desc_soup.get_text()[:300]
+
+                    date_display = ""
+                    if pub_date_str:
+                        try:
+                            pub_dt = datetime.strptime(pub_date_str, "%a, %d %b %Y %H:%M:%S %Z")
+                            date_display = pub_dt.strftime("%Y-%m-%d")
+                        except ValueError:
+                            date_display = pub_date_str
+
+                    all_results.append({
+                        "title": title,
+                        "snippet": snippet if snippet else title,
+                        "date": date_display,
+                        "source": source,
+                    })
+                except Exception:
+                    continue
+
+        except Exception:
+            continue
+
+    # Sort by date descending and limit
+    all_results.sort(key=lambda x: x.get("date", ""), reverse=True)
+    return all_results[:limit]

tradingagents/dataflows/interface.py

@@ -42,8 +42,8 @@ def clear_request_cache():
 
 # Import from vendor-specific modules
 from .local import get_YFin_data, get_finnhub_news, get_finnhub_company_insider_sentiment, get_finnhub_company_insider_transactions, get_simfin_balance_sheet, get_simfin_cashflow, get_simfin_income_statements, get_reddit_global_news, get_reddit_company_news
-from .y_finance import get_YFin_data_online, get_stock_stats_indicators_window, get_balance_sheet as get_yfinance_balance_sheet, get_cashflow as get_yfinance_cashflow, get_income_statement as get_yfinance_income_statement, get_insider_transactions as get_yfinance_insider_transactions
-from .google import get_google_news
+from .y_finance import get_YFin_data_online, get_stock_stats_indicators_window, get_balance_sheet as get_yfinance_balance_sheet, get_cashflow as get_yfinance_cashflow, get_income_statement as get_yfinance_income_statement, get_insider_transactions as get_yfinance_insider_transactions, get_fundamentals as get_yfinance_fundamentals
+from .google import get_google_news, get_google_global_news
 from .openai import get_stock_news_openai, get_global_news_openai, get_fundamentals_openai
 from .alpha_vantage import (
     get_stock as get_alpha_vantage_stock,
@@ -122,6 +122,7 @@ VENDOR_METHODS = {
     },
     # fundamental_data
     "get_fundamentals": {
+        "yfinance": get_yfinance_fundamentals,
         "alpha_vantage": get_alpha_vantage_fundamentals,
         "openai": get_fundamentals_openai,
     },
@@ -148,8 +149,9 @@ VENDOR_METHODS = {
         "local": [get_finnhub_news, get_reddit_company_news, get_google_news],
     },
     "get_global_news": {
+        "google": get_google_global_news,
         "openai": get_global_news_openai,
-        "local": get_reddit_global_news
+        "local": get_reddit_global_news,
     },
     "get_insider_sentiment": {
         "local": get_finnhub_company_insider_sentiment

tradingagents/dataflows/y_finance.py

@@ -461,6 +461,63 @@ def get_income_statement(
         return f"Error retrieving income statement for {normalized_ticker}: {str(e)}"
 
 
+def get_fundamentals(
+    ticker: Annotated[str, "ticker symbol of the company"],
+    curr_date: Annotated[str, "current date for reference"] = None,
+) -> str:
+    """Get comprehensive company fundamentals from yfinance (.info)."""
+    try:
+        normalized_ticker = normalize_symbol(ticker, target="yfinance")
+        ticker_obj = yf.Ticker(normalized_ticker)
+        info = ticker_obj.info
+
+        if not info or len(info) < 5:
+            return f"No fundamentals data found for symbol '{normalized_ticker}'"
+
+        # Select the most useful keys for analysis
+        key_groups = {
+            "Valuation": ["marketCap", "enterpriseValue", "trailingPE", "forwardPE",
+                          "priceToBook", "priceToSalesTrailing12Months", "enterpriseToRevenue",
+                          "enterpriseToEbitda"],
+            "Profitability": ["profitMargins", "operatingMargins", "grossMargins",
+                              "returnOnAssets", "returnOnEquity", "revenueGrowth",
+                              "earningsGrowth", "earningsQuarterlyGrowth"],
+            "Dividends": ["dividendRate", "dividendYield", "payoutRatio",
+                          "fiveYearAvgDividendYield", "trailingAnnualDividendRate"],
+            "Financial Health": ["totalCash", "totalDebt", "debtToEquity",
+                                 "currentRatio", "quickRatio", "freeCashflow",
+                                 "operatingCashflow", "totalRevenue", "ebitda"],
+            "Trading": ["currentPrice", "targetHighPrice", "targetLowPrice",
+                        "targetMeanPrice", "recommendationKey", "numberOfAnalystOpinions",
+                        "fiftyTwoWeekHigh", "fiftyTwoWeekLow", "fiftyDayAverage",
+                        "twoHundredDayAverage", "beta", "volume", "averageVolume"],
+            "Company Info": ["sector", "industry", "fullTimeEmployees", "country", "city"],
+        }
+
+        sections = []
+        sections.append(f"# Fundamentals for {normalized_ticker}")
+        if curr_date:
+            sections.append(f"# As of: {curr_date}")
+        sections.append(f"# Company: {info.get('longName', info.get('shortName', ticker))}")
+        sections.append("")
+
+        for group_name, keys in key_groups.items():
+            group_lines = []
+            for key in keys:
+                val = info.get(key)
+                if val is not None:
+                    group_lines.append(f"  {key}: {val}")
+            if group_lines:
+                sections.append(f"## {group_name}")
+                sections.extend(group_lines)
+                sections.append("")
+
+        return "\n".join(sections)
+
+    except Exception as e:
+        return f"Error retrieving fundamentals for {ticker}: {str(e)}"
+
+
 def get_insider_transactions(
     ticker: Annotated[str, "ticker symbol of the company"]
 ):

tradingagents/default_config.py

@@ -13,6 +13,7 @@ DEFAULT_CONFIG = {
     "deep_think_llm": "o4-mini",
     "quick_think_llm": "gpt-4o-mini",
     "backend_url": "https://api.openai.com/v1",
+    "llm_temperature": 0.2,  # Low temperature for deterministic financial analysis
     # Anthropic-specific config for Claude models (using aliases for Claude Max subscription)
     "anthropic_config": {
         "deep_think_llm": "opus",  # Claude Opus 4.5 for deep analysis

tradingagents/graph/signal_processing.py

@@ -1,5 +1,6 @@
 # TradingAgents/graph/signal_processing.py
 
+import re
 from langchain_openai import ChatOpenAI
 
 
@@ -12,37 +13,97 @@ class SignalProcessor:
 
     def process_signal(self, full_signal: str) -> dict:
         """
-        Process a full trading signal to extract the core decision and hold_days.
+        Process a full trading signal to extract the core decision, hold_days,
+        confidence, and risk level.
 
         Args:
             full_signal: Complete trading signal text
 
         Returns:
-            Dict with 'decision' (BUY/SELL/HOLD) and 'hold_days' (int or None)
+            Dict with 'decision', 'hold_days', 'confidence', 'risk'
         """
         messages = [
             (
                 "system",
                 "You are an efficient assistant designed to analyze paragraphs or financial reports "
-                "provided by a group of analysts. Extract two pieces of information:\n"
+                "provided by a group of analysts. Extract the following information:\n"
                 "1. The investment decision: SELL, BUY, or HOLD\n"
-                "2. The recommended holding period in trading days (only for BUY or HOLD decisions)\n\n"
+                "2. The recommended holding period in trading days (only for BUY or HOLD decisions)\n"
+                "3. The confidence level of the decision: HIGH, MEDIUM, or LOW\n"
+                "4. The risk level of the investment: HIGH, MEDIUM, or LOW\n\n"
                 "Respond in exactly this format (nothing else):\n"
                 "DECISION: <BUY|SELL|HOLD>\n"
-                "HOLD_DAYS: <number|N/A>\n\n"
+                "HOLD_DAYS: <number|N/A>\n"
+                "CONFIDENCE: <HIGH|MEDIUM|LOW>\n"
+                "RISK_LEVEL: <HIGH|MEDIUM|LOW>\n\n"
                 "For SELL decisions, always use HOLD_DAYS: N/A\n"
-                "For BUY or HOLD decisions, extract the number of days if mentioned, otherwise default to 5.",
+                "For BUY or HOLD decisions, extract the EXACT number of days mentioned in the report. "
+                "Look for phrases like 'N-day hold', 'N trading days', 'hold for N days', "
+                "'N-day horizon', 'over N days'. If no specific number is mentioned, use 5.\n"
+                "For CONFIDENCE and RISK_LEVEL, infer from the tone and content of the report. Default to MEDIUM if unclear.",
             ),
             ("human", full_signal),
         ]
 
         response = self.quick_thinking_llm.invoke(messages).content
-        return self._parse_signal_response(response)
+        result = self._parse_signal_response(response)
+
+        # If LLM returned default hold_days (5) or failed to extract, try regex on original text
+        if result["decision"] != "SELL" and result["hold_days"] == 5:
+            regex_days = self._extract_hold_days_regex(full_signal)
+            if regex_days is not None:
+                result["hold_days"] = regex_days
+
+        return result
+
+    @staticmethod
+    def _extract_hold_days_regex(text: str) -> int | None:
+        """Extract hold period from text using regex patterns.
+
+        Looks for common patterns like '15-day hold', 'hold for 45 days',
+        '30 trading days', 'N-day horizon', etc.
+        """
+        patterns = [
+            # "15-day hold", "45-day horizon", "30-day period"
+            r'(\d+)[\s-]*(?:day|trading[\s-]*day)[\s-]*(?:hold|horizon|period|timeframe)',
+            # "hold for 15 days", "holding period of 45 days"
+            r'(?:hold|holding)[\s\w]*?(?:for|of|period\s+of)[\s]*(\d+)[\s]*(?:trading\s+)?days?',
+            # "setting 45 trading days"
+            r'setting\s+(\d+)\s+(?:trading\s+)?days',
+            # "over 15 days", "within 30 days"
+            r'(?:over|within|next)\s+(\d+)\s+(?:trading\s+)?days',
+            # "N trading days (~2 months)" pattern
+            r'(\d+)\s+trading\s+days?\s*\(',
+        ]
+
+        candidates = []
+        for pattern in patterns:
+            for match in re.finditer(pattern, text, re.IGNORECASE):
+                days = int(match.group(1))
+                if 1 <= days <= 90:
+                    candidates.append(days)
+
+        if not candidates:
+            return None
+
+        # If multiple matches, prefer the one that appears in the conclusion
+        # (last ~500 chars of text, which is typically the RATIONALE section)
+        conclusion = text[-500:]
+        for pattern in patterns:
+            for match in re.finditer(pattern, conclusion, re.IGNORECASE):
+                days = int(match.group(1))
+                if 1 <= days <= 90:
+                    return days
+
+        # Fall back to most common candidate
+        return max(set(candidates), key=candidates.count)
 
     def _parse_signal_response(self, response: str) -> dict:
-        """Parse the structured LLM response into decision and hold_days."""
+        """Parse the structured LLM response into decision, hold_days, confidence, risk."""
         decision = "HOLD"
         hold_days = None
+        confidence = "MEDIUM"
+        risk = "MEDIUM"
 
         for line in response.strip().split("\n"):
             line = line.strip()
@@ -63,6 +124,16 @@ class SignalProcessor:
                         hold_days = max(1, min(90, hold_days))
                     except (ValueError, TypeError):
                         hold_days = None
+            elif upper.startswith("CONFIDENCE:"):
+                raw = upper.split(":", 1)[1].strip()
+                raw = raw.replace("*", "").strip()
+                if raw in ("HIGH", "MEDIUM", "LOW"):
+                    confidence = raw
+            elif upper.startswith("RISK_LEVEL:") or upper.startswith("RISK:"):
+                raw = upper.split(":", 1)[1].strip()
+                raw = raw.replace("*", "").strip()
+                if raw in ("HIGH", "MEDIUM", "LOW"):
+                    risk = raw
 
         # Enforce: SELL never has hold_days; BUY/HOLD default to 5 if missing
         if decision == "SELL":
@@ -70,4 +141,4 @@ class SignalProcessor:
         elif hold_days is None:
             hold_days = 5  # Default hold period
 
-        return {"decision": decision, "hold_days": hold_days}
+        return {"decision": decision, "hold_days": hold_days, "confidence": confidence, "risk": risk}

tradingagents/graph/trading_graph.py

@@ -81,17 +81,18 @@ class TradingAgentsGraph:
             exist_ok=True,
         )
 
-        # Initialize LLMs
+        # Initialize LLMs with low temperature for deterministic financial analysis
+        llm_temp = self.config.get("llm_temperature", 0.2)
         if self.config["llm_provider"].lower() == "openai" or self.config["llm_provider"] == "ollama" or self.config["llm_provider"] == "openrouter":
-            self.deep_thinking_llm = ChatOpenAI(model=self.config["deep_think_llm"], base_url=self.config["backend_url"])
-            self.quick_thinking_llm = ChatOpenAI(model=self.config["quick_think_llm"], base_url=self.config["backend_url"])
+            self.deep_thinking_llm = ChatOpenAI(model=self.config["deep_think_llm"], base_url=self.config["backend_url"], temperature=llm_temp)
+            self.quick_thinking_llm = ChatOpenAI(model=self.config["quick_think_llm"], base_url=self.config["backend_url"], temperature=llm_temp)
         elif self.config["llm_provider"].lower() == "anthropic":
             # Use ClaudeMaxLLM to leverage Claude Max subscription via CLI
-            self.deep_thinking_llm = ClaudeMaxLLM(model=self.config["deep_think_llm"])
-            self.quick_thinking_llm = ClaudeMaxLLM(model=self.config["quick_think_llm"])
+            self.deep_thinking_llm = ClaudeMaxLLM(model=self.config["deep_think_llm"], temperature=llm_temp)
+            self.quick_thinking_llm = ClaudeMaxLLM(model=self.config["quick_think_llm"], temperature=llm_temp)
         elif self.config["llm_provider"].lower() == "google":
-            self.deep_thinking_llm = ChatGoogleGenerativeAI(model=self.config["deep_think_llm"])
-            self.quick_thinking_llm = ChatGoogleGenerativeAI(model=self.config["quick_think_llm"])
+            self.deep_thinking_llm = ChatGoogleGenerativeAI(model=self.config["deep_think_llm"], temperature=llm_temp)
+            self.quick_thinking_llm = ChatGoogleGenerativeAI(model=self.config["quick_think_llm"], temperature=llm_temp)
         else:
             raise ValueError(f"Unsupported LLM provider: {self.config['llm_provider']}")
         
@@ -116,7 +117,10 @@ class TradingAgentsGraph:
         self.tool_nodes = self._create_tool_nodes()
 
         # Initialize components
-        self.conditional_logic = ConditionalLogic()
+        self.conditional_logic = ConditionalLogic(
+            max_debate_rounds=self.config.get("max_debate_rounds", 1),
+            max_risk_discuss_rounds=self.config.get("max_risk_discuss_rounds", 1),
+        )
         self.graph_setup = GraphSetup(
             self.quick_thinking_llm,
             self.deep_thinking_llm,
@@ -250,16 +254,18 @@ class TradingAgentsGraph:
         self._save_to_frontend_db(trade_date, final_state)
         add_log("info", "system", f"Database save completed in {_time.time() - t0:.1f}s")
 
-        # Extract and log the final decision + hold_days
+        # Extract and log the final decision + hold_days + confidence + risk
         signal_result = self.process_signal(final_state["final_trade_decision"])
         final_decision = signal_result["decision"]
         hold_days = signal_result.get("hold_days")
+        confidence = signal_result.get("confidence", "MEDIUM")
+        risk = signal_result.get("risk", "MEDIUM")
         total_elapsed = _time.time() - pipeline_start
         hold_info = f", hold {hold_days}d" if hold_days else ""
-        add_log("success", "system", f"✅ Analysis complete for {company_name}: {final_decision}{hold_info} (total: {total_elapsed:.0f}s)")
+        add_log("success", "system", f"✅ Analysis complete for {company_name}: {final_decision}{hold_info}, confidence={confidence}, risk={risk} (total: {total_elapsed:.0f}s)")
 
-        # Return decision, hold_days, and processed signal
-        return final_state, final_decision, hold_days
+        # Return decision, hold_days, confidence, risk
+        return final_state, final_decision, hold_days, confidence, risk
 
     def _log_state(self, trade_date, final_state):
         """Log the final state to a JSON file."""