diff --git a/backend/app/api/dependencies.py b/backend/app/api/dependencies.py
index 66364436..2640043c 100644
--- a/backend/app/api/dependencies.py
+++ b/backend/app/api/dependencies.py
@@ -1,11 +1,14 @@
 """
 Shared dependencies for API routes
 """
+import logging
 from typing import Optional, Dict, Any
 from fastapi import Depends, HTTPException, Header
 from backend.app.services.trading_service import TradingService, trading_service
 from backend.app.services.auth_utils import verify_access_token
 
+logger = logging.getLogger(__name__)
+
 
 def get_trading_service() -> TradingService:
     """Dependency to get trading service instance"""
@@ -17,24 +20,29 @@ async def get_current_user_optional(
 ) -> Optional[Dict[str, Any]]:
     """
     Get current user from JWT token (optional - returns None if not authenticated)
-    
+
     Use this for endpoints that work both with and without authentication.
+    All exceptions are caught to prevent 500 errors on malformed tokens.
     """
     if not authorization or not authorization.startswith("Bearer "):
         return None
-    
-    token = authorization.replace("Bearer ", "")
-    payload = verify_access_token(token)
-    
-    if not payload:
+
+    try:
+        token = authorization.replace("Bearer ", "")
+        payload = verify_access_token(token)
+
+        if not payload:
+            return None
+
+        return {
+            "id": payload.get("sub"),
+            "email": payload.get("email"),
+            "name": payload.get("name"),
+            "avatar_url": payload.get("avatar_url"),
+        }
+    except Exception as e:
+        logger.warning(f"Token validation error in optional auth: {type(e).__name__}")
         return None
-    
-    return {
-        "id": payload.get("sub"),
-        "email": payload.get("email"),
-        "name": payload.get("name"),
-        "avatar_url": payload.get("avatar_url"),
-    }
 
 
 async def get_current_user_required(
diff --git a/backend/app/api/routes.py b/backend/app/api/routes.py
index d947964f..0b2a9ded 100644
--- a/backend/app/api/routes.py
+++ b/backend/app/api/routes.py
@@ -93,14 +93,21 @@ async def run_analysis(
     # Log with user info for tracking
     user_info = f"user={current_user['email']}" if current_user else "user=anonymous"
     logger.info(f"Creating analysis task for {request.ticker} on {request.analysis_date} ({user_info})")
-    
+
     # Create task in Redis with user info
-    task_id = task_manager.create_task({
-        "ticker": request.ticker,
-        "analysis_date": request.analysis_date,
-        "user_id": current_user["id"] if current_user else None,
-        "user_email": current_user["email"] if current_user else None,
-    })
+    try:
+        task_id = task_manager.create_task({
+            "ticker": request.ticker,
+            "analysis_date": request.analysis_date,
+            "user_id": current_user["id"] if current_user else None,
+            "user_email": current_user["email"] if current_user else None,
+        })
+    except Exception as e:
+        logger.error(f"Failed to create analysis task: {type(e).__name__}: {str(e)}", exc_info=True)
+        raise HTTPException(
+            status_code=503,
+            detail=f"Task creation failed ({type(e).__name__}). The task service may be temporarily unavailable. Please try again."
+        )
     
     # Start background analysis
     def run_background_analysis():
diff --git a/backend/app/api/user.py b/backend/app/api/user.py
index f14e9bf4..ab1a2a6e 100644
--- a/backend/app/api/user.py
+++ b/backend/app/api/user.py
@@ -7,7 +7,8 @@ from uuid import UUID
 from fastapi import APIRouter, HTTPException, Depends, Header
 from pydantic import BaseModel
 from sqlalchemy.ext.asyncio import AsyncSession
-from sqlalchemy import select, delete, func
+from sqlalchemy import select, delete, func, text
+from sqlalchemy.orm import defer
 from datetime import datetime
 
 from backend.app.db import get_db, User, UserSettings, Report
@@ -155,7 +156,8 @@ async def get_reports(
     market_type: Optional[str] = None,
     language: Optional[str] = None,
     limit: int = 100,
-    offset: int = 0
+    offset: int = 0,
+    include_result: bool = True,
 ):
     """Get user's reports with optional filtering and pagination
 
@@ -164,6 +166,8 @@ async def get_reports(
         language: Filter by language (en, zh-TW)
         limit: Maximum number of reports to return (default 100, max 500)
         offset: Number of reports to skip for pagination
+        include_result: Whether to include result JSONB field (default True for backward compat).
+                        Set to False for lightweight list queries.
     """
     # Cap limit at 500 to prevent memory issues
     limit = min(limit, 500)
@@ -171,31 +175,35 @@ async def get_reports(
     # Build query with filters
     query = select(Report).where(Report.user_id == user.id)
 
+    # Skip loading the large result JSONB column when not needed
+    if not include_result:
+        query = query.options(defer(Report.result))
+
     if market_type:
         query = query.where(Report.market_type == market_type)
 
     if language:
-        query = query.where(Report.language == language)
+        query = query.where(func.coalesce(Report.language, "zh-TW") == language)
 
     # Order by created_at DESC and apply pagination
     query = query.order_by(Report.created_at.desc()).offset(offset).limit(limit)
 
     result = await db.execute(query)
     reports = result.scalars().all()
-    
-    # Process reports to strip large payloads for the list view
+
+    # Process reports
     optimized_reports = []
     for r in reports:
-        # Create a copy of the result to avoid modifying SQLAlchemy objects directly
-        if r.result and isinstance(r.result, dict):
-            # Shallow copy the dictionary
+        if include_result and r.result and isinstance(r.result, dict):
+            # Shallow copy the dictionary and strip massive reports field
             optimized_result = dict(r.result)
-            # Remove the massive reports field if it exists
             if "reports" in optimized_result:
                 optimized_result["reports"] = None
+        elif include_result:
+            optimized_result = r.result or {}
         else:
-            optimized_result = r.result
-            
+            optimized_result = {}
+
         optimized_reports.append(
             ReportResponse(
                 id=str(r.id),
@@ -263,6 +271,40 @@ async def create_report(
     }
 
 
+@router.get("/reports/counts")
+async def get_report_counts(
+    user: User = Depends(get_current_user_required),
+    db: AsyncSession = Depends(get_db),
+    language: Optional[str] = None,
+):
+    """Get report counts by market_type without loading full rows (lightweight query)"""
+    base_query = (
+        select(
+            Report.market_type,
+            func.count(Report.id).label("count")
+        )
+        .where(Report.user_id == user.id)
+        .group_by(Report.market_type)
+    )
+
+    if language:
+        base_query = base_query.where(
+            func.coalesce(Report.language, "zh-TW") == language
+        )
+
+    result = await db.execute(base_query)
+    rows = result.all()
+
+    counts = {"us": 0, "twse": 0, "tpex": 0}
+    for market_type, count in rows:
+        if market_type in counts:
+            counts[market_type] = count
+
+    total = sum(counts.values())
+
+    return {"counts": counts, "total": total}
+
+
 @router.get("/reports/{report_id}")
 async def get_report(
     report_id: str,
@@ -299,39 +341,34 @@ async def cleanup_duplicate_reports(
     user: User = Depends(get_current_user_required),
     db: AsyncSession = Depends(get_db)
 ):
-    """Remove duplicate reports, keeping only the most recent one per (ticker, analysis_date, market_type, language)"""
-    # Fetch all user reports ordered newest first
-    result = await db.execute(
-        select(Report)
-        .where(Report.user_id == user.id)
-        .order_by(Report.created_at.desc())
-    )
-    all_reports = result.scalars().all()
-
-    seen: set = set()
-    ids_to_delete: list = []
-
-    for report in all_reports:
-        # Normalize language
-        lang = report.language or "zh-TW"
-        key = (report.ticker, report.analysis_date, report.market_type, lang)
-        if key in seen:
-            ids_to_delete.append(report.id)
-        else:
-            seen.add(key)
-
-    if ids_to_delete:
-        await db.execute(
-            delete(Report)
-            .where(Report.user_id == user.id)
-            .where(Report.id.in_(ids_to_delete))
+    """Remove duplicate reports using SQL-level deduplication.
+    Keeps only the most recent one per (ticker, analysis_date, market_type, language).
+    """
+    # Use SQL window function to find duplicates without loading all rows into memory
+    result = await db.execute(text("""
+        DELETE FROM reports
+        WHERE id IN (
+            SELECT id FROM (
+                SELECT id,
+                    ROW_NUMBER() OVER (
+                        PARTITION BY user_id, ticker, analysis_date, market_type,
+                                     COALESCE(language, 'zh-TW')
+                        ORDER BY created_at DESC
+                    ) as rn
+                FROM reports
+                WHERE user_id = :user_id
+            ) ranked
+            WHERE rn > 1
         )
-        await db.commit()
+    """), {"user_id": str(user.id)})
+
+    deleted_count = result.rowcount
+    await db.commit()
 
     return {
         "success": True,
-        "deleted": len(ids_to_delete),
-        "message": f"Cleaned up {len(ids_to_delete)} duplicate reports"
+        "deleted": deleted_count,
+        "message": f"Cleaned up {deleted_count} duplicate reports"
     }
 
 
diff --git a/backend/app/db/database.py b/backend/app/db/database.py
index 54ac475e..b92df752 100644
--- a/backend/app/db/database.py
+++ b/backend/app/db/database.py
@@ -77,6 +77,8 @@ async def init_db():
             await conn.execute(text("CREATE INDEX IF NOT EXISTS ix_reports_language ON reports (language);"))
             # Add composite index for user + market_type + language queries
             await conn.execute(text("CREATE INDEX IF NOT EXISTS ix_reports_user_market_lang ON reports (user_id, market_type, language);"))
+            # Covering index for counts query (GROUP BY market_type with language filter)
+            await conn.execute(text("CREATE INDEX IF NOT EXISTS ix_reports_user_market_lang_count ON reports (user_id, market_type, COALESCE(language, 'zh-TW'));"))
         except Exception as e:
             print(f"Skipping manual migration (might be SQLite or syntax not supported): {e}")
     
diff --git a/backend/app/main.py b/backend/app/main.py
index 20322a47..ca12fb4d 100644
--- a/backend/app/main.py
+++ b/backend/app/main.py
@@ -171,10 +171,10 @@ async def root():
 
 
 @app.exception_handler(Exception)
-async def global_exception_handler(request, exc):
-    """Global exception handler - masks sensitive data in errors"""
+async def global_exception_handler(request: Request, exc: Exception):
+    """Global exception handler - returns appropriate status codes and masks sensitive data"""
     error_msg = str(exc)
-    
+
     # Mask any API keys that might be in the error message
     import re
     patterns = ["sk-", "sk-ant-", "xai-", "AIza"]
@@ -184,14 +184,32 @@ async def global_exception_handler(request, exc):
             f'{pattern}**********',
             error_msg
         )
-    
-    logger.error(f"Unhandled exception: {error_msg}", exc_info=True)
+
+    # Determine appropriate status code based on exception type
+    error_type = type(exc).__name__
+    status_code = 500
+
+    if isinstance(exc, ValueError):
+        status_code = 400
+    elif isinstance(exc, (ConnectionError, ConnectionRefusedError, TimeoutError)):
+        status_code = 503
+    elif isinstance(exc, PermissionError):
+        status_code = 403
+    elif isinstance(exc, FileNotFoundError):
+        status_code = 404
+
+    logger.error(
+        f"Unhandled exception on {request.method} {request.url.path}: "
+        f"[{error_type}] {error_msg}",
+        exc_info=True
+    )
     return JSONResponse(
-        status_code=500,
+        status_code=status_code,
         content={
-            "error": "Internal server error",
+            "error": "Internal server error" if status_code == 500 else error_type,
             "detail": error_msg,
-            "type": type(exc).__name__,
+            "type": error_type,
+            "path": str(request.url.path),
         },
     )
 
diff --git a/frontend/lib/report-utils.ts b/frontend/lib/report-utils.ts
new file mode 100644
index 00000000..709dd648
--- /dev/null
+++ b/frontend/lib/report-utils.ts
@@ -0,0 +1,81 @@
+/**
+ * Shared utilities for report deduplication, language detection, and date parsing.
+ * Single source of truth — imported by history/page.tsx, analysis/page.tsx, results/page.tsx, etc.
+ */
+
+/**
+ * Normalize report language. All null/undefined values become "zh-TW"
+ * to match backend behavior (COALESCE(language, 'zh-TW')).
+ */
+export function normalizeLanguage(
+  language?: string | null
+): "en" | "zh-TW" {
+  if (language === "en") return "en";
+  return "zh-TW";
+}
+
+/**
+ * Detect report language from content (for backward compatibility with old reports
+ * that don't have a language field stored).
+ * Checks trader_investment_plan for Chinese/English keywords.
+ */
+export function detectReportLanguage(reports: any): "en" | "zh-TW" {
+  const traderPlan = reports?.trader_investment_plan;
+  if (!traderPlan || typeof traderPlan !== "string") {
+    // If no trader plan, check other reports for Chinese characters
+    const allText = JSON.stringify(reports || {});
+    const chineseRegex = /[\u4e00-\u9fa5]/;
+    return chineseRegex.test(allText) ? "zh-TW" : "en";
+  }
+
+  // Check for Chinese decision keywords
+  const chineseKeywords = ["買入", "賣出", "持有", "最終交易提案"];
+  for (const keyword of chineseKeywords) {
+    if (traderPlan.includes(keyword)) {
+      return "zh-TW";
+    }
+  }
+
+  // Check for English decision keywords
+  const englishKeywords = ["buy", "sell", "hold", "final trading proposal"];
+  const lowerPlan = traderPlan.toLowerCase();
+  for (const keyword of englishKeywords) {
+    if (lowerPlan.includes(keyword)) {
+      return "en";
+    }
+  }
+
+  // Fallback: check for Chinese characters in the content
+  const chineseRegex = /[\u4e00-\u9fa5]/;
+  return chineseRegex.test(traderPlan) ? "zh-TW" : "en";
+}
+
+/**
+ * Generate a unique signature for report deduplication.
+ * Uses stable key fields: ticker + date + market_type + language.
+ * Language is normalized to "zh-TW" when null/undefined to match backend behavior.
+ */
+export function getReportSignature(report: {
+  ticker: string;
+  analysis_date: string;
+  market_type?: string;
+  language?: string | null;
+}): string {
+  const lang = normalizeLanguage(report.language);
+  return `${report.ticker}_${report.analysis_date}_${report.market_type || "us"}_${lang}`;
+}
+
+/**
+ * Parse a date string from the backend as UTC.
+ * Backend stores created_at in UTC but may not always include timezone info.
+ * This ensures the date is correctly interpreted as UTC so the browser
+ * converts it to the user's local timezone for display.
+ */
+export function parseUTCDate(dateStr: string): Date {
+  // If the string already has timezone info (Z, +, or - offset), parse directly
+  if (dateStr.endsWith("Z") || /[+-]\d{2}:\d{2}$/.test(dateStr)) {
+    return new Date(dateStr);
+  }
+  // Otherwise, append 'Z' to treat as UTC
+  return new Date(dateStr + "Z");
+}
diff --git a/frontend/lib/reports-db.ts b/frontend/lib/reports-db.ts
index 0bde5760..b8309ec8 100644
--- a/frontend/lib/reports-db.ts
+++ b/frontend/lib/reports-db.ts
@@ -5,6 +5,7 @@
 
 import Dexie, { type Table } from "dexie";
 import type { AnalysisResponse } from "./types";
+import { normalizeLanguage } from "./report-utils";
 
 // Saved report interface
 export interface SavedReport {
@@ -16,6 +17,8 @@ export interface SavedReport {
   task_id?: string; // Original task ID
   result: AnalysisResponse; // Full analysis result
   language?: "en" | "zh-TW"; // Language of the report (for filtering)
+  cloud_id?: string; // Corresponding cloud report ID (for sync tracking)
+  pending_sync?: boolean; // Whether report is waiting to be synced to cloud
 }
 
 // Database class extending Dexie
@@ -32,6 +35,11 @@ class ReportsDatabase extends Dexie {
     this.version(2).stores({
       reports: "++id, ticker, market_type, analysis_date, saved_at, language",
     });
+    // Version 3: Added cloud_id and pending_sync for sync tracking
+    this.version(3).stores({
+      reports:
+        "++id, ticker, market_type, analysis_date, saved_at, language, cloud_id, pending_sync",
+    });
   }
 }
 
@@ -123,22 +131,32 @@ export async function getReportCountByMarketType(): Promise<{
 }
 
 /**
- * Check if a report with the same ticker and analysis_date already exists
+ * Check if a report with the same signature already exists.
+ * Supports optional market_type and language for precise matching.
  */
 export async function checkDuplicateReport(
   ticker: string,
   analysis_date: string,
+  market_type?: "us" | "twse" | "tpex",
+  language?: "en" | "zh-TW",
 ): Promise<SavedReport | undefined> {
+  const normalizedLang = normalizeLanguage(language);
   return await db.reports
     .where("ticker")
     .equals(ticker)
-    .and((report) => report.analysis_date === analysis_date)
+    .and((report) => {
+      if (report.analysis_date !== analysis_date) return false;
+      if (market_type && report.market_type !== market_type) return false;
+      if (normalizeLanguage(report.language) !== normalizedLang) return false;
+      return true;
+    })
     .first();
 }
 
 /**
  * Check if a report exists by ticker, date, market type, and language
- * Used for bidirectional sync to prevent duplicates
+ * Used for bidirectional sync to prevent duplicates.
+ * Language is normalized so null/undefined matches "zh-TW".
  */
 export async function findExistingReport(
   ticker: string,
@@ -146,6 +164,7 @@ export async function findExistingReport(
   market_type: "us" | "twse" | "tpex",
   language?: "en" | "zh-TW",
 ): Promise<SavedReport | undefined> {
+  const normalizedLang = normalizeLanguage(language);
   return await db.reports
     .where("ticker")
     .equals(ticker)
@@ -153,7 +172,7 @@ export async function findExistingReport(
       (report) =>
         report.analysis_date === analysis_date &&
         report.market_type === market_type &&
-        report.language === language
+        normalizeLanguage(report.language) === normalizedLang
     )
     .first();
 }