fix breaking news banner

app/components/news.py

@@ -353,10 +353,11 @@ def display_category_breakdown(stats: dict):
 def display_breaking_news_banner(df: pd.DataFrame):
     """Display breaking news banner at the top with TradingView styling and ML-based impact score."""
 
-    breaking = df[df['is_breaking'] == True] if not df.empty and 'is_breaking' in df.columns else pd.DataFrame()
-
-    if not breaking.empty:
-        latest = breaking.iloc[0]
+    # With ML-based scoring, we trust that the passed DataFrame already contains
+    # the highest-impact news, so no need to filter by is_breaking
+    # (The scorer already selected the most impactful news)
+    if not df.empty:
+        latest = df.iloc[0]
 
         # Escape HTML
         summary = html_module.escape(latest.get('summary', '').strip())

app/pages/05_Dashboard.py

@@ -311,6 +311,9 @@ with st.spinner("🔍 Fetching latest financial & tech news in parallel..."):
             'ai_tech': future_ai_tech
         }
 
+        # Track errors to display later (avoid cluttering top of page)
+        fetch_errors = []
+
         for source_name, future in futures.items():
             try:
                 result_df, error = future.result(timeout=90)  # 90 second timeout per source
@@ -318,25 +321,25 @@ with st.spinner("🔍 Fetching latest financial & tech news in parallel..."):
                 if source_name == 'twitter':
                     twitter_df = result_df
                     if error:
-                        st.warning(error)
+                        fetch_errors.append(error)
                 elif source_name == 'reddit':
                     reddit_df = result_df
                     if error:
-                        st.warning(error)
+                        fetch_errors.append(error)
                 elif source_name == 'rss':
                     rss_all_df = result_df
                     if error:
-                        st.warning(error)
-                elif source_name == 'ai_tech':
-                    ai_tech_df = result_df
-                    if error:
-                        st.warning(error)
+                        fetch_errors.append(error)
                     # Get main page news subset for RSS
                     if not rss_all_df.empty and 'from_web' in rss_all_df.columns:
                         rss_main_df = rss_all_df[rss_all_df['from_web'] == True].copy()
+                elif source_name == 'ai_tech':
+                    ai_tech_df = result_df
+                    if error:
+                        fetch_errors.append(error)
 
             except Exception as e:
-                st.warning(f"Error fetching {source_name} news: {e}")
+                fetch_errors.append(f"Error fetching {source_name} news: {e}")
 
 # Clear force refresh flag after fetching is complete
 if force_refresh:
@@ -379,8 +382,15 @@ if not all_news_df.empty:
         breaking_df = pd.DataFrame([breaking_news_items[0]])
         display_breaking_news_banner(breaking_df)
     else:
-        # If no high-impact news found, show informational message
-        st.info("📊 Monitoring financial markets - no high-impact breaking news at this time.")
+        # If no high-impact news found, show informational message with score
+        if breaking_news_items:
+            top_score = breaking_news_items[0]['breaking_score']
+            st.info(f"📊 Monitoring financial markets - highest impact score: {top_score:.1f}/100 (threshold: 40)")
+        else:
+            st.info("📊 Monitoring financial markets - no news items available for scoring")
+else:
+    # No news data available at all
+    st.info("📊 Loading financial news - breaking news banner will appear when data is available")
 
 st.markdown("---")
 
@@ -520,6 +530,12 @@ with col4:
         </style>
         """, unsafe_allow_html=True)
 
+# Display fetch errors in expander (less intrusive)
+if 'fetch_errors' in locals() and fetch_errors:
+    with st.expander("⚠️ Source Fetch Warnings", expanded=False):
+        for error in fetch_errors:
+            st.caption(f"• {error}")
+
 # Auto-refresh logic
 if auto_refresh:
     import time