diff --git a/src/main.rs b/src/main.rs
index ec520a5..504e223 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -1,8 +1,8 @@
-use chrono::{NaiveDate, Datelike};
+use chrono::{NaiveDate, Datelike, Local};
 use fantoccini::{ClientBuilder, Locator};
 use serde::{Deserialize, Serialize};
 use serde_json::{Map, Value};
-use std::{path::PathBuf, process::Command};
+use std::{collections::{HashMap, HashSet}, path::PathBuf, process::Command};
 use tokio::{
     fs,
     signal,
@@ -22,6 +22,17 @@ struct EconomicEvent {
     description: String,
 }
 
+#[derive(Debug, Serialize, Deserialize, Clone)]
+struct EventChange {
+    date: String,
+    event: String,
+    country: String,
+    field_changed: String,
+    old_value: String,
+    new_value: String,
+    detected_at: String,
+}
+
 #[derive(Debug)]
 struct ChunkInfo {
     start_date: String,
@@ -254,11 +265,48 @@ fn calculate_next_start_date(events: &[EconomicEvent]) -> anyhow::Result<String>
     Ok(next.format("%Y-%m-%d").to_string())
 }
 
+fn event_key(event: &EconomicEvent) -> String {
+    format!("{}|{}|{}", event.date, event.time, event.event)
+}
+
+/// Compare two events and detect changes in future data
+fn detect_changes(old: &EconomicEvent, new: &EconomicEvent, now: &str) -> Vec<EventChange> {
+    let mut changes = Vec::new();
+    let timestamp = Local::now().format("%Y-%m-%d %H:%M:%S").to_string();
+    
+    // Only track changes for future events
+    if new.date.as_str() <= now {
+        return changes;
+    }
+
+    let fields = [
+        ("actual", &old.actual, &new.actual),
+        ("forecast", &old.forecast, &new.forecast),
+        ("previous", &old.previous, &new.previous),
+        ("description", &old.description, &new.description),
+    ];
+
+    for (field_name, old_val, new_val) in fields {
+        if old_val != new_val {
+            changes.push(EventChange {
+                date: new.date.clone(),
+                event: new.event.clone(),
+                country: new.country.clone(),
+                field_changed: field_name.to_string(),
+                old_value: old_val.to_string(),
+                new_value: new_val.to_string(),
+                detected_at: timestamp.clone(),
+            });
+        }
+    }
+
+    changes
+}
+
 /// Scan the economic_events directory for existing chunks
 async fn scan_existing_chunks() -> anyhow::Result<Vec<ChunkInfo>> {
     let events_dir = PathBuf::from("economic_events");
     
-    // Create directory if it doesn't exist
     if !events_dir.exists() {
         fs::create_dir_all(&events_dir).await?;
         println!("📁 Created economic_events directory");
@@ -272,11 +320,9 @@ async fn scan_existing_chunks() -> anyhow::Result<Vec<ChunkInfo>> {
         let path = entry.path();
         if path.extension().and_then(|s| s.to_str()) == Some("json") {
             if let Some(filename) = path.file_stem().and_then(|s| s.to_str()) {
-                // Parse filename: chunk_{startdate}_{enddate}.json
                 if let Some(dates) = filename.strip_prefix("chunk_") {
                     let parts: Vec<&str> = dates.split('_').collect();
                     if parts.len() == 2 {
-                        // Load and count events
                         if let Ok(content) = fs::read_to_string(&path).await {
                             if let Ok(events) = serde_json::from_str::<Vec<EconomicEvent>>(&content) {
                                 chunks.push(ChunkInfo {
@@ -302,7 +348,7 @@ async fn scan_existing_chunks() -> anyhow::Result<Vec<ChunkInfo>> {
                 chunk.start_date, chunk.end_date, chunk.event_count);
         }
     } else {
-        println!("📭 No existing chunks found");
+        println!("🔭 No existing chunks found");
     }
 
     Ok(chunks)
@@ -310,7 +356,7 @@ async fn scan_existing_chunks() -> anyhow::Result<Vec<ChunkInfo>> {
 
 /// Calculate target end date: first day of month, 3 months from now
 fn calculate_target_end_date() -> String {
-    let now = chrono::Local::now().naive_local().date();
+    let now = Local::now().naive_local().date();
     let three_months_ahead = if now.month() + 3 > 12 {
         NaiveDate::from_ymd_opt(now.year() + 1, (now.month() + 3) % 12, 1)
     } else {
@@ -320,82 +366,162 @@ fn calculate_target_end_date() -> String {
     three_months_ahead.format("%Y-%m-%d").to_string()
 }
 
-/// Determine what date range needs to be scraped based on existing data
-fn determine_scrape_range(chunks: &[ChunkInfo], target_end: &str) -> Option<(String, String)> {
-    let now = chrono::Local::now().naive_local().date().format("%Y-%m-%d").to_string();
-    
-    if chunks.is_empty() {
-        // No data exists, start from beginning
-        println!("📭 No existing data - scraping from 2007-02-13 to {}", target_end);
-        return Some(("2007-02-13".to_string(), target_end.to_string()));
-    }
-    
-    // Find the latest date in existing chunks
-    let latest_chunk_date = chunks.iter()
-        .map(|c| &c.end_date)
-        .max()
-        .cloned()
-        .unwrap_or_else(|| "2007-02-13".to_string());
-    
-    println!("📊 Latest existing data: {}", latest_chunk_date);
-    
-    if latest_chunk_date >= now {
-        // Data is ahead of current date - update from now to target
-        println!("🔄 Data exists beyond today - updating from {} to {}", now, target_end);
-        Some((now, target_end.to_string()))
-    } else {
-        // Data is behind - continue from where it left off
-        let next_start = parse_date(&latest_chunk_date)
-            .and_then(|d| d.succ_opt())
-            .map(|d| d.format("%Y-%m-%d").to_string())
-            .unwrap_or_else(|| latest_chunk_date.clone());
-            
-        println!("➡️ Continuing from {} to {}", next_start, target_end);
-        Some((next_start, target_end.to_string()))
-    }
-}
-
-/// Save a chunk to disk
-async fn save_chunk(events: &[EconomicEvent], start: &str, end: &str) -> anyhow::Result<()> {
-    let events_dir = PathBuf::from("economic_events");
-    fs::create_dir_all(&events_dir).await?;
-
-    let filename = format!("chunk_{}_{}.json", start, end);
-    let filepath = events_dir.join(&filename);
-
-    let json = serde_json::to_string_pretty(events)?;
-    fs::write(&filepath, json).await?;
-    
-    println!("💾 Saved chunk: {} ({} events)", filename, events.len());
-    Ok(())
-}
-
-/// Load all events from existing chunks
-async fn load_all_events(chunks: &[ChunkInfo]) -> anyhow::Result<Vec<EconomicEvent>> {
-    let mut all_events = Vec::new();
+/// Load all events from existing chunks into a HashMap
+async fn load_existing_events(chunks: &[ChunkInfo]) -> anyhow::Result<HashMap<String, EconomicEvent>> {
+    let mut event_map = HashMap::new();
 
     for chunk in chunks {
         if let Ok(content) = fs::read_to_string(&chunk.path).await {
             if let Ok(events) = serde_json::from_str::<Vec<EconomicEvent>>(&content) {
-                all_events.extend(events);
+                for event in events {
+                    event_map.insert(event_key(&event), event);
+                }
             }
         }
     }
 
-    println!("📥 Loaded {} events from existing chunks", all_events.len());
-    Ok(all_events)
+    println!("📥 Loaded {} events from existing chunks", event_map.len());
+    Ok(event_map)
 }
 
-/// Scrape events for a specific date range and save chunks immediately
-async fn scrape_date_range(
+/// Save or append changes to monthly change files
+async fn save_changes(changes: &[EventChange]) -> anyhow::Result<()> {
+    if changes.is_empty() {
+        return Ok(());
+    }
+
+    let events_dir = PathBuf::from("economic_events");
+    fs::create_dir_all(&events_dir).await?;
+
+    // Group changes by month
+    let mut changes_by_month: HashMap<String, Vec<EventChange>> = HashMap::new();
+    
+    for change in changes {
+        if let Some(date) = parse_date(&change.date) {
+            let month_key = format!("{:02}_{}", date.month(), date.year());
+            changes_by_month.entry(month_key).or_default().push(change.clone());
+        }
+    }
+
+    // Save each month's changes
+    for (month_key, month_changes) in changes_by_month {
+        let filename = format!("event_changes_{}.json", month_key);
+        let filepath = events_dir.join(&filename);
+
+        // Load existing changes if file exists
+        let mut all_changes = if filepath.exists() {
+            let content = fs::read_to_string(&filepath).await?;
+            serde_json::from_str::<Vec<EventChange>>(&content).unwrap_or_default()
+        } else {
+            Vec::new()
+        };
+
+        // Append new changes
+        all_changes.extend(month_changes);
+
+        // Save combined changes
+        let json = serde_json::to_string_pretty(&all_changes)?;
+        fs::write(&filepath, json).await?;
+        
+        println!("🔄 Updated changes file: {} ({} total changes)", filename, all_changes.len());
+    }
+
+    Ok(())
+}
+
+/// Reorganize events into optimal chunks and save them
+async fn save_optimized_chunks(events: HashMap<String, EconomicEvent>) -> anyhow::Result<()> {
+    if events.is_empty() {
+        return Ok(());
+    }
+
+    let events_dir = PathBuf::from("economic_events");
+    fs::create_dir_all(&events_dir).await?;
+
+    // Convert to sorted vector
+    let mut all_events: Vec<EconomicEvent> = events.into_values().collect();
+    all_events.sort_by(|a, b| a.date.cmp(&b.date));
+
+    // Group events by date ranges (chunks of ~100 days or similar)
+    let mut chunks: Vec<Vec<EconomicEvent>> = Vec::new();
+    let mut current_chunk = Vec::new();
+    let mut current_start_date: Option<NaiveDate> = None;
+
+    for event in all_events {
+        if let Some(event_date) = parse_date(&event.date) {
+            if let Some(start) = current_start_date {
+                // Start new chunk if we've gone 100+ days or have 500+ events
+                if (event_date - start).num_days() > 100 || current_chunk.len() >= 500 {
+                    chunks.push(current_chunk);
+                    current_chunk = Vec::new();
+                    current_start_date = Some(event_date);
+                }
+            } else {
+                current_start_date = Some(event_date);
+            }
+            current_chunk.push(event);
+        }
+    }
+
+    if !current_chunk.is_empty() {
+        chunks.push(current_chunk);
+    }
+
+    // Delete old chunk files
+    let mut entries = fs::read_dir(&events_dir).await?;
+    while let Some(entry) = entries.next_entry().await? {
+        let path = entry.path();
+        if let Some(filename) = path.file_stem().and_then(|s| s.to_str()) {
+            if filename.starts_with("chunk_") {
+                fs::remove_file(&path).await?;
+            }
+        }
+    }
+
+    // Save new optimized chunks
+    for chunk in chunks {
+        if chunk.is_empty() {
+            continue;
+        }
+
+        let start = chunk.iter()
+            .filter_map(|e| parse_date(&e.date))
+            .min()
+            .unwrap()
+            .format("%Y-%m-%d")
+            .to_string();
+
+        let end = chunk.iter()
+            .filter_map(|e| parse_date(&e.date))
+            .max()
+            .unwrap()
+            .format("%Y-%m-%d")
+            .to_string();
+
+        let filename = format!("chunk_{}_{}.json", start, end);
+        let filepath = events_dir.join(&filename);
+
+        let json = serde_json::to_string_pretty(&chunk)?;
+        fs::write(&filepath, json).await?;
+        
+        println!("💾 Saved optimized chunk: {} ({} events)", filename, chunk.len());
+    }
+
+    Ok(())
+}
+
+/// Scrape and update data with change tracking
+async fn scrape_and_update(
     client: &fantoccini::Client,
     start: &str,
     end: &str,
-) -> anyhow::Result<()> {
+    existing_events: &mut HashMap<String, EconomicEvent>,
+) -> anyhow::Result<Vec<EventChange>> {
     println!("\n🎯 Scraping range: {} to {}", start, end);
     
     let mut current_start = start.to_string();
-    let mut chunk_number = 0;
+    let mut all_changes = Vec::new();
+    let now = Local::now().naive_local().date().format("%Y-%m-%d").to_string();
 
     loop {
         set_date_range(client, &current_start, end).await?;
@@ -407,24 +533,24 @@ async fn scrape_date_range(
             break;
         }
 
-        chunk_number += 1;
         println!("   📦 Fetched {} events", events.len());
 
-        // Calculate actual date range of this chunk
-        let chunk_start = events.iter()
-            .filter_map(|e| parse_date(&e.date))
-            .min()
-            .map(|d| d.format("%Y-%m-%d").to_string())
-            .unwrap_or_else(|| current_start.clone());
+        // Process events: detect changes and update map
+        for new_event in events.clone() {
+            let key = event_key(&new_event);
             
-        let chunk_end = events.iter()
-            .filter_map(|e| parse_date(&e.date))
-            .max()
-            .map(|d| d.format("%Y-%m-%d").to_string())
-            .unwrap_or_else(|| end.to_string());
-
-        // Save chunk immediately
-        save_chunk(&events, &chunk_start, &chunk_end).await?;
+            if let Some(old_event) = existing_events.get(&key) {
+                // Event exists - check for changes
+                let changes = detect_changes(old_event, &new_event, &now);
+                if !changes.is_empty() {
+                    println!("   🔄 Detected {} change(s) in: {}", changes.len(), new_event.event);
+                    all_changes.extend(changes);
+                }
+            }
+            
+            // Insert or update event
+            existing_events.insert(key, new_event);
+        }
 
         let next = match calculate_next_start_date(&events) {
             Ok(n) => n,
@@ -443,31 +569,78 @@ async fn scrape_date_range(
         sleep(Duration::from_secs(2)).await;
     }
 
-    Ok(())
+    Ok(all_changes)
 }
 
-/// Main scraping logic with persistent storage
-async fn scrape_with_persistence(
-    client: &fantoccini::Client,
-) -> anyhow::Result<()> {
-    // Calculate target end date (3 months ahead, 1st of month)
+/// Main logic with intelligent update handling
+async fn run_intelligent_update(client: &fantoccini::Client) -> anyhow::Result<()> {
+    let now = Local::now().naive_local().date().format("%Y-%m-%d").to_string();
     let target_end = calculate_target_end_date();
+    
+    println!("📅 Today: {}", now);
     println!("🎯 Target end date: {}", target_end);
     
-    // Scan for existing chunks
-    let existing_chunks = scan_existing_chunks().await?;
+    // Load existing chunks
+    let chunks = scan_existing_chunks().await?;
+    let mut existing_events = load_existing_events(&chunks).await?;
     
-    // Determine what range needs to be scraped
-    let scrape_range = determine_scrape_range(&existing_chunks, &target_end);
-    
-    if let Some((start, end)) = scrape_range {
-        // Scrape the needed range (saves chunks automatically)
-        scrape_date_range(client, &start, &end).await?;
-        println!("\n✅ Scraping complete!");
-    } else {
-        println!("✅ All data is up to date!");
+    if existing_events.is_empty() {
+        // No existing data - full scrape from beginning
+        println!("\n🔭 No existing data - starting fresh scrape from 2007-02-13");
+        let changes = scrape_and_update(client, "2007-02-13", &target_end, &mut existing_events).await?;
+        save_changes(&changes).await?;
+        save_optimized_chunks(existing_events).await?;
+        return Ok(());
     }
-
+    
+    // Find date range of existing data
+    let dates: Vec<NaiveDate> = existing_events.values()
+        .filter_map(|e| parse_date(&e.date))
+        .collect();
+    
+    let min_date = dates.iter().min().unwrap().format("%Y-%m-%d").to_string();
+    let max_date = dates.iter().max().unwrap().format("%Y-%m-%d").to_string();
+    
+    println!("📊 Existing data range: {} to {}", min_date, max_date);
+    
+    // Determine update strategy
+    if max_date < now {
+        // Case 1: Data is in the past, need to update from max_date to target
+        let next_start = parse_date(&max_date)
+            .and_then(|d| d.succ_opt())
+            .map(|d| d.format("%Y-%m-%d").to_string())
+            .unwrap_or(max_date);
+            
+        println!("\n📈 Updating from end of existing data: {} to {}", next_start, target_end);
+        let changes = scrape_and_update(client, &next_start, &target_end, &mut existing_events).await?;
+        save_changes(&changes).await?;
+        save_optimized_chunks(existing_events).await?;
+    } else if max_date >= now {
+        // Case 2: Data extends to or beyond today, only refresh future data
+        println!("\n🔄 Data exists up to today - only refreshing future data: {} to {}", now, target_end);
+        
+        // Create a separate HashMap for only future events to avoid touching past data
+        let mut future_events: HashMap<String, EconomicEvent> = existing_events
+            .iter()
+            .filter(|(_, event)| event.date.as_str() >= now.as_str())
+            .map(|(k, v)| (k.clone(), v.clone()))
+            .collect();
+        
+        println!("📋 {} future events will be refreshed", future_events.len());
+        
+        // Scrape and update only future events
+        let changes = scrape_and_update(client, &now, &target_end, &mut future_events).await?;
+        save_changes(&changes).await?;
+        
+        // Merge future events back into existing events (past data untouched)
+        for (key, event) in future_events {
+            existing_events.insert(key, event);
+        }
+        
+        save_optimized_chunks(existing_events).await?;
+    }
+    
+    println!("\n✅ Update complete!");
     Ok(())
 }
 
@@ -528,28 +701,16 @@ async fn main() -> anyhow::Result<()> {
         sleep(Duration::from_secs(2)).await;
     }
 
-    // Run persistent scraping
-    scrape_with_persistence(&client).await?;
+    // Run intelligent update
+    run_intelligent_update(&client).await?;
 
-    // Load and display summary
+    // Display final summary
     let chunks = scan_existing_chunks().await?;
-    let all_events = load_all_events(&chunks).await?;
+    let final_events = load_existing_events(&chunks).await?;
     
     println!("\n📊 FINAL SUMMARY:");
     println!("   • Total chunks: {}", chunks.len());
-    println!("   • Total events: {}", all_events.len());
-    
-    if !chunks.is_empty() {
-        let dates: Vec<String> = all_events.iter()
-            .filter_map(|e| parse_date(&e.date))
-            .map(|d| d.format("%Y-%m-%d").to_string())
-            .collect();
-        if !dates.is_empty() {
-            let min = dates.iter().min().unwrap();
-            let max = dates.iter().max().unwrap();
-            println!("   • Date range: {} to {}", min, max);
-        }
-    }
+    println!("   • Total events: {}", final_events.len());
 
     client.close().await?;
     chromedriver.kill()?;