Refactor trend detection logic and update date filtering in main.py. Removed width parameter from detect_trends method and enhanced peak detection to include valleys and peaks between consecutive points.

main.py

@@ -5,13 +5,17 @@ from cycle_detector import CycleDetector
 
 # Load data from CSV file instead of database
 data = pd.read_csv('data/btcusd_1-day_data.csv')
+
 # Convert datetime column to datetime type
-one_month_ago = pd.to_datetime('2025-04-05')
+start_date = pd.to_datetime('2025-04-01')
-daily_data = data[pd.to_datetime(data['datetime']) >= one_month_ago]
+stop_date = pd.to_datetime('2025-05-06')
+
+daily_data = data[(pd.to_datetime(data['datetime']) >= start_date) & 
+                  (pd.to_datetime(data['datetime']) < stop_date)]
 print(f"Number of data points: {len(daily_data)}")
 
 trend_detector = TrendDetectorSimple(daily_data, verbose=True)
-trends = trend_detector.detect_trends(width=1)
+trends = trend_detector.detect_trends()
 trend_detector.plot_trends(trends)
 
 #trend_detector = TrendDetectorMACD(daily_data, True)

trend_detector_simple.py

@@ -33,7 +33,7 @@ class TrendDetectorSimple:
         
         self.logger.info(f"Initialized TrendDetectorSimple with {len(self.data)} data points")
 
-    def detect_trends(self, width):
+    def detect_trends(self):
         """
         Detect trends by identifying local minima and maxima in the price data
         using scipy.signal.find_peaks.
@@ -45,13 +45,13 @@ class TrendDetectorSimple:
         Returns:
         - DataFrame with columns for timestamps, prices, and trend indicators
         """
-        self.logger.info(f"Detecting trends with width {width}")
+        self.logger.info(f"Detecting trends")
         
         df = self.data.copy()
         close_prices = df['close'].values
         
-        max_peaks, _ = find_peaks(close_prices, width=width)
+        max_peaks, _ = find_peaks(close_prices)
-        min_peaks, _ = find_peaks(-close_prices, width=width)
+        min_peaks, _ = find_peaks(-close_prices)
         
         self.logger.info(f"Found {len(min_peaks)} local minima and {len(max_peaks)} local maxima")
         
@@ -63,7 +63,38 @@ class TrendDetectorSimple:
         for peak in min_peaks:
             df.at[peak, 'is_min'] = True
         
-
+        # Find consecutive peaks and add the valley in between
+        if len(max_peaks) >= 2:
+            for i in range(len(max_peaks) - 1):
+                start_idx = max_peaks[i]
+                end_idx = max_peaks[i + 1]
+                
+                # Find the minimum value between these two peaks
+                segment = close_prices[start_idx:end_idx + 1]
+                valley_idx = start_idx + np.argmin(segment)
+                
+                # Add to min_peaks if not already there
+                if valley_idx not in min_peaks:
+                    min_peaks = np.append(min_peaks, valley_idx)
+                    df.at[valley_idx, 'is_min'] = True
+                    self.logger.info(f"Added valley at index {valley_idx} between peaks {start_idx} and {end_idx}")
+        
+        # Find consecutive valleys and add the peak in between
+        if len(min_peaks) >= 2:
+            min_peaks = np.sort(min_peaks)  # Sort after potential additions
+            for i in range(len(min_peaks) - 1):
+                start_idx = min_peaks[i]
+                end_idx = min_peaks[i + 1]
+                
+                # Find the maximum value between these two valleys
+                segment = close_prices[start_idx:end_idx + 1]
+                peak_idx = start_idx + np.argmax(segment)
+                
+                # Add to max_peaks if not already there
+                if peak_idx not in max_peaks:
+                    max_peaks = np.append(max_peaks, peak_idx)
+                    df.at[peak_idx, 'is_max'] = True
+                    self.logger.info(f"Added peak at index {peak_idx} between valleys {start_idx} and {end_idx}")
         
         result = df[['datetime', 'close', 'is_min', 'is_max']]