Upload trading_bot/data_fetcher.py with huggingface_hub

trading_bot/data_fetcher.py

@@ -0,0 +1,118 @@
+"""
+Data Fetcher - Market Data from Multiple Sources
+==================================================
+Fetches OHLCV data from exchanges via ccxt.
+"""
+import pandas as pd
+import numpy as np
+from datetime import datetime, timedelta
+from typing import Optional, List
+
+
+def fetch_ohlcv_ccxt(symbol: str = 'BTC/USDT', timeframe: str = '1h', 
+                      exchange_name: str = 'binance', limit: int = 1000,
+                      since: Optional[datetime] = None) -> pd.DataFrame:
+    """Fetch OHLCV data using ccxt."""
+    import ccxt
+    
+    exchange_class = getattr(ccxt, exchange_name)
+    exchange = exchange_class({'enableRateLimit': True})
+    
+    if since:
+        since_ms = int(since.timestamp() * 1000)
+    else:
+        since_ms = None
+    
+    all_ohlcv = []
+    fetched = 0
+    batch_limit = min(limit, 1000)
+    
+    while fetched < limit:
+        ohlcv = exchange.fetch_ohlcv(symbol, timeframe, since=since_ms, limit=batch_limit)
+        if not ohlcv:
+            break
+        all_ohlcv.extend(ohlcv)
+        fetched += len(ohlcv)
+        since_ms = ohlcv[-1][0] + 1
+        if len(ohlcv) < batch_limit:
+            break
+    
+    df = pd.DataFrame(all_ohlcv, columns=['timestamp', 'open', 'high', 'low', 'close', 'volume'])
+    df['timestamp'] = pd.to_datetime(df['timestamp'], unit='ms')
+    df.set_index('timestamp', inplace=True)
+    df = df[~df.index.duplicated(keep='last')]
+    df.sort_index(inplace=True)
+    
+    return df
+
+
+def generate_sample_data(days: int = 365, timeframe: str = '1h', 
+                         start_price: float = 40000, volatility: float = 0.02) -> pd.DataFrame:
+    """
+    Generate realistic synthetic OHLCV data for testing.
+    Uses geometric Brownian motion with regime switching.
+    """
+    np.random.seed(42)
+    
+    tf_minutes = {
+        '1m': 1, '5m': 5, '15m': 15, '30m': 30,
+        '1h': 60, '4h': 240, '1d': 1440
+    }
+    minutes = tf_minutes.get(timeframe, 60)
+    total_bars = int(days * 24 * 60 / minutes)
+    
+    # Regime switching: trending vs ranging
+    regimes = []
+    current_regime = 'trending'
+    regime_length = 0
+    for _ in range(total_bars):
+        regime_length += 1
+        if current_regime == 'trending' and regime_length > np.random.geometric(1/100):
+            current_regime = 'ranging'
+            regime_length = 0
+        elif current_regime == 'ranging' and regime_length > np.random.geometric(1/50):
+            current_regime = 'trending'
+            regime_length = 0
+        regimes.append(current_regime)
+    
+    # Generate prices
+    prices = [start_price]
+    for i in range(1, total_bars):
+        if regimes[i] == 'trending':
+            drift = np.random.choice([-1, 1]) * 0.0002
+            vol = volatility * 1.2
+        else:
+            drift = 0
+            vol = volatility * 0.6
+        
+        ret = drift + vol * np.random.randn() * np.sqrt(minutes / 1440)
+        prices.append(prices[-1] * (1 + ret))
+    
+    prices = np.array(prices)
+    
+    # Generate OHLCV
+    dates = pd.date_range(end=datetime.now(), periods=total_bars, freq=f'{minutes}min')
+    
+    opens = prices.copy()
+    closes = prices.copy()
+    noise = np.abs(np.random.randn(total_bars)) * volatility * prices * 0.5
+    highs = np.maximum(opens, closes) + noise
+    lows = np.minimum(opens, closes) - noise
+    lows = np.maximum(lows, prices * 0.9)  # prevent negative
+    
+    # Volume with patterns
+    base_volume = 1000
+    volume = base_volume * (1 + np.abs(np.random.randn(total_bars)) * 2)
+    # Higher volume on big moves
+    moves = np.abs(np.diff(np.log(prices), prepend=np.log(prices[0])))
+    volume *= (1 + moves * 50)
+    
+    df = pd.DataFrame({
+        'open': opens,
+        'high': highs,
+        'low': lows,
+        'close': closes,
+        'volume': volume
+    }, index=dates)
+    
+    return df