import os import pandas as pd import numpy as np from datetime import datetime, timedelta from binance.client import Client from sklearn.model_selection import train_test_split from sklearn.ensemble import RandomForestClassifier from sklearn.metrics import classification_report import ta # Connect to Binance (Fill your own API keys if live) # client = Client(api_key, api_secret) client = Client() # File to store the historical data DATA_FILE = "btc_data.csv" symbol = "BTCUSDT" interval = Client.KLINE_INTERVAL_4HOUR # Load existing data or download fresh if os.path.exists(DATA_FILE): print("Loading existing data...") df = pd.read_csv(DATA_FILE, index_col=0, parse_dates=True) last_timestamp = df.index[-1] # Binance gives data in 15min intervals, so move forward start_time = last_timestamp + timedelta(minutes=15) start_str = start_time.strftime("%d %B %Y %H:%M:%S") print(f"Downloading new data from {start_str}...") new_klines = client.get_historical_klines(symbol, interval, start_str) if new_klines: new_df = pd.DataFrame(new_klines, columns=['timestamp', 'open', 'high', 'low', 'close', 'volume', 'close_time', 'quote_av', 'trades', 'tb_base_av', 'tb_quote_av', 'ignore']) new_df = new_df[['timestamp', 'open', 'high', 'low', 'close', 'volume']] new_df[['open', 'high', 'low', 'close', 'volume']] = new_df[['open', 'high', 'low', 'close', 'volume']].astype(float) new_df['timestamp'] = pd.to_datetime(new_df['timestamp'], unit='ms') new_df = new_df.set_index('timestamp') # Append and remove any duplicates (just in case) df = pd.concat([df, new_df]) df = df[~df.index.duplicated(keep='first')] df.to_csv(DATA_FILE) else: print("Downloading all data from scratch...") klinesT = client.get_historical_klines(symbol, interval, "01 December 2021") df = pd.DataFrame(klinesT, columns=['timestamp', 'open', 'high', 'low', 'close', 'volume', 'close_time', 'quote_av', 'trades', 'tb_base_av', 'tb_quote_av', 'ignore']) df = df[['timestamp', 'open', 'high', 'low', 'close', 'volume']] df[['open', 'high', 'low', 'close', 'volume']] = df[['open', 'high', 'low', 'close', 'volume']].astype(float) df['timestamp'] = pd.to_datetime(df['timestamp'], unit='ms') df = df.set_index('timestamp') df.to_csv(DATA_FILE) # Feature Engineering: Add technical indicators df['rsi'] = ta.momentum.RSIIndicator(df['close'], window=14).rsi() df['sma_fast'] = df['close'].rolling(window=5).mean() df['sma_slow'] = df['close'].rolling(window=20).mean() df['macd'] = ta.trend.MACD(df['close']).macd() df['ema'] = df['close'].ewm(span=10, adjust=False).mean() # Create target: 1 if next close > current close, else 0 df['target'] = np.where(df['close'].shift(-1) > df['close'], 1, 0) # Drop rows with NaN values df = df.dropna() # Features and Target features = ['rsi', 'sma_fast', 'sma_slow', 'macd', 'ema'] X = df[features] y = df['target'] # Train/Test split X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, shuffle=False) # Train Random Forest model = RandomForestClassifier(n_estimators=100, random_state=42) model.fit(X_train, y_train) # Evaluate y_pred = model.predict(X_test) print(classification_report(y_test, y_pred)) # Predict next movement latest_features = X.iloc[-1].values.reshape(1, -1) predicted_direction = model.predict(latest_features) print(f"Predicted next movement: {'UP' if predicted_direction[0] == 1 else 'DOWN'}")