Create app.py

app.py

@@ -0,0 +1,145 @@
+import gym
+import numpy as np
+import matplotlib.pyplot as plt
+import requests
+import pandas as pd
+from datetime import datetime, timedelta
+from stable_baselines3 import PPO
+from stable_baselines3.common.vec_env import DummyVecEnv
+from gym import spaces
+import time
+import firebase_admin
+from firebase_admin import credentials, db
+import os
+
+cred = credentials.Certificate("credentials.json")
+firebase_admin.initialize_app(cred, {"databaseURL": "https://socail-swap-default-rtdb.asia-southeast1.firebasedatabase.app/"})
+ref = db.reference()
+stopmodel = False
+buy_signals = []
+sell_signals = []
+
+class TradingEnv(gym.Env):
+    def __init__(self, data, window_size=50):
+        super(TradingEnv, self).__init__()
+        self.data = data
+        self.window_size = window_size
+        self.current_step = window_size
+        self.action_space = spaces.Discrete(3)
+        self.observation_space = spaces.Box(
+            low=0, high=1, shape=(window_size, 2), dtype=np.float32)
+        
+    def reset(self):
+        self.current_step = self.window_size
+        return self._get_observation()
+    
+    def _get_observation(self):
+        window_data = self.data[self.current_step-self.window_size:self.current_step]
+        obs = window_data[['Close', 'EMA']].values
+        obs = (obs - obs.min()) / (obs.max() - obs.min())
+        return obs
+    
+    def step(self, action):
+        reward = 0
+        done = False
+        self.current_step += 1
+
+        if self.current_step >= len(self.data):
+            done = True
+        else:
+            if action == 1:  
+                reward = self.data['Close'].iloc[self.current_step] - self.data['Close'].iloc[self.current_step - 1]
+            elif action == 2: 
+                reward = self.data['Close'].iloc[self.current_step - 1] - self.data['Close'].iloc[self.current_step]
+
+        return self._get_observation(), reward, done, {}
+    
+def fetch_data(symbol='ETHUSDT', interval='1h', start_date='2021-01-01'):
+    end_date = datetime.utcnow()
+    start_date = datetime.strptime(start_date, '%Y-%m-%d')
+    klines = []
+    
+    while start_date < end_date:
+        url = f'https://api.binance.com/api/v3/klines?symbol={symbol}&interval={interval}&startTime={int(start_date.timestamp() * 1000)}'
+        response = requests.get(url)
+        data = response.json()
+        if not data:
+            break
+        klines += data
+        start_date = datetime.utcfromtimestamp(data[-1][6] / 1000)
+    
+    df = pd.DataFrame(klines, columns=['timestamp', 'Open', 'High', 'Low', 'Close', 'Volume', 'Close_time', 'Quote_asset_volume', 'Number_of_trades', 'Taker_buy_base_asset_volume', 'Taker_buy_quote_asset_volume', 'Ignore'])
+    df['Close'] = df['Close'].astype(float)
+    df['timestamp'] = pd.to_datetime(df['timestamp'], unit='ms')
+    df.set_index('timestamp', inplace=True)
+    return df[['Close']]
+
+def calculate_ema(data, span=20):
+    data['EMA'] = data['Close'].ewm(span=span, adjust=False).mean()
+    return data
+
+def run_model():
+    while True:
+        try:
+            new_data = fetch_data(start_date=(datetime.utcnow() - timedelta(days=3)).strftime('%Y-%m-%d'))
+            new_data = calculate_ema(new_data)
+            
+            if len(new_data) < 50:
+                print("Not enough data to update the environment.")
+                time.sleep(3600)
+                continue
+            
+            env = DummyVecEnv([lambda: TradingEnv(new_data)])
+            model.set_env(env)
+            
+            obs = env.reset()
+            dates = new_data.index[50:]
+            prices = new_data['Close'][50:]
+            emas = new_data['EMA'][50:]
+            actions = []
+
+            for date, price, ema in zip(dates, prices, emas):
+                action, _ = model.predict(obs)
+                actions.append(action[0])
+                obs, _, done, _ = env.step(action)
+                if done:
+                    break
+
+            new_buy_signals = [(date, price, ema) for date, price, ema, action in zip(dates, prices, emas, actions) if action == 1 and date not in [signal[0] for signal in buy_signals]]
+            new_sell_signals = [(date, price, ema) for date, price, ema, action in zip(dates, prices, emas, actions) if action == 2 and date not in [signal[0] for signal in sell_signals]]
+
+            for signal in new_buy_signals:
+                if signal[0] not in [s[0] for s in buy_signals] and signal[0] not in [s[0] for s in sell_signals]:
+                    buy_signals.append(signal)
+                
+
+            for signal in new_sell_signals:
+                if signal[0] not in [s[0] for s in sell_signals] and signal[0] not in [s[0] for s in buy_signals]:
+                    sell_signals.append(signal)
+                    
+            buy_signals_data = [{'timestamp': signal[0].strftime('%Y-%m-%d %H:%M:%S'), 'type': 'b', 'price': round(signal[1], 2), 'ema': round(signal[2],2)} for signal in buy_signals]
+            sell_signals_data = [{'timestamp': signal[0].strftime('%Y-%m-%d %H:%M:%S'), 'type': 's', 'price': round(signal[1], 2), 'ema': round(signal[2],2)} for signal in sell_signals]
+
+            all_signals_data = buy_signals_data + sell_signals_data
+
+            ref.child('signals').child('data').set(all_signals_data)
+            time.sleep(3600) 
+        except Exception as e:
+            print(f"An error occurred: {e}")
+            break
+        
+if __name__ == "__main__":
+    data = fetch_data()
+    data = calculate_ema(data)
+    if len(data) < 50:
+        raise ValueError("Not enough data to fill the window size.")
+
+    env = DummyVecEnv([lambda: TradingEnv(data)])
+
+    if os.path.exists("./ppo_trading_agent.zip"):
+        model = PPO.load("ppo_trading_agent", env=env)
+    else:
+        model = PPO('MlpPolicy', env, verbose=1)
+        model.learn(total_timesteps=10000)
+    print("Ender function")
+    run_model()