ranking_client.py

from polygon import RESTClient
from config import POLYGON_API_KEY, FINANCIAL_PREP_API_KEY, MONGO_DB_USER, MONGO_DB_PASS, API_KEY, API_SECRET, BASE_URL, mongo_url
import threading
from concurrent.futures import ThreadPoolExecutor
from urllib.request import urlopen
from zoneinfo import ZoneInfo
from pymongo import MongoClient
import time
from datetime import datetime, timedelta
import alpaca
from alpaca.trading.client import TradingClient
from alpaca.data.timeframe import TimeFrame, TimeFrameUnit
from alpaca.data.historical.stock import StockHistoricalDataClient
from alpaca.trading.stream import TradingStream
from alpaca.data.live.stock import StockDataStream
from alpaca.data.requests import (
    StockBarsRequest,
    StockTradesRequest,
    StockQuotesRequest
)
from alpaca.trading.requests import (
    GetAssetsRequest, 
    MarketOrderRequest, 
    LimitOrderRequest, 
    StopOrderRequest, 
    StopLimitOrderRequest, 
    TakeProfitRequest, 
    StopLossRequest, 
    TrailingStopOrderRequest, 
    GetOrdersRequest, 
    ClosePositionRequest
)
from alpaca.trading.enums import ( 
    AssetStatus, 
    AssetExchange, 
    OrderSide, 
    OrderType, 
    TimeInForce, 
    OrderClass, 
    QueryOrderStatus
)
from alpaca.common.exceptions import APIError
from strategies.talib_indicators import *
import math
import yfinance as yf
import logging
from collections import Counter
from trading_client import market_status
from helper_files.client_helper import strategies, get_latest_price, get_ndaq_tickers, dynamic_period_selector
import time
from datetime import datetime 
import heapq 
import certifi
ca = certifi.where()
logging.basicConfig(
    level=logging.INFO,
    format='%(asctime)s - %(levelname)s - %(message)s',
    datefmt='%Y-%m-%d %H:%M:%S',
    handlers=[
        logging.FileHandler('rank_system.log'),  # Log messages to a file
        logging.StreamHandler()             # Log messages to the console
    ]
)
from control import rank_mode, time_delta_mode, time_delta_increment, time_delta_multiplicative,time_delta_balanced, rank_liquidity_limit, rank_asset_limit, profit_price_change_ratio_d1, profit_profit_time_d1, profit_price_change_ratio_d2, profit_profit_time_d2, profit_profit_time_else, loss_price_change_ratio_d1, loss_price_change_ratio_d2, loss_profit_time_d1, loss_profit_time_d2, loss_profit_time_else
from control import period_start, period_end, train_tickers
import json

def process_ticker(ticker, mongo_client):
   try:
      
      current_price = None
      
      while current_price is None:
         try:
            current_price = get_latest_price(ticker)
         except Exception as fetch_error:
            logging.warning(f"Error fetching price for {ticker}. Retrying... {fetch_error}")
            time.sleep(10)
      
      indicator_tb = mongo_client.IndicatorsDatabase
      indicator_collection = indicator_tb.Indicators
      for strategy in strategies:
         historical_data = None
         while historical_data is None:
            try:
               period = indicator_collection.find_one({'indicator': strategy.__name__})
               historical_data = get_data(ticker, mongo_client, period['ideal_period'])
            except Exception as fetch_error:
               logging.warning(f"Error fetching historical data for {ticker}. Retrying... {fetch_error}")
               time.sleep(60)
         db = mongo_client.trading_simulator  
         holdings_collection = db.algorithm_holdings
         print(f"Processing {strategy.__name__} for {ticker}")
         strategy_doc = holdings_collection.find_one({"strategy": strategy.__name__})
         if not strategy_doc:
            logging.warning(f"Strategy {strategy.__name__} not found in database. Skipping.")
            continue

         account_cash = strategy_doc["amount_cash"]
         total_portfolio_value = strategy_doc["portfolio_value"]

         
         portfolio_qty = strategy_doc["holdings"].get(ticker, {}).get("quantity", 0)

         simulate_trade(ticker, strategy, historical_data, current_price,
                        account_cash, portfolio_qty, total_portfolio_value, mongo_client)
         
      print(f"{ticker} processing completed.")
   except Exception as e:
      logging.error(f"Error in thread for {ticker}: {e}")

def simulate_trade(ticker, strategy, historical_data, current_price, account_cash, portfolio_qty, total_portfolio_value, mongo_client):
   """
   Simulates a trade based on the given strategy and updates MongoDB.
   """
    
   # Simulate trading action from strategy
   print(f"Simulating trade for {ticker} with strategy {strategy.__name__} and quantity of {portfolio_qty}")
   action, quantity = simulate_strategy(strategy, ticker, current_price, historical_data, account_cash, portfolio_qty, total_portfolio_value)
   
   # MongoDB setup
   
   db = mongo_client.trading_simulator
   holdings_collection = db.algorithm_holdings
   points_collection = db.points_tally
   
   # Find the strategy document in MongoDB
   strategy_doc = holdings_collection.find_one({"strategy": strategy.__name__})
   holdings_doc = strategy_doc.get("holdings", {})
   time_delta = db.time_delta.find_one({})['time_delta']
   
   
   # Update holdings and cash based on trade action
   if action in ["buy"] and strategy_doc["amount_cash"] - quantity * current_price > rank_liquidity_limit and quantity > 0 and ((portfolio_qty + quantity) * current_price) / total_portfolio_value < rank_asset_limit:
      logging.info(f"Action: {action} | Ticker: {ticker} | Quantity: {quantity} | Price: {current_price}")
      # Calculate average price if already holding some shares of the ticker
      if ticker in holdings_doc:
         current_qty = holdings_doc[ticker]["quantity"]
         new_qty = current_qty + quantity
         average_price = (holdings_doc[ticker]["price"] * current_qty + current_price * quantity) / new_qty
      else:
         new_qty = quantity
         average_price = current_price

      # Update the holdings document for the ticker. 
      holdings_doc[ticker] = {
            "quantity": new_qty,
            "price": average_price
      }

      # Deduct the cash used for buying and increment total trades
      holdings_collection.update_one(
         {"strategy": strategy.__name__},
         {
            "$set": {
                  "holdings": holdings_doc,
                  "amount_cash": strategy_doc["amount_cash"] - quantity * current_price,
                  "last_updated": datetime.now()
            },
            "$inc": {"total_trades": 1}
         },
         upsert=True
      )
      

   elif action in ["sell"] and str(ticker) in holdings_doc and holdings_doc[str(ticker)]["quantity"] > 0:
      
      logging.info(f"Action: {action} | Ticker: {ticker} | Quantity: {quantity} | Price: {current_price}")
      current_qty = holdings_doc[ticker]["quantity"]
        
      # Ensure we do not sell more than we have
      sell_qty = min(quantity, current_qty)
      holdings_doc[ticker]["quantity"] = current_qty - sell_qty
      
      price_change_ratio = current_price / holdings_doc[ticker]["price"] if ticker in holdings_doc else 1
      
      

      if current_price > holdings_doc[ticker]["price"]:
         #increment successful trades
         holdings_collection.update_one(
            {"strategy": strategy.__name__},
            {"$inc": {"successful_trades": 1}},
            upsert=True
         )
         
         # Calculate points to add if the current price is higher than the purchase price
         if price_change_ratio < profit_price_change_ratio_d1:
            points = time_delta * profit_profit_time_d1
         elif price_change_ratio < profit_price_change_ratio_d2:
            points = time_delta * profit_profit_time_d2
         else:
            points = time_delta * profit_profit_time_else
         
      else:
         # Calculate points to deduct if the current price is lower than the purchase price
         if holdings_doc[ticker]["price"] == current_price:
            holdings_collection.update_one(
               {"strategy": strategy.__name__},
               {"$inc": {"neutral_trades": 1}}
            )
            
         else:   
            
            holdings_collection.update_one(
               {"strategy": strategy.__name__},
               {"$inc": {"failed_trades": 1}},
               upsert=True
            )
         
         if price_change_ratio > loss_price_change_ratio_d1:
            points = -time_delta * loss_profit_time_d1
         elif price_change_ratio > loss_price_change_ratio_d2:
            points = -time_delta * loss_profit_time_d2
         else:
            points = -time_delta * loss_profit_time_else
         
      # Update the points tally
      points_collection.update_one(
         {"strategy": strategy.__name__},
         {
            "$set" : {
               "last_updated": datetime.now()
            },
            "$inc": {"total_points": points}
         },
         upsert=True
      )
      if holdings_doc[ticker]["quantity"] == 0:      
         del holdings_doc[ticker]
      # Update cash after selling
      holdings_collection.update_one(
         {"strategy": strategy.__name__},
         {
            "$set": {
               "holdings": holdings_doc,
               "amount_cash": strategy_doc["amount_cash"] + sell_qty * current_price,
               "last_updated": datetime.now()
            },
            "$inc": {"total_trades": 1}
         },
         upsert=True
      )

        
      # Remove the ticker if quantity reaches zero
      if holdings_doc[ticker]["quantity"] == 0:      
         del holdings_doc[ticker]
        
   else:
      logging.info(f"Action: {action} | Ticker: {ticker} | Quantity: {quantity} | Price: {current_price}")
   print(f"Action: {action} | Ticker: {ticker} | Quantity: {quantity} | Price: {current_price}")
   # Close the MongoDB connection

def update_portfolio_values(client):
   """
   still need to implement.
   we go through each strategy and update portfolio value buy cash + summation(holding * current price)
   """
   
   db = client.trading_simulator  
   holdings_collection = db.algorithm_holdings
   # Update portfolio values
   for strategy_doc in holdings_collection.find({}):
      # Calculate the portfolio value for the strategy
      portfolio_value = strategy_doc["amount_cash"]
      
      for ticker, holding in strategy_doc["holdings"].items():
          # The current price can be gotten through a cache system maybe
          # if polygon api is getting clogged - but that hasn't happened yet
          # Also implement in C++ or C instead of python
          # Get the current price of the ticker from the Polygon API
          # Use a cache system to store the latest prices
          # If the cache is empty, fetch the latest price from the Polygon API
          # Cache should be updated every 60 seconds 

          current_price = None
          while current_price is None:
            try:
               # get latest price shouldn't cache - we should also do a delay
               current_price = get_latest_price(ticker)
            except:
               print(f"Error fetching price for {ticker}. Retrying...")
               time.sleep(120)
               # Will sleep 120 seconds before retrying to get latest price
          print(f"Current price of {ticker}: {current_price}")
          # Calculate the value of the holding
          holding_value = holding["quantity"] * current_price
          # Add the holding value to the portfolio value
          portfolio_value += holding_value
          
      # Update the portfolio value in the strategy document
      holdings_collection.update_one({"strategy": strategy_doc["strategy"]}, {"$set": {"portfolio_value": portfolio_value}}, upsert=True)

   # Update MongoDB with the modified strategy documents
   

def update_ranks(client):
   """"
   based on portfolio values, rank the strategies to use for actual trading_simulator
   """
   
   db = client.trading_simulator
   points_collection = db.points_tally
   rank_collection = db.rank
   algo_holdings = db.algorithm_holdings
   """
   delete all documents in rank collection first
   """
   rank_collection.delete_many({})
   """
   Reason why delete rank is so that rank is intially null and
   then we can populate it in the order we wish
   now update rank based on successful_trades - failed
   """
   q = []
   for strategy_doc in algo_holdings.find({}):
      """
      based on (points_tally (less points pops first), failed-successful(more negtive pops first), portfolio value (less value pops first), and then strategy_name), we add to heapq.
      """
      strategy_name = strategy_doc["strategy"]
      if strategy_name == "test" or strategy_name == "test_strategy":
         continue
      if points_collection.find_one({"strategy": strategy_name})["total_points"] > 0:
         
         heapq.heappush(q, (points_collection.find_one({"strategy": strategy_name})["total_points"] * 2 + (strategy_doc["portfolio_value"]), strategy_doc["successful_trades"] - strategy_doc["failed_trades"], strategy_doc["amount_cash"], strategy_doc["strategy"]))
      else:
         heapq.heappush(q, (strategy_doc["portfolio_value"], strategy_doc["successful_trades"] - strategy_doc["failed_trades"], strategy_doc["amount_cash"], strategy_doc["strategy"]))
   rank = 1
   while q:
      
      _, _, _, strategy_name = heapq.heappop(q)
      rank_collection.insert_one({"strategy": strategy_name, "rank": rank})
      rank+=1
   
   """
   Delete historical database so new one can be used tomorrow
   """
   db = client.HistoricalDatabase
   collection = db.HistoricalDatabase
   collection.delete_many({})
   print("Successfully updated ranks")
   print("Successfully deleted historical database")
   
def main():  
   """  
   Main function to control the workflow based on the market's status.  
   """  
   if rank_mode == 'live':
      ndaq_tickers = []  
      early_hour_first_iteration = True
      post_market_hour_first_iteration = True
   
   
      while True: 
         mongo_client = MongoClient(mongo_url, tlsCAFile=ca)
      
         status = mongo_client.market_data.market_status.find_one({})["market_status"]
      
         if status == "open":  
            # Connection pool is not thread safe. Create a new client for each thread.
            # We can use ThreadPoolExecutor to manage threads - maybe use this but this risks clogging
            # resources if we have too many threads or if a thread is on stall mode
            # We can also use multiprocessing.Pool to manage threads
         
            if not ndaq_tickers:
               logging.info("Market is open. Processing strategies.")  
               ndaq_tickers = get_ndaq_tickers(mongo_client, FINANCIAL_PREP_API_KEY)

            threads = []

            for ticker in ndaq_tickers:
               thread = threading.Thread(target=process_ticker, args=(ticker, mongo_client))
               threads.append(thread)
               thread.start()

            # Wait for all threads to complete
            for thread in threads:
               thread.join()

         
         

            logging.info("Finished processing all strategies. Waiting for 30 seconds.")
            time.sleep(30)  
      
         elif status == "early_hours":  
               # During early hour, currently we only support prep
               # However, we should add more features here like premarket analysis
            
               if early_hour_first_iteration is True:  
               
                  ndaq_tickers = get_ndaq_tickers(mongo_client, FINANCIAL_PREP_API_KEY)  
                  early_hour_first_iteration = False  
                  post_market_hour_first_iteration = True
                  logging.info("Market is in early hours. Waiting for 30 seconds.")  
               time.sleep(30)  
  
         elif status == "closed":  
            # Performs post-market analysis for next trading day
            # Will only run once per day to reduce clogging logging
            # Should self-implementing a delete log process after a certain time - say 1 year
        
            if post_market_hour_first_iteration is True:
               early_hour_first_iteration = True
               logging.info("Market is closed. Performing post-market analysis.") 
               post_market_hour_first_iteration = False
               # Update time delta based on the mode
               
               if time_delta_mode == 'additive':
                  mongo_client.trading_simulator.time_delta.update_one({}, {"$inc": {"time_delta": time_delta_increment}})
               elif time_delta_mode == 'multiplicative':
                  mongo_client.trading_simulator.time_delta.update_one({}, {"$mul": {"time_delta": time_delta_multiplicative}})
               elif time_delta_mode == 'balanced':
                  """
                  retrieve time_delta first
                  """
                  time_delta = mongo_client.trading_simulator.time_delta.find_one({})['time_delta']
                  mongo_client.trading_simulator.time_delta.update_one({}, {"$inc": {"time_delta": time_delta_balanced * time_delta}})
            
               #Update ranks
               update_portfolio_values(mongo_client)
               # We keep reusing the same mongo client and never close to reduce the number within the connection pool

               update_ranks(mongo_client)
            time.sleep(60)  
         else:  
            logging.error("An error occurred while checking market status.")  
            time.sleep(60)
         mongo_client.close()
   elif rank_mode == 'train':
      """
      initialize
      """
      ticker_price_history = {}
      trading_simulator = {}
      points = {}
      """
      need it for time_delta component and we need to adapt time delta for multiple modes - multiplicative, balanced or additive
      """
      for strategy in strategies:
         points[strategy.__name__] = 0
         trading_simulator[strategy.__name__] = {
               "holdings": {},
               "amount_cash": 50000,
               "total_trades": 0,
               "successful_trades": 0,
               "neutral_trades": 0,
               "failed_trades": 0,
               "portfolio_value": 50000
         }
      ideal_period = {}
      time_delta = 0.01
      mongo_client = MongoClient(mongo_url, tlsCAFile=ca)
      db = mongo_client.IndicatorsDatabase
      indicator_collection = db.Indicators
      for strategy in strategies:
         period = indicator_collection.find_one({'indicator': strategy.__name__})
         ideal_period[strategy.__name__] = period['ideal_period']
      
      for ticker in train_tickers:
         data = yf.Ticker(ticker).history(start=period_start, end=period_end, interval="1d")
         ticker_price_history[ticker] = data
      
      """
      now we have the data loaded, we need to simulate strategy for each day from start day to end day. create a loop that goes from start to end date
      """
      # Now simulate strategy for each day from start date to end date
      start_date = datetime.strptime(period_start, "%Y-%m-%d")
      end_date = datetime.strptime(period_end, "%Y-%m-%d")
      current_date = start_date
      
      def get_historical_data(ticker, current_date, period):
         period_start_date = {
               "1mo": current_date - timedelta(days=30),
               "3mo": current_date - timedelta(days=90),
               "6mo": current_date - timedelta(days=180),
               "1y": current_date - timedelta(days=365),
               "2y": current_date - timedelta(days=730)
         }
         start_date = period_start_date[period]
         
         return ticker_price_history[ticker].loc[start_date.strftime('%Y-%m-%d'):current_date.strftime('%Y-%m-%d')]
      
      def local_update_portfolio_values(current_date):
         active_count = 0
         for strategy in strategies:
               trading_simulator[strategy.__name__]["portfolio_value"] = trading_simulator[strategy.__name__]["amount_cash"]
               """
               update portfolio value here
               """
               amount = 0
               for ticker in trading_simulator[strategy.__name__]["holdings"]:
                  qty = trading_simulator[strategy.__name__]["holdings"][ticker]["quantity"]
                  current_price = ticker_price_history[ticker].loc[current_date.strftime('%Y-%m-%d')]["Close"]
                  amount += qty * current_price
               cash = trading_simulator[strategy.__name__]["amount_cash"]
               trading_simulator[strategy.__name__]["portfolio_value"] = amount + cash
               if trading_simulator[strategy.__name__]["portfolio_value"] != 50000:
                  active_count += 1
         return active_count
      while current_date <= end_date:
         print(f"Simulating strategies for date: {current_date.strftime('%Y-%m-%d')}")
         if current_date.weekday() >= 5:
            current_date += timedelta(days=1)
            continue
         if current_date.strftime('%Y-%m-%d') not in ticker_price_history[train_tickers[0]].index:
            current_date += timedelta(days=1)
            continue
         for ticker in train_tickers:
               """
               what we need to simulate:
               1. strategy - completed
               2. historical data - must give historical data that is ideal period days/months/years before the current date to current date
               3. current_price - get from trading simulator
               4. account_cash - get from trading_simulator
               5. holdings - get from trading_simulator
               6. total_portfolio_value
               """
               if current_date.strftime('%Y-%m-%d') in ticker_price_history[ticker].index:
                  daily_data = ticker_price_history[ticker].loc[current_date.strftime('%Y-%m-%d')]
                  current_price = daily_data['Close']
                  for strategy in strategies:
                     historical_data = get_historical_data(ticker, current_date, ideal_period[strategy.__name__])
                     account_cash = trading_simulator[strategy.__name__]["amount_cash"]
                     portfolio_qty = trading_simulator[strategy.__name__]["holdings"].get(ticker, {}).get("quantity", 0)
                     total_portfolio_value = trading_simulator[strategy.__name__]["portfolio_value"]
                     decision, qty = simulate_strategy(
                           strategy, ticker, current_price, historical_data, account_cash, portfolio_qty, total_portfolio_value
                     )
                     print(f"{strategy.__name__} - {decision} - {qty} - {ticker}")
                     """
                     now simulate the trade
                     """
                     if decision == "buy" and trading_simulator[strategy.__name__]["amount_cash"] > rank_liquidity_limit and qty > 0 and ((portfolio_qty + qty) * current_price) / total_portfolio_value < rank_asset_limit:
                        trading_simulator[strategy.__name__]["amount_cash"] -= qty * current_price
                        
                        if ticker in trading_simulator[strategy.__name__]["holdings"]:
                           trading_simulator[strategy.__name__]["holdings"][ticker]["quantity"] += qty
                        else:
                           trading_simulator[strategy.__name__]["holdings"][ticker] = {"quantity": qty}
                        
                        trading_simulator[strategy.__name__]["holdings"][ticker]["price"] = current_price
                        trading_simulator[strategy.__name__]["total_trades"] += 1
                           
                     elif decision == "sell" and trading_simulator[strategy.__name__]["holdings"].get(ticker, {}).get("quantity", 0) >= qty:
                        trading_simulator[strategy.__name__]["amount_cash"] += qty * current_price
                        if current_price > trading_simulator[strategy.__name__]["holdings"][ticker]["price"]:
                           trading_simulator[strategy.__name__]["successful_trades"] += 1
                           points[strategy.__name__] = points.get(strategy.__name__, 0) + time_delta * profit_profit_time_d1
                        elif current_price == trading_simulator[strategy.__name__]["holdings"][ticker]["price"]:
                           trading_simulator[strategy.__name__]["neutral_trades"] += 1
                        else:
                           trading_simulator[strategy.__name__]["failed_trades"] += 1
                           points[strategy.__name__] = points.get(strategy.__name__, 0) - time_delta * loss_profit_time_d1
                        trading_simulator[strategy.__name__]["holdings"][ticker]["quantity"] -= qty
                        if trading_simulator[strategy.__name__]["holdings"][ticker]["quantity"] == 0:
                           del trading_simulator[strategy.__name__]["holdings"][ticker]
                        elif trading_simulator[strategy.__name__]["holdings"][ticker]["quantity"] < 0:
                           Exception("Quantity cannot be negative")
                        trading_simulator[strategy.__name__]["total_trades"] += 1
         active_count = local_update_portfolio_values(current_date) 
         """
         log history of trading_simulator and points
         """
         logging.info(f"Trading simulator: {trading_simulator}")
         logging.info(f"Points: {points}")
         logging.info(f"Date: {current_date.strftime('%Y-%m-%d')}")
         logging.info(f"time_delta: {time_delta}")
         logging.info(f"Active count: {active_count}")
         logging.info("-------------------------------------------------")
         
         """
         Update time_delta based on the mode
         """
         if time_delta_mode == 'additive':
               time_delta += time_delta_increment
         elif time_delta_mode == 'multiplicative':
               time_delta *= time_delta_multiplicative
         elif time_delta_mode == 'balanced':
               time_delta += time_delta_balanced * time_delta


         current_date += timedelta(days=1)
         time.sleep(10)
            
      results = {
            "trading_simulator": trading_simulator,
            "points": points,
            "date": current_date.strftime('%Y-%m-%d'),
            "time_delta": time_delta
         }
         
      with open('training_results.json', 'w') as json_file:
         json.dump(results, json_file, indent=4)

      """
      output onto console top 10 strategies with highest portfolio values and top 10 strategies with highest points
      """
      top_portfolio_values = sorted(trading_simulator.items(), key=lambda x: x[1]["portfolio_value"], reverse=True)[:10]
      top_points = sorted(trading_simulator.items(), key=lambda x: x[1]["points"], reverse=True)[:10]
      print("Top 10 strategies with highest portfolio values")
      for strategy, value in top_portfolio_values:
         print(f"{strategy} - {value['portfolio_value']}")
      print("Top 10 strategies with highest points")
      for strategy, value in top_points:
         print(f"{strategy} - {value['points']}")
   elif rank_mode == 'test':
      return None
   
  
if __name__ == "__main__": 
   
   main()