Trading & ML Enhancements: Position Limits, Signal Validation, Agent Training, and Environments module update

.gitignore

@@ -279,4 +279,5 @@ research/*
 add_stage.sh
 catalog/
 *.session*
-*trades.db
+*trades.db
+current_open_positions.csv

packages/itbot/agents/__init__.py

@@ -1 +1,4 @@
 from .agent import *
+from .basic_ml_agent import *
+
+# ITBAgent => https://github.com/asavinov/intelligent-trading-bot

packages/itbot/agents/agent.py

@@ -1,6 +1,10 @@
 import asyncio
 from abc import ABC, abstractmethod
-from typing import List, Optional
+import os
+import joblib
+from typing import Dict, List, Optional
+
+import pandas as pd
 from packages.itbot.itbot import Signal
 import threading
 import logging
@@ -13,31 +17,73 @@ class Agent(ABC):
     The agent continuously waits for data and sends signals asynchronously as an event-driven program.
     """
 
-    def __init__(self, logger: Optional[Logger] = None):
+    def __init__(
+        self,
+        selected_symbols: List[str] = ["EURUSD", "BTCUSD", "XAUUSD"],
+        logger: Optional[Logger] = None,
+    ):
+        """Initialize the trading bot with logging and a list of trading symbols.
+
+        Args:
+            selected_symbols (List[str]): List of symbols to trade. Default is ["EURUSD", "BTCUSD","XAUUSD"].
+            logger (Optional[Logger]): An optional logger instance. If not provided, a default logger will be created.
+        """
         # Set up logging
-        self.logger = logger or Logger(name="it_bot", log_level=logging.DEBUG, filename="ITBot.log")
+        self.logger = logger or Logger(name="it_bot", level=logging.DEBUG, filename="ITBot.log")
 
-        self.model = None
+        # Initialize the signal queue for asynchronous handling
         self.signals_queue = asyncio.Queue()
-        self.loop = None
 
-    @abstractmethod
-    def load_model(self, model_path: str) -> None:
+        # Initialize the event loop
+        self.loop = asyncio.get_event_loop()  # Use the current event loop
+
+        # Store the selected symbols for trading
+        self.selected_symbols = selected_symbols
+        self.models = {
+            symbol: None for symbol in self.selected_symbols
+        }  # Initialize models for symbols
+
+        # Log the initialized symbols
+        self.logger.debug(f"Initialized with symbols: {self.selected_symbols}")
+
+    def load_models(self, model_path: str) -> None:
         """
-        Load the trained model from the specified path.
+        Load the trained models from the specified path.
 
         Args:
-            model_path (str): The file path of the trained model.
+            model_path (str): The directory path where the trained models are stored.
+
+        Raises:
+            ValueError: If the provided path is not a directory or if the model file is not found.
         """
-        pass
+        # Check if the provided path is a valid directory
+        if not os.path.isdir(model_path):
+            raise ValueError(f"The provided path '{model_path}' is not a valid directory.")
+
+        # Load models for each symbol from the specified directory
+        for symbol in self.selected_symbols:
+            model_file = os.path.join(
+                model_path, f"{symbol}_voting.joblib"
+            )  # Adjust the file naming as needed
+
+            # Check if the model file exists
+            if os.path.isfile(model_file):
+                try:
+                    self.models[symbol] = joblib.load(model_file)
+                    self.logger.info(f"Model for {symbol} loaded successfully from {model_file}.")
+                except Exception as e:
+                    self.logger.error(f"Error loading model for {symbol}: {e}")
+            else:
+                self.logger.warning(f"No model file found for {symbol} at {model_file}.")
 
     @abstractmethod
-    async def generate_signals(self, data: dict) -> List[Signal]:
+    async def generate_signals(self, symbol: str, data: pd.DataFrame) -> List[Signal]:
         """
-        Asynchronously generate trading signals using the loaded model and input data.
+        Generate trading signals using the loaded model and input data.
 
         Args:
-            data (dict): The data for the agent to make decisions.
+            symbol (str): The trading symbol for which signals are being generated.
+            data (pd.DataFrame): The input data used by the agent to make trading decisions.
 
         Returns:
             List[Signal]: A list of trading signals generated by the model.