""" 美股交易智能体 - 主控制器（LLM 驱动版） 只进行市场分析和通知，不执行模拟交易 """ import asyncio from typing import Dict, Any, List, Optional from datetime import datetime, timedelta import pandas as pd from app.utils.logger import logger from app.config import get_settings from app.services.yfinance_service import get_yfinance_service from app.services.feishu_service import get_feishu_service from app.services.telegram_service import get_telegram_service from app.crypto_agent.llm_signal_analyzer import LLMSignalAnalyzer class StockAgent: """美股交易信号智能体（LLM 驱动，仅分析通知）""" def __init__(self): """初始化智能体""" self.settings = get_settings() self.yfinance = get_yfinance_service() self.feishu = get_feishu_service() self.telegram = get_telegram_service() self.llm_analyzer = LLMSignalAnalyzer() # 状态管理 self.last_signals: Dict[str, Dict[str, Any]] = {} self.signal_cooldown: Dict[str, datetime] = {} # 配置 self.symbols = self.settings.stock_symbols.split(',') # 运行状态 self.running = False self._event_loop = None self._task = None logger.info(f"美股智能体初始化完成，监控股票: {self.symbols}") async def start(self): """启动智能体""" if self.running: logger.warning("美股智能体已在运行中") return self.running = True self._event_loop = asyncio.get_event_loop() logger.info("美股智能体已启动") # 启动分析任务 self._task = asyncio.create_task(self._analysis_loop()) async def stop(self): """停止智能体""" self.running = False if self._task: self._task.cancel() try: await self._task except asyncio.CancelledError: pass logger.info("美股智能体已停止") async def _analysis_loop(self): """分析循环 - 只在美股交易时间内运行""" while self.running: try: # 检查是否在美股交易时间 if not self._is_market_hours(): # 不在交易时间，等待 10 分钟后再次检查 logger.debug("非美股交易时间，等待中...") await asyncio.sleep(600) continue # 在交易时间内，分析所有股票 for symbol in self.symbols: if not self.running: break await self.analyze_symbol(symbol) # 等待 1 小时后进行下次分析 logger.info("本次分析完成，等待 1 小时后进行下次分析...") await asyncio.sleep(3600) except Exception as e: logger.error(f"分析循环出错: {e}") await asyncio.sleep(60) # 出错后等待 1 分钟再重试 def _is_market_hours(self) -> bool: """ 判断当前是否在美股交易时间 美股交易时间: 周一至周五 9:30-16:00 (EST) 北京时间: - 冬令时 (11月-3月): 22:30-05:00 (次日) - 夏令时 (3月-11月): 21:30-04:00 (次日) Returns: 是否在交易时间 """ from datetime import datetime # 获取当前时间 now = datetime.now() # 检查是否为周末 if now.weekday() >= 5: # 5=周六, 6=周日 return False # 获取当前小时和分钟 hour = now.hour minute = now.minute current_time = hour * 100 + minute # 转换为数字，如 2130 表示 21:30 # 判断夏令时/冬令时（简单判断：3-11月为夏令时） is_summer = 3 <= now.month <= 11 if is_summer: # 夏令时: 21:30-04:00 (次日) # 即 2130-2359 或 0000-0400 if current_time >= 2130 or current_time < 400: return True else: # 冬令时: 22:30-05:00 (次日) # 即 2230-2359 或 0000-0500 if current_time >= 2230 or current_time < 500: return True return False async def analyze_symbol(self, symbol: str): """ 分析单个股票 Args: symbol: 股票代码 """ try: # 1. 获取多时间周期数据 data = self.yfinance.get_multi_timeframe_data(symbol) # 2. 验证数据完整性 if not self._validate_data(data): logger.warning(f"{symbol} 数据不完整，跳过本次分析") return # 3. 获取当前价格 ticker = self.yfinance.get_ticker(symbol) if not ticker: logger.warning(f"无法获取 {symbol} 当前价格") return current_price = ticker['lastPrice'] logger.info(f"\n{'='*60}") logger.info(f"📊 分析 {symbol} @ ${current_price:,.2f}") logger.info(f"{'='*60}") # 4. LLM 分析 logger.info(f"\n🤖 【LLM 分析中...】") result = await self.llm_analyzer.analyze( symbol, data, symbols=self.symbols, position_info=None # 美股不跟踪持仓 ) # 输出分析摘要 summary = result.get('analysis_summary', '无') logger.info(f" 市场状态: {summary}") # 输出新闻情绪 news_sentiment = result.get('news_sentiment', '') news_impact = result.get('news_impact', '') if news_sentiment: sentiment_icon = {'positive': '📈', 'negative': '📉', 'neutral': '➖'}.get(news_sentiment, '') logger.info(f" 新闻情绪: {sentiment_icon} {news_sentiment}") if news_impact: logger.info(f" 消息影响: {news_impact}") # 输出关键价位 levels = result.get('key_levels', {}) if levels.get('support') or levels.get('resistance'): support_str = ', '.join([f"${s:,.2f}" for s in levels.get('support', [])[:2]]) resistance_str = ', '.join([f"${r:,.2f}" for r in levels.get('resistance', [])[:2]]) logger.info(f" 支撑位: {support_str or '-'}") logger.info(f" 阻力位: {resistance_str or '-'}") # 5. 处理信号 signals = result.get('signals', []) if not signals: logger.info(f"\n⏸️ 结论: 无交易信号，继续观望") return # 输出所有信号 logger.info(f"\n🎯 【发现 {len(signals)} 个信号】") for sig in signals: signal_type = sig.get('type', 'unknown') type_map = {'short_term': '短线', 'medium_term': '中线', 'long_term': '长线'} type_text = type_map.get(signal_type, signal_type) action = sig.get('action', 'wait') action_map = {'buy': '🟢 做多', 'sell': '🔴 做空'} action_text = action_map.get(action, action) grade = sig.get('grade', 'D') confidence = sig.get('confidence', 0) grade_icon = {'A': '⭐⭐⭐', 'B': '⭐⭐', 'C': '⭐', 'D': ''}.get(grade, '') logger.info(f"\n {type_text} {action_text} [{grade}级{grade_icon}] {confidence}%") # 6. 过滤并通知最佳信号 best_signal = self._get_best_signal(signals) if not best_signal: logger.info(f"\n⏸️ 信号质量不高，不发送通知") return # 检查置信度阈值 threshold = self.settings.stock_llm_threshold * 100 if best_signal.get('confidence', 0) < threshold: logger.info(f"\n⏸️ 置信度不足 ({best_signal.get('confidence', 0)}% < {threshold}%)") return # 检查冷却时间 if not self._should_send_signal(symbol, best_signal): logger.info(f"\n⏸️ 信号冷却中，不发送通知") return # 发送通知 await self._send_signal_notification(symbol, best_signal, current_price) # 更新状态 self.last_signals[symbol] = best_signal self.signal_cooldown[symbol] = datetime.now() except Exception as e: logger.error(f"❌ 分析 {symbol} 出错: {e}") import traceback logger.error(traceback.format_exc()) def _get_best_signal(self, signals: List[Dict[str, Any]]) -> Optional[Dict[str, Any]]: """获取最佳信号""" # 过滤掉 D 级信号 valid_signals = [s for s in signals if s.get('grade', 'D') != 'D'] if not valid_signals: return None # 按等级和置信度排序 grade_order = {'A': 0, 'B': 1, 'C': 2} valid_signals.sort(key=lambda x: ( grade_order.get(x.get('grade', 'C'), 3), -x.get('confidence', 0) )) return valid_signals[0] def _should_send_signal(self, symbol: str, signal: Dict[str, Any]) -> bool: """判断是否应该发送信号""" action = signal.get('action', 'wait') if action == 'wait': return False # 检查冷却时间（60分钟内不重复发送相同方向的信号） if symbol in self.signal_cooldown: cooldown_end = self.signal_cooldown[symbol] + timedelta(minutes=60) if datetime.now() < cooldown_end: if symbol in self.last_signals: if self.last_signals[symbol].get('action') == action: logger.debug(f"{symbol} 信号冷却中，跳过") return False return True async def _send_signal_notification( self, symbol: str, signal: Dict[str, Any], current_price: float ): """发送信号通知""" try: # 使用正确的方法格式化信号 card = self.llm_analyzer.format_feishu_card(signal, symbol) title = card['title'] content = card['content'] # 发送到飞书 await self.feishu.send_markdown(title, content) # 发送到 Telegram await self.telegram.send_message(self.llm_analyzer.format_signal_message(signal, symbol)) logger.info(f"✅ 信号通知已发送: {title}") except Exception as e: logger.error(f"发送通知失败: {e}") def _validate_data(self, data: Dict[str, pd.DataFrame]) -> bool: """验证数据完整性""" required_intervals = ['1d', '1h'] for interval in required_intervals: if interval not in data or data[interval].empty: return False if len(data[interval]) < 20: return False return True async def analyze_once(self, symbol: str) -> Dict[str, Any]: """单次分析（用于测试或手动触发）""" data = self.yfinance.get_multi_timeframe_data(symbol) if not self._validate_data(data): return {'error': '数据不完整'} result = await self.llm_analyzer.analyze( symbol, data, symbols=self.symbols, position_info=None ) return result def get_status(self) -> Dict[str, Any]: """获取智能体状态""" return { 'running': self.running, 'symbols': self.symbols, 'mode': 'LLM 驱动（仅分析通知）', 'last_signals': { symbol: { 'type': sig.get('type'), 'action': sig.get('action'), 'confidence': sig.get('confidence'), 'grade': sig.get('grade') } for symbol, sig in self.last_signals.items() } } # 全局单例 _stock_agent: Optional[StockAgent] = None def get_stock_agent() -> StockAgent: """获取美股智能体单例""" global _stock_agent if _stock_agent is None: _stock_agent = StockAgent() return _stock_agent