""" 新闻流量情绪分析模块 实现情绪指数、情绪分类、流量阶段判断、情绪动量计算 """ import logging from datetime import datetime from typing import Dict, List, Optional, Tuple import statistics logging.basicConfig(level=logging.INFO) logger = logging.getLogger(__name__) class SentimentAnalyzer: """情绪分析器""" def __init__(self): # 情绪分类阈值 self.sentiment_thresholds = { 'extremely_pessimistic': 20, # 极度悲观 'pessimistic': 40, # 悲观 'neutral': 60, # 中性 'optimistic': 80, # 乐观 'extremely_optimistic': 100, # 极度乐观 } # 流量阶段定义 self.flow_stages = { 'startup': '启动', # 刚开始发酵 'acceleration': '加速', # 增速加快 'divergence': '分歧', # 多空分歧 'consensus': '一致', # 流量高潮（危险！） 'decline': '退潮', # 热度下降 } # 阶段判断阈值 self.stage_thresholds = { 'startup_growth': 0.20, # 启动期增速阈值 'acceleration_growth': 0.20, # 加速期增速阈值 'divergence_volatility': 0.30, # 分歧期波动率阈值 'consensus_k': 1.5, # 一致期K值阈值 'decline_growth': -0.20, # 退潮期增速阈值 } # 正面/负面关键词（用于情绪分析） self.positive_keywords = [ '利好', '大涨', '暴涨', '涨停', '新高', '突破', '牛市', '反弹', '加仓', '买入', '增持', '推荐', '看好', '机遇', '政策支持', '业绩预增', '超预期', '景气度', '高增长', ] self.negative_keywords = [ '利空', '大跌', '暴跌', '跌停', '新低', '破位', '熊市', '回调', '减仓', '卖出', '减持', '风险', '看空', '危机', '政策收紧', '业绩下滑', '不及预期', '亏损', '退市', ] def calculate_sentiment_index(self, platforms_data: List[Dict], stock_news: List[Dict] = None) -> Dict: """ 计算情绪指数（0-100） 基于以下因素： 1. 流量规模（占40%） 2. 财经平台活跃度（占30%） 3. 正负面关键词比例（占30%） Returns: { 'sentiment_index': int, 'flow_factor': int, 'finance_factor': int, 'keyword_factor': int, 'sentiment_class': str, 'analysis': str, } """ # 1. 流量规模因子（40%） total_news = 0 finance_news = 0 for platform_data in platforms_data: if not platform_data.get('success'): continue count = platform_data.get('count', 0) total_news += count if platform_data.get('category') == 'finance': finance_news += count # 流量分数：基于新闻数量 if total_news >= 500: flow_factor = 90 elif total_news >= 300: flow_factor = 70 elif total_news >= 150: flow_factor = 50 elif total_news >= 50: flow_factor = 30 else: flow_factor = 10 # 2. 财经平台活跃度因子（30%） if total_news > 0: finance_ratio = finance_news / total_news finance_factor = min(int(finance_ratio * 200), 100) else: finance_factor = 50 # 3. 关键词情绪因子（30%） positive_count = 0 negative_count = 0 if stock_news: for news in stock_news: title = news.get('title', '') content = news.get('content', '') text = f"{title} {content}" for kw in self.positive_keywords: if kw in text: positive_count += 1 break for kw in self.negative_keywords: if kw in text: negative_count += 1 break total_sentiment_news = positive_count + negative_count if total_sentiment_news > 0: positive_ratio = positive_count / total_sentiment_news keyword_factor = int(positive_ratio * 100) else: keyword_factor = 50 # 中性 # 4. 综合计算情绪指数 sentiment_index = int( flow_factor * 0.4 + finance_factor * 0.3 + keyword_factor * 0.3 ) # 限制范围 sentiment_index = max(0, min(100, sentiment_index)) # 5. 情绪分类 sentiment_class = self.classify_sentiment(sentiment_index) # 6. 生成分析 analysis = self._generate_sentiment_analysis( sentiment_index, sentiment_class, flow_factor, finance_factor, keyword_factor ) return { 'sentiment_index': sentiment_index, 'flow_factor': flow_factor, 'finance_factor': finance_factor, 'keyword_factor': keyword_factor, 'positive_count': positive_count, 'negative_count': negative_count, 'sentiment_class': sentiment_class, 'analysis': analysis, } def classify_sentiment(self, index: int) -> str: """ 情绪分类 极度悲观(<20) / 悲观(20-40) / 中性(40-60) / 乐观(60-80) / 极度乐观(>80) """ if index < 20: return "极度悲观" elif index < 40: return "悲观" elif index < 60: return "中性" elif index < 80: return "乐观" else: return "极度乐观" def _generate_sentiment_analysis(self, index: int, sentiment_class: str, flow: int, finance: int, keyword: int) -> str: """生成情绪分析文本""" if sentiment_class == "极度乐观": return f"情绪指数{index}，市场极度乐观！流量爆发({flow})，财经活跃({finance})。警告：可能是情绪顶部，注意获利了结。" elif sentiment_class == "乐观": return f"情绪指数{index}，市场情绪乐观。题材正在发酵，可关注龙头机会，但需注意节奏。" elif sentiment_class == "中性": return f"情绪指数{index}，市场情绪中性。缺乏明确方向，建议观望为主。" elif sentiment_class == "悲观": return f"情绪指数{index}，市场情绪偏悲观。负面因素较多，控制仓位，等待转机。" else: # 极度悲观 return f"情绪指数{index}，市场极度悲观！恐慌情绪蔓延。可能是超跌机会，但需谨慎左侧布局。" def determine_flow_stage(self, history_scores: List[int], current_score: int, current_k: float = None) -> Dict: """ 判断流量阶段 流量阶段： 1. 启动 - 刚开始发酵，关注 2. 加速 - 增速加快，参与 3. 分歧 - 多空分歧，谨慎 4. 一致 - 流量高潮，危险！准备跑路 5. 退潮 - 热度下降，及时止盈止损 Returns: { 'stage': str, 'stage_name': str, 'confidence': int, 'signal': str, # 关注/参与/谨慎/危险/离场 'analysis': str, } """ if len(history_scores) < 3: return { 'stage': 'unknown', 'stage_name': '未知', 'confidence': 0, 'signal': '观察', 'analysis': '历史数据不足，继续观察积累数据', } # 计算增长率序列 all_scores = history_scores + [current_score] growth_rates = [] for i in range(1, len(all_scores)): if all_scores[i-1] > 0: rate = (all_scores[i] - all_scores[i-1]) / all_scores[i-1] growth_rates.append(rate) # 计算关键指标 recent_growth_rates = growth_rates[-3:] if len(growth_rates) >= 3 else growth_rates avg_growth = sum(recent_growth_rates) / len(recent_growth_rates) if recent_growth_rates else 0 # 波动率（标准差） if len(recent_growth_rates) >= 2: volatility = statistics.stdev(recent_growth_rates) else: volatility = 0 # 计算K值（如果没有提供） if current_k is None and len(all_scores) >= 2: previous_score = all_scores[-2] current_k = current_score / previous_score if previous_score > 0 else 1.0 elif current_k is None: current_k = 1.0 # 判断上升/下降趋势 positive_count = sum(1 for r in recent_growth_rates if r > 0) negative_count = sum(1 for r in recent_growth_rates if r < 0) # 阶段判断逻辑 stage = 'unknown' stage_name = '未知' confidence = 50 signal = '观察' analysis = '' # 1. 一致阶段（最危险）- K值>1.5 且 高增速 if current_k >= 1.5 and avg_growth > 0.3: stage = 'consensus' stage_name = '一致' confidence = 90 signal = '危险' analysis = f"流量高潮！K值={current_k:.2f}，增速{avg_growth*100:.1f}%。市场一致看多，这往往是顶部信号。立即减仓或清仓！" # 2. 退潮阶段 - 连续下降 elif negative_count >= 2 and avg_growth < -0.15: stage = 'decline' stage_name = '退潮' confidence = 85 signal = '离场' analysis = f"流量退潮！连续下降，增速{avg_growth*100:.1f}%。题材热度消退，及时止盈止损，不要恋战。" # 3. 分歧阶段 - 高波动，涨跌交替 elif volatility > 0.25 and positive_count > 0 and negative_count > 0: stage = 'divergence' stage_name = '分歧' confidence = 75 signal = '谨慎' analysis = f"多空分歧！波动率{volatility*100:.1f}%，市场观点不一。高抛低吸，控制仓位，设好止损。" # 4. 加速阶段 - 持续上涨，增速加快 elif avg_growth > 0.2 and current_k > 1.1 and positive_count >= 2: stage = 'acceleration' stage_name = '加速' confidence = 80 signal = '参与' analysis = f"流量加速！增速{avg_growth*100:.1f}%，K值{current_k:.2f}。题材正在快速发酵，可参与龙头，但注意仓位。" # 5. 启动阶段 - 刚开始上涨 elif avg_growth > 0.05 and positive_count >= 2: stage = 'startup' stage_name = '启动' confidence = 70 signal = '关注' analysis = f"流量启动！增速{avg_growth*100:.1f}%，题材刚开始发酵。可以关注，等待确认后介入。" # 6. 其他情况 else: stage = 'stable' stage_name = '平稳' confidence = 60 signal = '观察' analysis = f"流量平稳。增速{avg_growth*100:.1f}%，无明显趋势。保持观望，等待方向明确。" return { 'stage': stage, 'stage_name': stage_name, 'confidence': confidence, 'signal': signal, 'analysis': analysis, 'avg_growth': round(avg_growth * 100, 1), 'volatility': round(volatility * 100, 1), 'current_k': round(current_k, 2), } def calculate_momentum(self, history_data: List[Dict]) -> Dict: """ 计算情绪动量 情绪动量 = 当前变化速率 / 平均变化速率 动量 > 1.5: 情绪加速 动量 ≈ 1: 情绪稳定 动量 < 0.5: 情绪减速 Args: history_data: 历史数据列表，每项包含 'sentiment_index' 和 'timestamp' Returns: { 'momentum': float, 'momentum_level': str, 'trend': str, 'analysis': str, } """ if len(history_data) < 3: return { 'momentum': 1.0, 'momentum_level': '正常', 'trend': '数据不足', 'analysis': '历史数据不足，无法计算动量', } # 提取情绪指数序列 sentiment_values = [d.get('sentiment_index', 50) for d in history_data] # 计算变化率序列 changes = [] for i in range(1, len(sentiment_values)): change = sentiment_values[i] - sentiment_values[i-1] changes.append(abs(change)) # 计算平均变化率 avg_change = sum(changes) / len(changes) if changes else 1 # 计算当前变化率（最近的变化） current_change = abs(changes[-1]) if changes else 0 # 计算动量 if avg_change > 0: momentum = round(current_change / avg_change, 2) else: momentum = 1.0 # 确定动量级别和趋势 if momentum >= 2.0: momentum_level = '极高' trend = '急剧变化' analysis = f"情绪动量{momentum}，情绪正在急剧变化！市场可能出现转折点，密切关注。" elif momentum >= 1.5: momentum_level = '高' trend = '加速变化' analysis = f"情绪动量{momentum}，情绪变化正在加速。趋势可能强化或反转。" elif momentum >= 0.8: momentum_level = '正常' trend = '稳定' analysis = f"情绪动量{momentum}，情绪变化平稳，市场处于正常状态。" elif momentum >= 0.3: momentum_level = '低' trend = '减速' analysis = f"情绪动量{momentum}，情绪变化正在减速，可能进入盘整期。" else: momentum_level = '极低' trend = '停滞' analysis = f"情绪动量{momentum}，情绪几乎没有变化，市场陷入僵局。" # 判断方向 if len(sentiment_values) >= 2: direction = sentiment_values[-1] - sentiment_values[-2] if direction > 0: trend += "（向上）" elif direction < 0: trend += "（向下）" return { 'momentum': momentum, 'momentum_level': momentum_level, 'trend': trend, 'current_change': current_change, 'avg_change': round(avg_change, 1), 'analysis': analysis, } def run_full_sentiment_analysis(self, platforms_data: List[Dict], stock_news: List[Dict], history_scores: List[int], current_score: int, history_sentiments: List[Dict] = None) -> Dict: """ 运行完整的情绪分析 Returns: { 'sentiment': Dict, # 情绪指数 'flow_stage': Dict, # 流量阶段 'momentum': Dict, # 情绪动量 'summary': str, # 总结 'risk_level': str, # 风险等级 'advice': str, # 操作建议 } """ # 1. 计算情绪指数 sentiment = self.calculate_sentiment_index(platforms_data, stock_news) # 2. 判断流量阶段 flow_stage = self.determine_flow_stage(history_scores, current_score) # 3. 计算情绪动量 if history_sentiments and len(history_sentiments) >= 3: momentum = self.calculate_momentum(history_sentiments) else: momentum = { 'momentum': 1.0, 'momentum_level': '正常', 'trend': '数据不足', 'analysis': '历史情绪数据不足', } # 4. 综合风险评估 risk_level, advice = self._assess_risk(sentiment, flow_stage, momentum) # 5. 生成总结 summary = self._generate_summary(sentiment, flow_stage, momentum, risk_level) return { 'sentiment': sentiment, 'flow_stage': flow_stage, 'momentum': momentum, 'summary': summary, 'risk_level': risk_level, 'advice': advice, 'timestamp': datetime.now().strftime('%Y-%m-%d %H:%M:%S'), } def _assess_risk(self, sentiment: Dict, flow_stage: Dict, momentum: Dict) -> Tuple[str, str]: """综合风险评估""" risk_score = 0 # 情绪因素 sentiment_index = sentiment['sentiment_index'] if sentiment_index > 85: risk_score += 3 # 过度乐观 elif sentiment_index < 25: risk_score += 2 # 过度悲观 # 流量阶段因素 stage = flow_stage['stage'] if stage == 'consensus': risk_score += 4 # 一致阶段最危险 elif stage == 'decline': risk_score += 3 elif stage == 'divergence': risk_score += 2 # 动量因素 momentum_value = momentum['momentum'] if momentum_value > 2.0: risk_score += 2 # 变化过快 # 确定风险等级和建议 if risk_score >= 7: risk_level = "极高" advice = "立即减仓或清仓！市场处于极端状态，控制风险为第一要务。" elif risk_score >= 5: risk_level = "高" advice = "谨慎操作，建议减仓。不追高，设置严格止损。" elif risk_score >= 3: risk_level = "中等" advice = "正常操作，注意仓位控制。逢高减仓，逢低观察。" elif risk_score >= 1: risk_level = "低" advice = "可适度参与，关注龙头机会。" else: risk_level = "极低" advice = "风险较低，可积极参与，但仍需设置止损。" return risk_level, advice def _generate_summary(self, sentiment: Dict, flow_stage: Dict, momentum: Dict, risk_level: str) -> str: """生成综合总结""" lines = [ f"【情绪】{sentiment['sentiment_class']}（{sentiment['sentiment_index']}分）", f"【阶段】{flow_stage['stage_name']}期，信号：{flow_stage['signal']}", f"【动量】{momentum['momentum_level']}，趋势{momentum['trend']}", f"【风险】{risk_level}", ] return '\n'.join(lines) # 全局实例 sentiment_analyzer = SentimentAnalyzer() # 测试代码 if __name__ == "__main__": print("=== 测试情绪分析模块 ===") # 模拟数据 platforms_data = [ {'success': True, 'category': 'finance', 'count': 100}, {'success': True, 'category': 'social', 'count': 200}, {'success': True, 'category': 'news', 'count': 150}, ] stock_news = [ {'title': 'AI板块大涨，龙头股涨停', 'content': '利好政策推动'}, {'title': '新能源概念股反弹', 'content': '业绩预增超预期'}, {'title': '市场观望情绪浓厚', 'content': '回调风险加大'}, ] history_scores = [300, 350, 420, 500, 580] current_score = 650 history_sentiments = [ {'sentiment_index': 55}, {'sentiment_index': 60}, {'sentiment_index': 68}, {'sentiment_index': 72}, ] # 运行完整分析 result = sentiment_analyzer.run_full_sentiment_analysis( platforms_data, stock_news, history_scores, current_score, history_sentiments ) print(f"\n情绪指数: {result['sentiment']['sentiment_index']} ({result['sentiment']['sentiment_class']})") print(f"流量阶段: {result['flow_stage']['stage_name']} - {result['flow_stage']['signal']}") print(f"情绪动量: {result['momentum']['momentum']} ({result['momentum']['momentum_level']})") print(f"风险等级: {result['risk_level']}") print(f"\n===总结===\n{result['summary']}") print(f"\n操作建议: {result['advice']}")