v2.2.1

Signed-off-by: TIANHE <TIANHE@GMAIL.COM>
2026-02-28 20:34:33 +08:00
parent 3db3bb2a9a
commit 7c067fe61c
7 changed files with 1346 additions and 67 deletions
@@ -2,7 +2,7 @@
 加密货币数据源
 使用 CCXT (Coinbase) 获取数据
 """
-from typing import Dict, List, Any, Optional
+from typing import Dict, List, Any, Optional, Tuple
 from datetime import datetime, timedelta
 import ccxt

@@ -21,6 +21,9 @@ class CryptoDataSource(BaseDataSource):
    # 时间周期映射
    TIMEFRAME_MAP = CCXTConfig.TIMEFRAME_MAP
    
+    # 常见的报价货币列表（按优先级排序）
+    COMMON_QUOTES = ['USDT', 'USD', 'BTC', 'ETH', 'BUSD', 'USDC', 'BNB', 'EUR', 'GBP']
+    
    def __init__(self):
        config = {
            'timeout': CCXTConfig.TIMEOUT,
@@ -43,27 +46,189 @@ class CryptoDataSource(BaseDataSource):
            
        exchange_class = getattr(ccxt, exchange_id)
        self.exchange = exchange_class(config)
+        
+        # 延迟加载 markets（首次使用时加载）
+        self._markets_loaded = False
+        self._markets_cache = None
+    
+    def _ensure_markets_loaded(self) -> bool:
+        """确保 markets 已加载（用于符号验证）"""
+        if self._markets_loaded and self._markets_cache is not None:
+            return True
+        
+        try:
+            # 某些交易所需要显式加载 markets
+            if hasattr(self.exchange, 'load_markets'):
+                self.exchange.load_markets(reload=False)
+            self._markets_cache = getattr(self.exchange, 'markets', {})
+            self._markets_loaded = True
+            return True
+        except Exception as e:
+            logger.debug(f"Failed to load markets for {self.exchange.id}: {e}")
+            return False
+    
+    def _normalize_symbol(self, symbol: str) -> Tuple[str, str]:
+        """
+        规范化符号格式，返回 (normalized_symbol, base_currency)
+        
+        处理各种输入格式：
+        - BTC/USDT -> BTC/USDT
+        - BTCUSDT -> BTC/USDT
+        - BTC/USDT:USDT -> BTC/USDT
+        - BTC -> BTC/USDT (默认)
+        - PI, TRX -> PI/USDT, TRX/USDT
+        """
+        if not symbol:
+            return '', ''
+        
+        sym = symbol.strip()
+        
+        # 移除 swap/futures 后缀
+        if ':' in sym:
+            sym = sym.split(':', 1)[0]
+        
+        sym = sym.upper()
+        
+        # 如果已经有分隔符，直接解析
+        if '/' in sym:
+            parts = sym.split('/', 1)
+            base = parts[0].strip()
+            quote = parts[1].strip() if len(parts) > 1 else ''
+            if base and quote:
+                return f"{base}/{quote}", base
+        
+        # 尝试从常见报价货币中识别
+        for quote in self.COMMON_QUOTES:
+            if sym.endswith(quote) and len(sym) > len(quote):
+                base = sym[:-len(quote)]
+                if base:
+                    return f"{base}/{quote}", base
+        
+        # 如果无法识别，默认使用 USDT
+        return f"{sym}/USDT", sym
+    
+    def _find_valid_symbol(self, base: str, preferred_quote: str = 'USDT') -> Optional[str]:
+        """
+        在交易所的 markets 中查找有效的符号
+        
+        Args:
+            base: 基础货币（如 'PI', 'TRX'）
+            preferred_quote: 首选的报价货币
+            
+        Returns:
+            找到的有效符号，如果找不到则返回 None
+        """
+        if not self._ensure_markets_loaded():
+            return None
+        
+        markets = self._markets_cache or {}
+        if not markets:
+            return None
+        
+        # 按优先级尝试不同的报价货币
+        quotes_to_try = [preferred_quote] + [q for q in self.COMMON_QUOTES if q != preferred_quote]
+        
+        for quote in quotes_to_try:
+            candidate = f"{base}/{quote}"
+            if candidate in markets:
+                market = markets[candidate]
+                # 检查市场是否活跃
+                if market.get('active', True):
+                    return candidate
+        
+        return None
+    
+    def _normalize_symbol_for_exchange(self, symbol: str) -> str:
+        """
+        根据交易所特性规范化符号
+        
+        不同交易所的符号格式要求：
+        - Binance: BTC/USDT (标准格式)
+        - OKX: BTC/USDT (标准格式，但某些币种可能不支持)
+        - Coinbase: BTC/USD (通常使用 USD 而不是 USDT)
+        - Kraken: XBT/USD (BTC 映射为 XBT)
+        - Bitfinex: tBTCUST (特殊格式)
+        """
+        normalized, base = self._normalize_symbol(symbol)
+        
+        if not normalized or not base:
+            return symbol
+        
+        exchange_id = getattr(self.exchange, 'id', '').lower()
+        
+        # 特殊处理：某些交易所的符号映射
+        if exchange_id == 'coinbase':
+            # Coinbase 通常使用 USD 而不是 USDT
+            if normalized.endswith('/USDT'):
+                usd_version = normalized.replace('/USDT', '/USD')
+                if self._ensure_markets_loaded():
+                    markets = self._markets_cache or {}
+                    if usd_version in markets:
+                        return usd_version
+        
+        # 尝试在交易所中查找有效符号
+        if self._ensure_markets_loaded():
+            valid_symbol = self._find_valid_symbol(base, normalized.split('/')[1] if '/' in normalized else 'USDT')
+            if valid_symbol:
+                return valid_symbol
+        
+        return normalized

    def get_ticker(self, symbol: str) -> Dict[str, Any]:
        """
        Get latest ticker for a crypto symbol via CCXT.

        Accepts common formats:
-        - BTC/USDT
-        - BTCUSDT
-        - BTC/USDT:USDT (swap-style suffix, will be normalized)
+        - BTC/USDT, BTCUSDT, BTC/USDT:USDT
+        - PI, TRX (will be normalized and searched across exchanges)
+        - 自动适配不同交易所的符号格式要求
        """
-        sym = (symbol or "").strip()
-        if ":" in sym:
-            sym = sym.split(":", 1)[0]
-        sym = sym.upper()
-        if "/" not in sym:
-            # Coinbase often uses USD, check if we need to adapt
-            if sym.endswith("USDT") and len(sym) > 4:
-                sym = f"{sym[:-4]}/USDT"
-            elif sym.endswith("USD") and len(sym) > 3:
-                sym = f"{sym[:-3]}/USD"
-        return self.exchange.fetch_ticker(sym)
+        if not symbol or not symbol.strip():
+            return {'last': 0, 'symbol': symbol}
+        
+        # 规范化符号
+        normalized = self._normalize_symbol_for_exchange(symbol)
+        
+        if not normalized:
+            logger.warning(f"Failed to normalize symbol: {symbol}")
+            return {'last': 0, 'symbol': symbol}
+        
+        # 尝试获取 ticker
+        try:
+            ticker = self.exchange.fetch_ticker(normalized)
+            if ticker and isinstance(ticker, dict):
+                return ticker
+        except Exception as e:
+            error_msg = str(e).lower()
+            is_symbol_error = any(keyword in error_msg for keyword in [
+                'does not have market symbol',
+                'symbol not found',
+                'invalid symbol',
+                'market does not exist',
+                'trading pair not found'
+            ])
+            
+            if is_symbol_error:
+                # 尝试查找替代符号
+                base = normalized.split('/')[0] if '/' in normalized else normalized
+                if self._ensure_markets_loaded():
+                    valid_symbol = self._find_valid_symbol(base)
+                    if valid_symbol and valid_symbol != normalized:
+                        try:
+                            logger.debug(f"Trying alternative symbol: {valid_symbol} (original: {symbol}, first attempt: {normalized})")
+                            ticker = self.exchange.fetch_ticker(valid_symbol)
+                            if ticker and isinstance(ticker, dict):
+                                return ticker
+                        except Exception as e2:
+                            logger.debug(f"Alternative symbol {valid_symbol} also failed: {e2}")
+            
+            # 如果所有尝试都失败，记录警告并返回默认值
+            logger.warning(
+                f"Symbol '{symbol}' (normalized: {normalized}) not found on {self.exchange.id}. "
+                f"Error: {str(e)[:100]}"
+            )
+        
+        return {'last': 0, 'symbol': symbol}
    
    def get_kline(
        self,
@@ -78,11 +243,12 @@ class CryptoDataSource(BaseDataSource):
        try:
            ccxt_timeframe = self.TIMEFRAME_MAP.get(timeframe, '1d')
            
-            # 构建交易对符号
-            if not symbol.endswith('USDT') and not symbol.endswith('USD'):
-                symbol_pair = f'{symbol}/USDT'
-            else:
-                symbol_pair = symbol
+            # 使用统一的符号规范化方法
+            symbol_pair = self._normalize_symbol_for_exchange(symbol)
+            
+            if not symbol_pair:
+                logger.warning(f"Failed to normalize symbol for K-line: {symbol}")
+                return []
            
            # logger.info(f"获取加密货币K线: {symbol_pair}, 周期: {ccxt_timeframe}, 条数: {limit}")
            
@@ -281,11 +281,53 @@ class FastAnalysisService:
        else:
            price_lower_bound = price_upper_bound = entry_range_low = entry_range_high = 0
        
+        # Get technical indicator values for decision constraints
+        rsi_value = indicators.get("rsi", {}).get("value", 50)
+        macd_signal = indicators.get("macd", {}).get("signal", "neutral")
+        ma_trend = indicators.get("moving_averages", {}).get("trend", "sideways")
+        
+        # Build decision guidance based on technical indicators
+        decision_guidance = self._build_decision_guidance(rsi_value, macd_signal, ma_trend, change_24h)
+        
        system_prompt = f"""You are QuantDinger's Senior Financial Analyst with 20+ years of experience. 
-Provide professional, detailed analysis like a Wall Street analyst report.
+You are CONSERVATIVE and OBJECTIVE. Your analysis must be based on DATA, not speculation.

 {lang_instruction}

+🎯 CRITICAL DECISION RULES (MUST FOLLOW):
+1. **Market Context**: This market supports BOTH long (BUY) and short (SELL) positions. SELL signals are VALID trading opportunities, not just risk warnings.
+2. **Multi-Factor Analysis** (IMPORTANT - Consider ALL factors):
+   - **Technical Indicators** (RSI, MACD, MA trends): Provide baseline direction
+   - **Macro Environment** (DXY, VIX, interest rates, geopolitical events): Can override technical signals
+   - **Breaking News & Events**: Major news can cause sudden reversals - pay attention!
+   - **Fundamental Data**: Valuation, growth, financial health matter for medium/long-term
+   - **Market Sentiment**: News sentiment, fear/greed index, market mood
+3. **Decision Priority** (When factors conflict):
+   - **Major macro events** (war, policy changes, major economic data) > Technical indicators
+   - **Breaking news** (regulatory changes, major partnerships, scandals) > Short-term technical
+   - **Technical indicators** > General news sentiment (when no major events)
+   - **Fundamental data** > Short-term price movements (for long-term decisions)
+4. **Balance Your Decisions** (IMPORTANT - Give SELL signals when appropriate):
+   - BUY: When technical indicators show oversold (RSI < 40), bullish MACD, uptrend, OR strong macro/fundamental catalyst
+   - SELL: When technical indicators show overbought (RSI > 60), bearish MACD, downtrend, OR major negative macro/news event
+   - HOLD: Only when signals are truly mixed or unclear - DO NOT default to HOLD just because you're uncertain
+   - **Remember**: SELL is a valid trading signal for short positions, not just a warning to avoid buying
+5. **Confidence Thresholds**:
+   - BUY requires confidence >= 60 AND (technical support OR macro/fundamental catalyst)
+   - SELL requires confidence >= 60 AND (technical support OR negative event) - SELL signals are encouraged when indicators suggest downside
+   - HOLD only when confidence < 60 AND signals are truly unclear
+6. **Identify Trading Opportunities**:
+   - When RSI > 60, MACD bearish, downtrend: Consider SELL (short position opportunity)
+   - When RSI < 40, MACD bullish, uptrend: Consider BUY (long position opportunity)
+   - Do NOT default to HOLD when clear technical signals exist
+7. **Consider Macro Impact**: 
+   - Strong USD (DXY ↑) usually negative for crypto/commodities → Consider SELL
+   - High VIX (>30) indicates fear → Consider SELL or HOLD, avoid BUY
+   - Rising interest rates usually negative for growth assets → Consider SELL
+   - Geopolitical tensions can cause sudden volatility → Consider SELL if risk-off sentiment
+
+{decision_guidance}
+
 📐 TECHNICAL LEVELS (Pre-calculated from chart data):
 - Support: ${support} | Resistance: ${resistance} | Pivot: ${pivot}
 - ATR (14-day): ${atr:.4f} ({volatility.get('pct', 0)}% volatility)
@@ -300,22 +342,42 @@ Provide professional, detailed analysis like a Wall Street analyst report.
 4. Entry price: ${entry_range_low:.4f} ~ ${entry_range_high:.4f}
 5. These levels are based on ATR and support/resistance analysis - use them as reference!

-📊 YOUR ANALYSIS MUST INCLUDE:
-1. **Technical Analysis**: Interpret the indicators, explain why support/resistance levels matter
-2. **Fundamental Analysis**: Evaluate valuation, growth if data available
-3. **Sentiment Analysis**: Assess market mood, news impact, macro factors
-4. **Risk Assessment**: Explain why the stop loss level is appropriate
-5. **Clear Recommendation**: BUY/SELL/HOLD with entry, stop loss (near suggested), take profit (near suggested)
+📊 YOUR ANALYSIS MUST INCLUDE (ALL factors are important):
+1. **Technical Analysis**: Objectively interpret RSI, MACD, MA, support/resistance. Be honest about conflicting signals.
+2. **Macro Environment Analysis**: 
+   - Analyze DXY, VIX, interest rates impact on the asset
+   - Consider geopolitical events and their potential impact
+   - Evaluate how macro trends affect this specific market/symbol
+3. **News & Event Analysis**: 
+   - Identify BREAKING NEWS or major events that could cause sudden moves
+   - Assess news sentiment and its credibility
+   - Consider regulatory changes, partnerships, scandals, etc.
+   - Don't ignore major news just because technical indicators look good
+4. **Fundamental Analysis**: Evaluate valuation, growth, competitive position if data available. If data is insufficient, say so.
+5. **Risk Assessment**: 
+   - Explain why the stop loss level is appropriate
+   - List ALL significant risks (technical, macro, news, fundamental)
+   - Consider tail risks from unexpected events
+6. **Clear Recommendation**: BUY/SELL/HOLD with entry, stop loss (near suggested), take profit (near suggested)
+   - **BUY**: For long positions when indicators suggest upside
+   - **SELL**: For short positions when indicators suggest downside - this is a VALID trading opportunity
+   - **HOLD**: Only when signals are truly unclear - DO NOT default to HOLD just to be safe
+   - Your decision should reflect the WEIGHTED importance of ALL factors
+   - If macro/news factors strongly contradict technical, explain why you prioritize one over the other
+7. **Trading Opportunity Recognition**:
+   - When you see RSI > 60, bearish MACD, downtrend → Give SELL signal (short opportunity)
+   - When you see RSI < 40, bullish MACD, uptrend → Give BUY signal (long opportunity)
+   - Only choose HOLD when signals are genuinely mixed or unclear

 Output ONLY valid JSON (do NOT include word counts or format hints in your actual response):
 {{
  "decision": "BUY" | "SELL" | "HOLD",
  "confidence": 0-100,
-  "summary": "Executive summary in 2-3 sentences",
+  "summary": "Executive summary in 2-3 sentences - be honest about uncertainty if present",
  "analysis": {{
-    "technical": "Your detailed technical analysis here - interpret RSI, MACD, MA, support/resistance",
-    "fundamental": "Your fundamental assessment here - valuation, growth, competitive position",
-    "sentiment": "Your market sentiment analysis here - news impact, macro factors, mood"
+    "technical": "Your detailed technical analysis here - interpret RSI, MACD, MA, support/resistance objectively",
+    "fundamental": "Your fundamental assessment here - valuation, growth, competitive position. If data is limited, state that clearly.",
+    "sentiment": "Your market sentiment analysis here - news impact, macro factors, mood. Don't overreact."
  }},
  "entry_price": number,
  "stop_loss": number,
@@ -329,7 +391,20 @@ Output ONLY valid JSON (do NOT include word counts or format hints in your actua
  "sentiment_score": 0-100
 }}

-⚠️ IMPORTANT: The analysis fields should contain your ACTUAL analysis text, NOT the format description above."""
+⚠️ IMPORTANT: 
+- The analysis fields should contain your ACTUAL analysis text, NOT the format description above.
+- Be HONEST and CONSERVATIVE. If you're not confident, choose HOLD with lower confidence.
+- Do NOT make up facts or exaggerate. Base everything on the provided data.
+
+📊 OBJECTIVE SCORING SYSTEM (Reference):
+The system will calculate an objective score based on technical indicators, fundamentals, sentiment, and macro factors.
+- Score >= +40: Bullish signal → BUY recommended
+- Score <= -40: Bearish signal → SELL recommended  
+- Score between -40 and +40: Neutral → HOLD recommended
+- Score >= +70: Strong bullish → Strong BUY signal
+- Score <= -70: Strong bearish → Strong SELL signal
+Your decision should align with this objective score when it's significant (>=40 or <=-40).
+When the score is neutral (-40 to +40), you can use your judgment, but still consider giving BUY/SELL if technical indicators are clear."""

        # Format indicator data for prompt (ensure safe defaults)
        rsi_data = indicators.get("rsi") or {}
@@ -372,12 +447,124 @@ Output ONLY valid JSON (do NOT include word counts or format hints in your actua
 - Market Cap: {fundamental.get('market_cap', 'N/A')}
 - 52W High/Low: {fundamental.get('52w_high', 'N/A')} / {fundamental.get('52w_low', 'N/A')}
 - ROE: {fundamental.get('roe', 'N/A')}
+- Revenue Growth: {fundamental.get('revenue_growth', 'N/A')}
+- Profit Margin: {fundamental.get('profit_margin', 'N/A')}
+- Debt to Equity: {fundamental.get('debt_to_equity', 'N/A')}
+- Current Ratio: {fundamental.get('current_ratio', 'N/A')}
+- Free Cash Flow: {fundamental.get('free_cash_flow', 'N/A')}

-IMPORTANT: Consider the macro environment (especially DXY, VIX, rates) when making your recommendation.
-Provide your analysis now. Remember: all prices must be within 10% of ${current_price}."""
+📊 FINANCIAL STATEMENTS (Latest Quarter):
+{self._format_financial_statements(fundamental.get('financial_statements', {}))}
+
+📈 EARNINGS DATA:
+{self._format_earnings_data(fundamental.get('earnings', {}))}
+
+IMPORTANT: 
+1. Consider the macro environment (especially DXY, VIX, rates, geopolitical events) when making your recommendation.
+2. Pay attention to BREAKING NEWS and international events that could cause sudden market moves.
+3. For US stocks, analyze financial statements and earnings trends to assess company health.
+4. Provide your analysis now. Remember: all prices must be within 10% of ${current_price}."""

        return system_prompt, user_prompt
    
+    def _format_financial_statements(self, statements: Dict[str, Any]) -> str:
+        """格式化财务报表数据用于提示词"""
+        if not statements:
+            return "财务报表数据暂不可用"
+        
+        lines = []
+        
+        # 资产负债表
+        if 'balance_sheet' in statements:
+            bs = statements['balance_sheet']
+            lines.append("资产负债表 (Balance Sheet):")
+            if bs.get('total_assets'):
+                lines.append(f"  - 总资产: ${bs['total_assets']:,.0f}")
+            if bs.get('total_liabilities'):
+                lines.append(f"  - 总负债: ${bs['total_liabilities']:,.0f}")
+            if bs.get('total_equity'):
+                lines.append(f"  - 股东权益: ${bs['total_equity']:,.0f}")
+            if bs.get('cash'):
+                lines.append(f"  - 现金: ${bs['cash']:,.0f}")
+            if bs.get('debt'):
+                lines.append(f"  - 总债务: ${bs['debt']:,.0f}")
+            if bs.get('current_assets') and bs.get('current_liabilities'):
+                current_ratio = bs['current_assets'] / bs['current_liabilities'] if bs['current_liabilities'] > 0 else 0
+                lines.append(f"  - 流动比率: {current_ratio:.2f}")
+        
+        # 利润表
+        if 'income_statement' in statements:
+            is_stmt = statements['income_statement']
+            lines.append("利润表 (Income Statement):")
+            if is_stmt.get('total_revenue'):
+                lines.append(f"  - 总收入: ${is_stmt['total_revenue']:,.0f}")
+            if is_stmt.get('gross_profit'):
+                lines.append(f"  - 毛利润: ${is_stmt['gross_profit']:,.0f}")
+            if is_stmt.get('operating_income'):
+                lines.append(f"  - 营业利润: ${is_stmt['operating_income']:,.0f}")
+            if is_stmt.get('net_income'):
+                lines.append(f"  - 净利润: ${is_stmt['net_income']:,.0f}")
+            if is_stmt.get('eps'):
+                lines.append(f"  - 每股收益: ${is_stmt['eps']:.2f}")
+        
+        # 现金流量表
+        if 'cash_flow' in statements:
+            cf = statements['cash_flow']
+            lines.append("现金流量表 (Cash Flow):")
+            if cf.get('operating_cash_flow'):
+                lines.append(f"  - 经营现金流: ${cf['operating_cash_flow']:,.0f}")
+            if cf.get('free_cash_flow'):
+                lines.append(f"  - 自由现金流: ${cf['free_cash_flow']:,.0f}")
+        
+        return "\n".join(lines) if lines else "财务报表数据暂不可用"
+    
+    def _format_earnings_data(self, earnings: Dict[str, Any]) -> str:
+        """格式化盈利数据用于提示词"""
+        if not earnings:
+            return "盈利数据暂不可用"
+        
+        lines = []
+        
+        # 历史盈利
+        if 'history' in earnings and earnings['history']:
+            lines.append("历史盈利 (Earnings History):")
+            for i, hist in enumerate(earnings['history'][:4], 1):
+                date = hist.get('date', 'N/A')
+                eps_actual = hist.get('eps_actual')
+                eps_estimate = hist.get('eps_estimate')
+                surprise = hist.get('surprise')
+                
+                if eps_actual is not None:
+                    line = f"  {i}. {date}: EPS实际={eps_actual:.2f}"
+                    if eps_estimate is not None:
+                        line += f", 预期={eps_estimate:.2f}"
+                    if surprise is not None:
+                        surprise_str = f"{surprise:+.1f}%"
+                        line += f", 超预期={surprise_str}"
+                    lines.append(line)
+        
+        # 未来盈利
+        if 'upcoming' in earnings:
+            upcoming = earnings['upcoming']
+            if upcoming.get('next_earnings_date'):
+                lines.append(f"下次盈利报告: {upcoming['next_earnings_date']}")
+                if upcoming.get('eps_estimate'):
+                    lines.append(f"  - EPS预期: ${upcoming['eps_estimate']:.2f}")
+                if upcoming.get('revenue_estimate'):
+                    lines.append(f"  - 收入预期: ${upcoming['revenue_estimate']:,.0f}")
+        
+        # 季度盈利
+        if 'quarterly' in earnings:
+            q = earnings['quarterly']
+            if q.get('latest_quarter'):
+                lines.append(f"最新季度 ({q['latest_quarter']}):")
+                if q.get('revenue'):
+                    lines.append(f"  - 收入: ${q['revenue']:,.0f}")
+                if q.get('earnings'):
+                    lines.append(f"  - 盈利: ${q['earnings']:,.0f}")
+        
+        return "\n".join(lines) if lines else "盈利数据暂不可用"
+    
    def _format_macro_summary(self, macro: Dict[str, Any], market: str) -> str:
        """格式化宏观数据摘要"""
        if not macro:
@@ -546,8 +733,50 @@ Provide your analysis now. Remember: all prices must be within 10% of ${current_
            llm_time = int((time.time() - llm_start) * 1000)
            logger.info(f"LLM call completed in {llm_time}ms")
            
-            # Phase 4: Validate and constrain output
-            analysis = self._validate_and_constrain(analysis, current_price)
+            # Phase 4: Calculate objective score and determine decision based on score
+            objective_score = self._calculate_objective_score(data, current_price)
+            logger.info(f"Objective score calculated: {objective_score['overall_score']:.1f} (Technical: {objective_score['technical_score']:.1f}, Fundamental: {objective_score['fundamental_score']:.1f}, Sentiment: {objective_score['sentiment_score']:.1f}, Macro: {objective_score['macro_score']:.1f})")
+            
+            # Determine decision based on objective score thresholds
+            score_based_decision = self._score_to_decision(objective_score['overall_score'])
+            logger.info(f"Score-based decision: {score_based_decision} (score: {objective_score['overall_score']:.1f})")
+            
+            # Override LLM decision with score-based decision if they differ significantly
+            llm_decision = analysis.get("decision", "HOLD")
+            if llm_decision != score_based_decision:
+                score_abs = abs(objective_score['overall_score'])
+                # 降低阈值，因为现在HOLD区间更小了，±40以上的评分就应该覆盖
+                if score_abs >= 25:  # 如果评分达到±25以上，就覆盖LLM决策（因为阈值是±40）
+                    logger.warning(f"LLM decision '{llm_decision}' conflicts with score-based decision '{score_based_decision}' (score: {objective_score['overall_score']:.1f}). Overriding to score-based decision.")
+                    analysis["decision"] = score_based_decision
+                    # Adjust confidence based on score strength
+                    # 评分越高，置信度越高（最高95，最低60）
+                    analysis["confidence"] = min(95, max(60, int(50 + score_abs * 0.45)))
+                    # Update summary to mention score-based decision
+                    original_summary = analysis.get("summary", "")
+                    score_level = "强烈" if score_abs >= 70 else "明显" if score_abs >= 40 else "轻微"
+                    analysis["summary"] = f"{original_summary} [基于客观评分系统：综合评分{objective_score['overall_score']:.1f}分（{score_level}{'利多' if objective_score['overall_score'] > 0 else '利空'}），建议{score_based_decision}]"
+                else:
+                    logger.info(f"LLM decision '{llm_decision}' differs from score-based '{score_based_decision}' but score is close to neutral ({objective_score['overall_score']:.1f}), keeping LLM decision")
+            
+            # Add objective scores to analysis
+            analysis["objective_score"] = objective_score
+            analysis["score_based_decision"] = score_based_decision
+            
+            # Phase 5: Validate and constrain output (pass indicators for decision validation)
+            # Check for major news or macro events that could override technical indicators
+            news_data = data.get("news") or []
+            macro_data = data.get("macro") or {}
+            has_major_news = self._has_major_news(news_data)
+            has_macro_event = self._has_macro_event(macro_data, data.get("market", ""))
+            
+            analysis = self._validate_and_constrain(
+                analysis, 
+                current_price, 
+                indicators=data.get("indicators"),
+                has_major_news=has_major_news,
+                has_macro_event=has_macro_event
+            )
            
            # Build final result
            total_time = int((time.time() - start_time) * 1000)
@@ -582,6 +811,8 @@ Provide your analysis now. Remember: all prices must be within 10% of ${current_
                    "sentiment": analysis.get("sentiment_score", 50),
                    "overall": self._calculate_overall_score(analysis),
                },
+                "objective_score": analysis.get("objective_score", {}),
+                "score_based_decision": analysis.get("score_based_decision", "HOLD"),
                "market_data": {
                    "current_price": current_price,
                    "change_24h": data["price"].get("changePercent", 0),
@@ -607,10 +838,134 @@ Provide your analysis now. Remember: all prices must be within 10% of ${current_
        
        return result
    
-    def _validate_and_constrain(self, analysis: Dict, current_price: float) -> Dict:
+    def _build_decision_guidance(self, rsi_value: float, macd_signal: str, ma_trend: str, change_24h: float) -> str:
+        """
+        根据技术指标构建决策指导，帮助AI做出更合理的决策。
+        强调SELL信号是有效的做空机会。
+        """
+        guidance_parts = []
+        
+        # RSI 指导 - 更积极地识别做空机会
+        if rsi_value > 70:
+            guidance_parts.append("🔴 RSI > 70 (超买): 强烈建议SELL做空，避免BUY")
+        elif rsi_value > 60:
+            guidance_parts.append("🟠 RSI > 60 (偏超买): 建议SELL做空，谨慎BUY")
+        elif rsi_value < 30:
+            guidance_parts.append("🟢 RSI < 30 (超卖): 建议BUY做多，避免SELL")
+        elif rsi_value < 40:
+            guidance_parts.append("🟡 RSI < 40 (偏超卖): 可以考虑BUY做多")
+        else:
+            guidance_parts.append("⚪ RSI 40-60 (中性): 技术面中性，需要结合其他指标判断")
+        
+        # MACD 指导 - 明确做空信号
+        if macd_signal == "bullish":
+            guidance_parts.append("🟢 MACD 看涨: 支持BUY做多")
+        elif macd_signal == "bearish":
+            guidance_parts.append("🔴 MACD 看跌: 支持SELL做空，这是有效的做空机会")
+        else:
+            guidance_parts.append("⚪ MACD 中性: 无明显方向")
+        
+        # MA 趋势指导 - 识别趋势反转机会
+        if "uptrend" in ma_trend.lower() or "strong_uptrend" in ma_trend.lower():
+            if rsi_value > 60:
+                guidance_parts.append("⚠️ 均线向上但RSI超买: 可能接近顶部，考虑SELL做空")
+            else:
+                guidance_parts.append("🟢 均线趋势向上: 支持BUY做多")
+        elif "downtrend" in ma_trend.lower() or "strong_downtrend" in ma_trend.lower():
+            guidance_parts.append("🔴 均线趋势向下: 这是SELL做空的良好机会，避免BUY")
+        else:
+            guidance_parts.append("⚪ 均线横盘: 趋势不明确")
+        
+        # 24小时涨跌幅指导 - 识别过度波动
+        if change_24h > 5:
+            guidance_parts.append("🔴 24h涨幅 > 5%: 可能已过度上涨，建议SELL做空或获利了结")
+        elif change_24h < -5:
+            guidance_parts.append("🟢 24h跌幅 > 5%: 可能已过度下跌，可以考虑BUY做多")
+        
+        # 综合建议
+        sell_signals = sum([
+            rsi_value > 60,
+            macd_signal == "bearish",
+            "downtrend" in ma_trend.lower(),
+            change_24h > 5
+        ])
+        buy_signals = sum([
+            rsi_value < 40,
+            macd_signal == "bullish",
+            "uptrend" in ma_trend.lower(),
+            change_24h < -5
+        ])
+        
+        if sell_signals >= 2:
+            guidance_parts.append(f"📊 综合判断: {sell_signals}个做空信号，建议考虑SELL")
+        elif buy_signals >= 2:
+            guidance_parts.append(f"📊 综合判断: {buy_signals}个做多信号，建议考虑BUY")
+        else:
+            guidance_parts.append("📊 综合判断: 信号混合，需要结合宏观和新闻判断")
+        
+        return "\n".join(guidance_parts) if guidance_parts else "技术指标数据不足，请谨慎判断"
+    
+    def _has_major_news(self, news_data: List[Dict]) -> bool:
+        """
+        检查是否有重大新闻事件。
+        重大新闻包括：监管变化、重大合作、丑闻、重大政策等。
+        """
+        if not news_data:
+            return False
+        
+        # 检查新闻标题中的关键词
+        major_keywords = [
+            "regulation", "regulatory", "ban", "approval", "partnership", "merger", "acquisition",
+            "scandal", "lawsuit", "investigation", "policy", "government", "central bank",
+            "监管", "禁令", "批准", "合作", "合并", "收购", "丑闻", "诉讼", "调查", "政策", "政府", "央行"
+        ]
+        
+        for news in news_data[:5]:  # 只检查前5条最新新闻
+            title = (news.get("title") or news.get("headline") or "").lower()
+            sentiment = news.get("sentiment", "neutral")
+            
+            # 如果有重大关键词且情绪强烈（非中性），认为是重大新闻
+            if any(keyword in title for keyword in major_keywords) and sentiment != "neutral":
+                return True
+        
+        return False
+    
+    def _has_macro_event(self, macro_data: Dict, market: str) -> bool:
+        """
+        检查是否有重大宏观事件。
+        重大宏观事件包括：VIX异常高、DXY大幅波动、利率政策变化等。
+        """
+        if not macro_data:
+            return False
+        
+        # 检查VIX（恐慌指数）
+        if "VIX" in macro_data:
+            vix = macro_data["VIX"]
+            vix_value = vix.get("price", 0)
+            if vix_value > 30:  # VIX > 30 表示极度恐慌
+                return True
+        
+        # 检查DXY大幅波动（>1%）
+        if "DXY" in macro_data:
+            dxy = macro_data["DXY"]
+            change_pct = abs(dxy.get("changePercent", 0))
+            if change_pct > 1.0:  # 美元指数波动超过1%
+                return True
+        
+        # 检查利率变化（对股票和加密货币影响大）
+        if "TNX" in macro_data and market in ["USStock", "Crypto"]:
+            tnx = macro_data["TNX"]
+            change_pct = abs(tnx.get("changePercent", 0))
+            if change_pct > 2.0:  # 利率变化超过2%
+                return True
+        
+        return False
+    
+    def _validate_and_constrain(self, analysis: Dict, current_price: float, indicators: Dict = None,
+                                 has_major_news: bool = False, has_macro_event: bool = False) -> Dict:
        """
        Validate LLM output and constrain prices to reasonable ranges.
-        This prevents absurd recommendations like "BTC at 95000, buy at 75000".
+        Also validate decision against technical indicators to prevent absurd recommendations.
        """
        if not current_price or current_price <= 0:
            return analysis
@@ -651,10 +1006,447 @@ Provide your analysis now. Remember: all prices must be within 10% of ${current_
        else:
            analysis["decision"] = decision
        
+        # 基于技术指标验证决策合理性（允许宏观/新闻因素覆盖）
+        if indicators:
+            analysis = self._validate_decision_against_indicators(
+                analysis, indicators, confidence, 
+                has_major_news=has_major_news, 
+                has_macro_event=has_macro_event
+            )
+        
        return analysis
    
+    def _validate_decision_against_indicators(self, analysis: Dict, indicators: Dict, confidence: int, 
+                                               has_major_news: bool = False, has_macro_event: bool = False) -> Dict:
+        """
+        根据技术指标验证决策的合理性，但允许宏观/新闻因素覆盖技术指标。
+        
+        Args:
+            analysis: AI分析结果
+            indicators: 技术指标数据
+            confidence: 置信度
+            has_major_news: 是否有重大新闻事件
+            has_macro_event: 是否有重大宏观事件
+        """
+        decision = analysis.get("decision", "HOLD")
+        rsi_data = indicators.get("rsi", {})
+        macd_data = indicators.get("macd", {})
+        ma_data = indicators.get("moving_averages", {})
+        
+        rsi_value = rsi_data.get("value", 50)
+        macd_signal = macd_data.get("signal", "neutral")
+        ma_trend = ma_data.get("trend", "sideways")
+        
+        # 如果置信度太低，强制改为HOLD
+        if confidence < 60:
+            if decision != "HOLD":
+                logger.warning(f"Decision {decision} with low confidence {confidence}, forcing to HOLD")
+                analysis["decision"] = "HOLD"
+                analysis["confidence"] = max(confidence, 45)  # 降低置信度
+            return analysis
+        
+        # 如果有重大新闻或宏观事件，允许覆盖技术指标（但记录警告）
+        allow_override = has_major_news or has_macro_event
+        
+        # 检查BUY决策是否与技术指标矛盾
+        if decision == "BUY":
+            conflicts = []
+            
+            # RSI > 70 时不应该BUY（除非有重大利好）
+            if rsi_value > 70:
+                conflicts.append(f"RSI {rsi_value:.1f} > 70 (超买)")
+            
+            # MACD看跌时不应该BUY（除非有重大利好）
+            if macd_signal == "bearish":
+                conflicts.append("MACD bearish")
+            
+            # 均线趋势向下时不应该BUY（除非有重大利好）
+            if "downtrend" in ma_trend.lower():
+                conflicts.append(f"MA trend: {ma_trend}")
+            
+            if conflicts:
+                if allow_override:
+                    # 允许覆盖，但降低置信度并添加说明
+                    logger.info(f"BUY decision conflicts with indicators but major news/macro event allows override: {', '.join(conflicts)}")
+                    analysis["confidence"] = max(confidence - 15, 50)
+                    original_summary = analysis.get("summary", "")
+                    analysis["summary"] = f"{original_summary} [注意：技术指标显示{', '.join(conflicts)}，但重大事件可能改变趋势]"
+                else:
+                    # 没有重大事件，强制改为HOLD
+                    logger.warning(f"BUY decision conflicts with indicators and no major event: {', '.join(conflicts)}. Forcing to HOLD")
+                    analysis["decision"] = "HOLD"
+                    analysis["confidence"] = max(confidence - 20, 40)
+                    original_summary = analysis.get("summary", "")
+                    analysis["summary"] = f"{original_summary} [注意：技术指标显示{', '.join(conflicts)}，建议观望]"
+        
+        # 检查SELL决策是否与技术指标矛盾（放宽限制，因为SELL是有效的做空机会）
+        elif decision == "SELL":
+            conflicts = []
+            
+            # 只有在强烈看涨信号时才阻止SELL（放宽条件）
+            # RSI < 30 且 MACD看涨 且 均线向上时，才认为矛盾
+            if rsi_value < 30 and macd_signal == "bullish" and "uptrend" in ma_trend.lower():
+                conflicts.append(f"Strong bullish signals (RSI {rsi_value:.1f} < 30, MACD bullish, uptrend)")
+            # 或者 RSI < 30 且 均线强烈向上
+            elif rsi_value < 30 and "strong_uptrend" in ma_trend.lower():
+                conflicts.append(f"Very strong uptrend with oversold RSI {rsi_value:.1f}")
+            
+            if conflicts:
+                if allow_override:
+                    # 允许覆盖，但降低置信度并添加说明
+                    logger.info(f"SELL decision conflicts with strong bullish indicators but major news/macro event allows override: {', '.join(conflicts)}")
+                    analysis["confidence"] = max(confidence - 15, 50)
+                    original_summary = analysis.get("summary", "")
+                    analysis["summary"] = f"{original_summary} [注意：技术指标显示{', '.join(conflicts)}，但重大事件可能改变趋势]"
+                else:
+                    # 只有在非常强烈的看涨信号时才改为HOLD
+                    logger.warning(f"SELL decision conflicts with very strong bullish indicators: {', '.join(conflicts)}. Forcing to HOLD")
+                    analysis["decision"] = "HOLD"
+                    analysis["confidence"] = max(confidence - 20, 40)
+                    original_summary = analysis.get("summary", "")
+                    analysis["summary"] = f"{original_summary} [注意：技术指标显示{', '.join(conflicts)}，建议观望]"
+        
+        return analysis
+    
+    def _calculate_objective_score(self, data: Dict[str, Any], current_price: float) -> Dict[str, float]:
+        """
+        基于客观数据计算量化评分系统
+        
+        返回一个-100到+100的分数：
+        - +100: 强烈利多（强烈BUY）
+        - +70到+100: 强烈利多（强烈BUY）
+        - +40到+70: 利多（BUY）
+        - -40到+40: 中性（HOLD）
+        - -70到-40: 利空（SELL）
+        - -100到-70: 强烈利空（强烈SELL）
+        - -100: 强烈利空（强烈SELL）
+        """
+        indicators = data.get("indicators") or {}
+        fundamental = data.get("fundamental") or {}
+        news = data.get("news") or []
+        macro = data.get("macro") or {}
+        price_data = data.get("price") or {}
+        
+        # 1. 技术指标评分 (-100 to +100)
+        technical_score = self._calculate_technical_score(indicators, price_data)
+        
+        # 2. 基本面评分 (-100 to +100)
+        fundamental_score = self._calculate_fundamental_score(fundamental, data.get("market", ""))
+        
+        # 3. 新闻情绪评分 (-100 to +100)
+        sentiment_score = self._calculate_sentiment_score(news)
+        
+        # 4. 宏观环境评分 (-100 to +100)
+        macro_score = self._calculate_macro_score(macro, data.get("market", ""))
+        
+        # 5. 综合评分（加权平均）
+        # 权重：技术40%，基本面25%，情绪20%，宏观15%
+        overall_score = (
+            technical_score * 0.40 +
+            fundamental_score * 0.25 +
+            sentiment_score * 0.20 +
+            macro_score * 0.15
+        )
+        
+        return {
+            "technical_score": technical_score,
+            "fundamental_score": fundamental_score,
+            "sentiment_score": sentiment_score,
+            "macro_score": macro_score,
+            "overall_score": overall_score
+        }
+    
+    def _calculate_technical_score(self, indicators: Dict, price_data: Dict) -> float:
+        """计算技术指标评分 (-100 to +100)"""
+        score = 0.0
+        weight_sum = 0.0
+        
+        # RSI 评分 (-50 to +50)
+        rsi_data = indicators.get("rsi", {})
+        rsi_value = rsi_data.get("value", 50)
+        if rsi_value > 0:
+            if rsi_value > 70:
+                rsi_score = -50  # 超买，强烈利空
+            elif rsi_value > 60:
+                rsi_score = -30  # 偏超买，利空
+            elif rsi_value < 30:
+                rsi_score = +50  # 超卖，强烈利多
+            elif rsi_value < 40:
+                rsi_score = +30  # 偏超卖，利多
+            else:
+                rsi_score = (50 - rsi_value) * 0.6  # 40-60之间，线性映射
+            score += rsi_score * 0.30
+            weight_sum += 0.30
+        
+        # MACD 评分 (-40 to +40)
+        macd_data = indicators.get("macd", {})
+        macd_signal = macd_data.get("signal", "neutral")
+        if macd_signal == "bullish":
+            macd_score = +40
+        elif macd_signal == "bearish":
+            macd_score = -40
+        else:
+            macd_score = 0
+        score += macd_score * 0.25
+        weight_sum += 0.25
+        
+        # 均线趋势评分 (-40 to +40)
+        ma_data = indicators.get("moving_averages", {})
+        ma_trend = ma_data.get("trend", "sideways")
+        if "strong_uptrend" in ma_trend.lower():
+            ma_score = +40
+        elif "uptrend" in ma_trend.lower():
+            ma_score = +25
+        elif "strong_downtrend" in ma_trend.lower():
+            ma_score = -40
+        elif "downtrend" in ma_trend.lower():
+            ma_score = -25
+        else:
+            ma_score = 0
+        score += ma_score * 0.25
+        weight_sum += 0.25
+        
+        # 24小时涨跌幅评分 (-20 to +20)
+        change_24h = price_data.get("changePercent", 0)
+        if change_24h > 10:
+            change_score = -20  # 过度上涨，利空
+        elif change_24h > 5:
+            change_score = -10
+        elif change_24h < -10:
+            change_score = +20  # 过度下跌，利多
+        elif change_24h < -5:
+            change_score = +10
+        else:
+            change_score = change_24h * 2  # 线性映射
+        score += change_score * 0.20
+        weight_sum += 0.20
+        
+        # 归一化到-100到+100
+        if weight_sum > 0:
+            score = score / weight_sum * 100
+        
+        return max(-100, min(100, score))
+    
+    def _calculate_fundamental_score(self, fundamental: Dict, market: str) -> float:
+        """计算基本面评分 (-100 to +100)"""
+        if market != "USStock" or not fundamental:
+            return 0.0  # 非美股或无基本面数据，返回中性
+        
+        score = 0.0
+        factors = 0
+        
+        # PE Ratio 评分
+        pe_ratio = fundamental.get("pe_ratio")
+        if pe_ratio and pe_ratio > 0:
+            if pe_ratio < 15:
+                pe_score = +20  # 低PE，利多
+            elif pe_ratio < 25:
+                pe_score = +10
+            elif pe_ratio > 50:
+                pe_score = -20  # 高PE，利空
+            elif pe_ratio > 35:
+                pe_score = -10
+            else:
+                pe_score = 0
+            score += pe_score
+            factors += 1
+        
+        # ROE 评分
+        roe = fundamental.get("roe")
+        if roe:
+            if roe > 20:
+                roe_score = +20  # 高ROE，利多
+            elif roe > 15:
+                roe_score = +10
+            elif roe < 5:
+                roe_score = -20  # 低ROE，利空
+            elif roe < 10:
+                roe_score = -10
+            else:
+                roe_score = 0
+            score += roe_score
+            factors += 1
+        
+        # 营收增长评分
+        revenue_growth = fundamental.get("revenue_growth")
+        if revenue_growth:
+            if revenue_growth > 20:
+                growth_score = +20  # 高增长，利多
+            elif revenue_growth > 10:
+                growth_score = +10
+            elif revenue_growth < -10:
+                growth_score = -20  # 负增长，利空
+            elif revenue_growth < 0:
+                growth_score = -10
+            else:
+                growth_score = 0
+            score += growth_score
+            factors += 1
+        
+        # 利润率评分
+        profit_margin = fundamental.get("profit_margin")
+        if profit_margin:
+            if profit_margin > 20:
+                margin_score = +15  # 高利润率，利多
+            elif profit_margin > 10:
+                margin_score = +7
+            elif profit_margin < 0:
+                margin_score = -15  # 亏损，利空
+            elif profit_margin < 5:
+                margin_score = -7
+            else:
+                margin_score = 0
+            score += margin_score
+            factors += 1
+        
+        # 债务权益比评分
+        debt_to_equity = fundamental.get("debt_to_equity")
+        if debt_to_equity:
+            if debt_to_equity < 0.5:
+                debt_score = +10  # 低负债，利多
+            elif debt_to_equity > 2.0:
+                debt_score = -10  # 高负债，利空
+            else:
+                debt_score = 0
+            score += debt_score
+            factors += 1
+        
+        # 归一化（如果有多个因素）
+        if factors > 0:
+            score = score / factors * 100 / 4  # 最大可能分数是4个因素各20分=80，归一化到100
+        
+        return max(-100, min(100, score))
+    
+    def _calculate_sentiment_score(self, news: List[Dict]) -> float:
+        """计算新闻情绪评分 (-100 to +100)"""
+        if not news:
+            return 0.0  # 无新闻，中性
+        
+        positive_count = 0
+        negative_count = 0
+        neutral_count = 0
+        
+        for item in news[:10]:  # 只看前10条
+            sentiment = item.get("sentiment", "neutral")
+            if sentiment == "positive":
+                positive_count += 1
+            elif sentiment == "negative":
+                negative_count += 1
+            else:
+                neutral_count += 1
+        
+        total = positive_count + negative_count + neutral_count
+        if total == 0:
+            return 0.0
+        
+        # 计算净情绪
+        net_sentiment = (positive_count - negative_count) / total
+        
+        # 映射到-100到+100
+        score = net_sentiment * 100
+        
+        return max(-100, min(100, score))
+    
+    def _calculate_macro_score(self, macro: Dict, market: str) -> float:
+        """计算宏观环境评分 (-100 to +100)"""
+        if not macro:
+            return 0.0  # 无宏观数据，中性
+        
+        score = 0.0
+        factors = 0
+        
+        # VIX 评分（恐慌指数）
+        vix = macro.get("VIX", {})
+        vix_value = vix.get("price", 0)
+        if vix_value > 0:
+            if vix_value > 30:
+                vix_score = -30  # 高恐慌，利空
+            elif vix_value > 20:
+                vix_score = -15
+            elif vix_value < 15:
+                vix_score = +15  # 低恐慌，利多
+            else:
+                vix_score = 0
+            score += vix_score
+            factors += 1
+        
+        # DXY 评分（美元指数）
+        dxy = macro.get("DXY", {})
+        dxy_value = dxy.get("price", 0)
+        dxy_change = dxy.get("changePercent", 0)
+        if dxy_value > 0:
+            # 对于加密货币和商品，强美元通常是利空
+            if market in ["Crypto", "Forex", "Futures"]:
+                if dxy_change > 1:
+                    dxy_score = -20  # 美元走强，利空
+                elif dxy_change < -1:
+                    dxy_score = +20  # 美元走弱，利多
+                else:
+                    dxy_score = 0
+            else:
+                dxy_score = 0  # 对股票影响较小
+            score += dxy_score
+            factors += 1
+        
+        # 利率评分（TNX）
+        tnx = macro.get("TNX", {})
+        tnx_change = tnx.get("changePercent", 0)
+        if tnx_change != 0:
+            # 利率上升对成长股和加密货币通常是利空
+            if market in ["Crypto", "USStock"]:
+                if tnx_change > 2:
+                    tnx_score = -20  # 利率大幅上升，利空
+                elif tnx_change < -2:
+                    tnx_score = +20  # 利率下降，利多
+                else:
+                    tnx_score = 0
+            else:
+                tnx_score = 0
+            score += tnx_score
+            factors += 1
+        
+        # 归一化
+        if factors > 0:
+            score = score / factors * 100 / 3  # 最大可能分数是3个因素各30分=90，归一化到100
+        
+        return max(-100, min(100, score))
+    
+    def _score_to_decision(self, score: float) -> str:
+        """
+        根据客观评分转换为决策
+        
+        优化后的阈值（缩小HOLD区间，使决策更明确）：
+        - score >= +40: BUY（利多）
+        - score <= -40: SELL（利空）
+        - -40 < score < +40: HOLD（中性）
+        
+        分级决策（可选，用于更细粒度的判断）：
+        - score >= +70: 强烈BUY
+        - +40 <= score < +70: BUY
+        - +10 < score < +40: 弱利多（倾向于BUY，但可HOLD）
+        - -10 <= score <= +10: 中性HOLD
+        - -40 < score < -10: 弱利空（倾向于SELL，但可HOLD）
+        - -70 < score <= -40: SELL
+        - score <= -70: 强烈SELL
+        """
+        # 使用±40作为主要阈值，缩小HOLD区间
+        if score >= 40:
+            return "BUY"
+        elif score <= -40:
+            return "SELL"
+        else:
+            return "HOLD"
+    
    def _calculate_overall_score(self, analysis: Dict) -> int:
-        """Calculate weighted overall score."""
+        """Calculate weighted overall score (legacy method, now uses objective score if available)."""
+        # 优先使用客观评分
+        if "objective_score" in analysis:
+            objective = analysis["objective_score"]
+            overall = objective.get("overall_score", 50)
+            # 转换为0-100格式（原系统使用）
+            return max(0, min(100, int(50 + overall * 0.5)))
+        
+        # 降级到LLM评分
        tech = analysis.get("technical_score", 50)
        fund = analysis.get("fundamental_score", 50)
        sent = analysis.get("sentiment_score", 50)
@@ -36,6 +36,50 @@ IBKRClient = None
 MT5Client = None


+def _normalize_symbol_for_order(symbol: str, market_type: str = "swap") -> str:
+    """
+    规范化符号格式，确保符号符合交易所要求。
+    
+    处理各种输入格式：
+    - BTC/USDT -> BTC/USDT
+    - BTCUSDT -> BTC/USDT
+    - BTC/USDT:USDT -> BTC/USDT
+    - PI, TRX -> PI/USDT, TRX/USDT (默认添加 /USDT)
+    
+    Args:
+        symbol: 原始符号
+        market_type: 市场类型 (spot/swap)
+        
+    Returns:
+        规范化后的符号
+    """
+    if not symbol:
+        return symbol
+    
+    sym = symbol.strip()
+    
+    # 移除 swap/futures 后缀
+    if ':' in sym:
+        sym = sym.split(':', 1)[0]
+    
+    sym = sym.upper()
+    
+    # 如果已经有分隔符，直接返回（假设格式正确）
+    if '/' in sym:
+        return sym
+    
+    # 尝试从常见报价货币中识别
+    common_quotes = ['USDT', 'USD', 'BTC', 'ETH', 'BUSD', 'USDC']
+    for quote in common_quotes:
+        if sym.endswith(quote) and len(sym) > len(quote):
+            base = sym[:-len(quote)]
+            if base:
+                return f"{base}/{quote}"
+    
+    # 如果无法识别，默认使用 USDT
+    return f"{sym}/USDT"
+
+
 def _signal_to_sides(signal_type: str) -> Tuple[str, str, bool]:
    """
    Returns (side, pos_side, reduce_only)
@@ -80,6 +124,9 @@ def place_order_from_signal(
    # Spot does not support short signals in this system.
    if mt == "spot" and ("short" in (signal_type or "").lower()):
        raise LiveTradingError("spot market does not support short signals")
+    
+    # 规范化符号格式（统一处理裸符号如 PI, TRX 等）
+    symbol = _normalize_symbol_for_order(symbol, market_type=mt)

    if isinstance(client, BinanceFuturesClient):
        return client.place_market_order(
@@ -14,12 +14,28 @@ from typing import Dict, Tuple


 def _split_base_quote(symbol: str) -> Tuple[str, str]:
+    """
+    分割符号为基础货币和报价货币。
+    
+    处理各种格式：
+    - BTC/USDT -> (BTC, USDT)
+    - BTCUSDT -> (BTCUSDT, "") - 需要进一步处理
+    - PI, TRX -> (PI, "") - 需要进一步处理
+    """
    s = (symbol or "").strip()
    if ":" in s:
        s = s.split(":", 1)[0]
    if "/" not in s:
-        # Already exchange-specific (best-effort)
-        return s, ""
+        # 尝试识别报价货币（常见格式：BASEQUOTE）
+        s_upper = s.upper()
+        common_quotes = ['USDT', 'USD', 'BTC', 'ETH', 'BUSD', 'USDC', 'BNB']
+        for quote in common_quotes:
+            if s_upper.endswith(quote) and len(s_upper) > len(quote):
+                base = s_upper[:-len(quote)]
+                if base:
+                    return base, quote
+        # 无法识别，返回原符号和空报价
+        return s_upper, ""
    base, quote = s.split("/", 1)
    return base.strip().upper(), quote.strip().upper()

@@ -313,7 +313,7 @@ class LLMService:

    def call_llm_api(self, messages: list, model: str = None, temperature: float = 0.7, 
                     use_fallback: bool = True, provider: LLMProvider = None,
-                     use_json_mode: bool = True) -> str:
+                     use_json_mode: bool = True, try_alternative_providers: bool = True) -> str:
        """
        Call LLM API with the specified or default provider.
        
@@ -324,6 +324,7 @@ class LLMService:
            use_fallback: Whether to try fallback model on failure
            provider: Override the service's default provider
            use_json_mode: Whether to request JSON output format (default True for analysis, False for code generation)
+            try_alternative_providers: Whether to try alternative providers when current provider fails with 403/402
        
        Returns:
            Generated text content
@@ -347,11 +348,22 @@ class LLMService:
        api_key = self.get_api_key(p)
        
        if not api_key:
-            raise ValueError(f"API key not configured for provider: {p.value}")
+            # If no API key for current provider, try to find any available provider
+            if try_alternative_providers:
+                for alt_provider in [LLMProvider.DEEPSEEK, LLMProvider.GROK, LLMProvider.OPENAI, LLMProvider.GOOGLE, LLMProvider.OPENROUTER]:
+                    if alt_provider != p and self.get_api_key(alt_provider):
+                        logger.warning(f"No API key for {p.value}, switching to {alt_provider.value}")
+                        p = alt_provider
+                        api_key = self.get_api_key(p)
+                        break
+            
+            if not api_key:
+                raise ValueError(f"API key not configured for provider: {p.value}. Please configure at least one LLM provider API key.")
        
        base_url = self.get_base_url(p)
        
        # Normalize model name for the provider
+        original_model = model
        model = self._normalize_model_for_provider(model, p)
        
        config = load_addon_config()
@@ -366,6 +378,7 @@ class LLMService:
                models_to_try.append(fallback)
        
        last_error = None
+        last_status_code = None
        
        for current_model in models_to_try:
            try:
@@ -384,13 +397,25 @@ class LLMService:
                    
            except requests.exceptions.HTTPError as e:
                error_detail = e.response.text if e.response else str(e)
-                logger.error(f"{p.value} API HTTP error ({current_model}): {error_detail}")
+                status_code = e.response.status_code if e.response else None
+                last_status_code = status_code
+                
+                logger.error(f"{p.value} API HTTP error ({current_model}): {status_code} - {error_detail}")
                last_error = str(e)
                
+                # 403/402 errors usually mean API key issue - try alternative provider
+                if status_code in (402, 403) and try_alternative_providers and current_model == models_to_try[-1]:
+                    # Only try alternative providers after all models in current provider failed
+                    logger.warning(f"{p.value} returned {status_code} (likely API key issue). Trying alternative providers...")
+                    return self._try_alternative_providers(
+                        messages, original_model, temperature, 
+                        use_json_mode, excluded_provider=p
+                    )
+                
                # Check for recoverable errors - try fallback model
                # 402: Payment required, 403: Forbidden (invalid key), 404: Model not found, 429: Rate limit
-                if e.response and e.response.status_code in (402, 403, 404, 429):
-                    logger.warning(f"{p.value} returned {e.response.status_code} for model {current_model}; trying fallback...")
+                if status_code in (402, 403, 404, 429):
+                    logger.warning(f"{p.value} returned {status_code} for model {current_model}; trying fallback...")
                    continue
                
                if not use_fallback or current_model == models_to_try[-1]:
@@ -408,9 +433,50 @@ class LLMService:
                if current_model == models_to_try[-1]:
                    raise
        
-        error_msg = f"All model calls failed. Last error: {last_error}"
+        error_msg = f"All model calls failed for {p.value}. Last error: {last_error}"
+        if last_status_code in (402, 403):
+            error_msg += f"\nStatus {last_status_code} usually means: API key invalid/expired, insufficient balance, or no access to model."
+            error_msg += f"\nPlease check your {p.value} API key configuration and account balance."
+        
        logger.error(error_msg)
        raise Exception(error_msg)
+    
+    def _try_alternative_providers(self, messages: list, model: str, temperature: float,
+                                  use_json_mode: bool, excluded_provider: LLMProvider = None) -> str:
+        """
+        Try alternative providers when current provider fails.
+        
+        Priority: DeepSeek > Grok > OpenAI > Google > OpenRouter
+        """
+        priority_order = [
+            LLMProvider.DEEPSEEK,
+            LLMProvider.GROK,
+            LLMProvider.OPENAI,
+            LLMProvider.GOOGLE,
+            LLMProvider.OPENROUTER,
+        ]
+        
+        for alt_provider in priority_order:
+            if alt_provider == excluded_provider:
+                continue
+            
+            api_key = self.get_api_key(alt_provider)
+            if not api_key:
+                continue
+            
+            logger.info(f"Trying alternative provider: {alt_provider.value}")
+            try:
+                return self.call_llm_api(
+                    messages, model, temperature,
+                    use_fallback=True, provider=alt_provider,
+                    use_json_mode=use_json_mode,
+                    try_alternative_providers=False  # Prevent infinite recursion
+                )
+            except Exception as e:
+                logger.warning(f"Alternative provider {alt_provider.value} also failed: {str(e)}")
+                continue
+        
+        raise Exception(f"All LLM providers failed. Please check your API key configurations.")

    # Legacy method for backward compatibility
    def call_openrouter_api(self, messages: list, model: str = None, temperature: float = 0.7, use_fallback: bool = True) -> str:
@@ -20,6 +20,7 @@ from datetime import datetime, timedelta
 from concurrent.futures import ThreadPoolExecutor, as_completed, TimeoutError

 import yfinance as yf
+import pandas as pd

 from app.data_sources import DataSourceFactory
 from app.services.kline import KlineService
@@ -552,10 +553,13 @@ class MarketDataCollector:
        return None
    
    def _get_us_fundamental(self, symbol: str) -> Optional[Dict[str, Any]]:
-        """美股基本面 - Finnhub + yfinance"""
+        """
+        美股基本面 - Finnhub + yfinance
+        包括：基础财务指标 + 财报数据（资产负债表、利润表、现金流量表）
+        """
        result = {}
        
-        # Finnhub
+        # === 1. 基础财务指标 (Finnhub) ===
        if self._finnhub_client:
            try:
                metrics = self._finnhub_client.company_basic_financials(symbol, 'all')
@@ -573,31 +577,193 @@ class MarketDataCollector:
                        'roe': m.get('roeTTM'),
                        'eps': m.get('epsBasicExclExtraItemsTTM'),
                        'revenue_growth': m.get('revenueGrowthTTMYoy'),
+                        'profit_margin': m.get('netProfitMarginTTM'),
+                        'debt_to_equity': m.get('totalDebtToEquityQuarterly'),
+                        'current_ratio': m.get('currentRatioQuarterly'),
+                        'quick_ratio': m.get('quickRatioQuarterly'),
                    })
            except Exception as e:
                logger.debug(f"Finnhub fundamental failed for {symbol}: {e}")
        
-        # yfinance 补充
-        if not result:
-            try:
-                ticker = yf.Ticker(symbol)
-                info = ticker.info or {}
-                result.update({
-                    'pe_ratio': info.get('trailingPE') or info.get('forwardPE'),
-                    'pb_ratio': info.get('priceToBook'),
-                    'market_cap': info.get('marketCap'),
-                    'dividend_yield': info.get('dividendYield'),
-                    'beta': info.get('beta'),
-                    '52w_high': info.get('fiftyTwoWeekHigh'),
-                    '52w_low': info.get('fiftyTwoWeekLow'),
-                    'roe': info.get('returnOnEquity'),
-                    'eps': info.get('trailingEps'),
-                })
-            except Exception as e:
-                logger.debug(f"yfinance fundamental failed for {symbol}: {e}")
+        # === 2. yfinance 补充基础指标 ===
+        try:
+            ticker = yf.Ticker(symbol)
+            info = ticker.info or {}
+            
+            # 补充缺失的基础指标
+            if not result.get('pe_ratio'):
+                result['pe_ratio'] = info.get('trailingPE') or info.get('forwardPE')
+            if not result.get('pb_ratio'):
+                result['pb_ratio'] = info.get('priceToBook')
+            if not result.get('market_cap'):
+                result['market_cap'] = info.get('marketCap')
+            if not result.get('dividend_yield'):
+                result['dividend_yield'] = info.get('dividendYield')
+            if not result.get('beta'):
+                result['beta'] = info.get('beta')
+            if not result.get('52w_high'):
+                result['52w_high'] = info.get('fiftyTwoWeekHigh')
+            if not result.get('52w_low'):
+                result['52w_low'] = info.get('fiftyTwoWeekLow')
+            if not result.get('roe'):
+                result['roe'] = info.get('returnOnEquity')
+            if not result.get('eps'):
+                result['eps'] = info.get('trailingEps')
+            
+            # 补充更多财务指标
+            result.update({
+                'revenue': info.get('totalRevenue'),
+                'gross_profit': info.get('grossProfits'),
+                'operating_margin': info.get('operatingMargins'),
+                'profit_margin': result.get('profit_margin') or info.get('profitMargins'),
+                'ebitda': info.get('ebitda'),
+                'debt': info.get('totalDebt'),
+                'cash': info.get('totalCash'),
+                'free_cash_flow': info.get('freeCashflow'),
+                'operating_cash_flow': info.get('operatingCashflow'),
+                'book_value': info.get('bookValue'),
+                'enterprise_value': info.get('enterpriseValue'),
+            })
+        except Exception as e:
+            logger.debug(f"yfinance fundamental failed for {symbol}: {e}")
+        
+        # === 3. 获取财报数据（资产负债表、利润表、现金流量表）===
+        financial_statements = self._get_financial_statements(symbol)
+        if financial_statements:
+            result['financial_statements'] = financial_statements
+        
+        # === 4. 获取盈利报告（Earnings）===
+        earnings_data = self._get_earnings_data(symbol)
+        if earnings_data:
+            result['earnings'] = earnings_data
        
        return result if result else None
    
+    def _get_financial_statements(self, symbol: str) -> Optional[Dict[str, Any]]:
+        """
+        获取财务报表数据（资产负债表、利润表、现金流量表）
+        
+        使用 yfinance 获取，包含最近几个季度的数据
+        """
+        try:
+            ticker = yf.Ticker(symbol)
+            statements = {}
+            
+            # 资产负债表 (Balance Sheet)
+            try:
+                balance_sheet = ticker.balance_sheet
+                if balance_sheet is not None and not balance_sheet.empty:
+                    # 获取最近4个季度
+                    latest_quarters = balance_sheet.columns[:4] if len(balance_sheet.columns) >= 4 else balance_sheet.columns
+                    statements['balance_sheet'] = {
+                        'latest_date': str(latest_quarters[0]) if len(latest_quarters) > 0 else None,
+                        'total_assets': float(balance_sheet.loc['Total Assets', latest_quarters[0]]) if 'Total Assets' in balance_sheet.index and len(latest_quarters) > 0 else None,
+                        'total_liabilities': float(balance_sheet.loc['Total Liab', latest_quarters[0]]) if 'Total Liab' in balance_sheet.index and len(latest_quarters) > 0 else None,
+                        'total_equity': float(balance_sheet.loc['Stockholders Equity', latest_quarters[0]]) if 'Stockholders Equity' in balance_sheet.index and len(latest_quarters) > 0 else None,
+                        'cash': float(balance_sheet.loc['Cash', latest_quarters[0]]) if 'Cash' in balance_sheet.index and len(latest_quarters) > 0 else None,
+                        'debt': float(balance_sheet.loc['Total Debt', latest_quarters[0]]) if 'Total Debt' in balance_sheet.index and len(latest_quarters) > 0 else None,
+                        'current_assets': float(balance_sheet.loc['Current Assets', latest_quarters[0]]) if 'Current Assets' in balance_sheet.index and len(latest_quarters) > 0 else None,
+                        'current_liabilities': float(balance_sheet.loc['Current Liabilities', latest_quarters[0]]) if 'Current Liabilities' in balance_sheet.index and len(latest_quarters) > 0 else None,
+                    }
+            except Exception as e:
+                logger.debug(f"Balance sheet fetch failed for {symbol}: {e}")
+            
+            # 利润表 (Income Statement)
+            try:
+                income_stmt = ticker.financials
+                if income_stmt is not None and not income_stmt.empty:
+                    latest_quarters = income_stmt.columns[:4] if len(income_stmt.columns) >= 4 else income_stmt.columns
+                    statements['income_statement'] = {
+                        'latest_date': str(latest_quarters[0]) if len(latest_quarters) > 0 else None,
+                        'total_revenue': float(income_stmt.loc['Total Revenue', latest_quarters[0]]) if 'Total Revenue' in income_stmt.index and len(latest_quarters) > 0 else None,
+                        'gross_profit': float(income_stmt.loc['Gross Profit', latest_quarters[0]]) if 'Gross Profit' in income_stmt.index and len(latest_quarters) > 0 else None,
+                        'operating_income': float(income_stmt.loc['Operating Income', latest_quarters[0]]) if 'Operating Income' in income_stmt.index and len(latest_quarters) > 0 else None,
+                        'net_income': float(income_stmt.loc['Net Income', latest_quarters[0]]) if 'Net Income' in income_stmt.index and len(latest_quarters) > 0 else None,
+                        'eps': float(income_stmt.loc['Basic EPS', latest_quarters[0]]) if 'Basic EPS' in income_stmt.index and len(latest_quarters) > 0 else None,
+                    }
+            except Exception as e:
+                logger.debug(f"Income statement fetch failed for {symbol}: {e}")
+            
+            # 现金流量表 (Cash Flow Statement)
+            try:
+                cashflow = ticker.cashflow
+                if cashflow is not None and not cashflow.empty:
+                    latest_quarters = cashflow.columns[:4] if len(cashflow.columns) >= 4 else cashflow.columns
+                    statements['cash_flow'] = {
+                        'latest_date': str(latest_quarters[0]) if len(latest_quarters) > 0 else None,
+                        'operating_cash_flow': float(cashflow.loc['Operating Cash Flow', latest_quarters[0]]) if 'Operating Cash Flow' in cashflow.index and len(latest_quarters) > 0 else None,
+                        'investing_cash_flow': float(cashflow.loc['Capital Expenditure', latest_quarters[0]]) if 'Capital Expenditure' in cashflow.index and len(latest_quarters) > 0 else None,
+                        'financing_cash_flow': float(cashflow.loc['Financing Cash Flow', latest_quarters[0]]) if 'Financing Cash Flow' in cashflow.index and len(latest_quarters) > 0 else None,
+                        'free_cash_flow': float(cashflow.loc['Free Cash Flow', latest_quarters[0]]) if 'Free Cash Flow' in cashflow.index and len(latest_quarters) > 0 else None,
+                    }
+            except Exception as e:
+                logger.debug(f"Cash flow statement fetch failed for {symbol}: {e}")
+            
+            return statements if statements else None
+            
+        except Exception as e:
+            logger.debug(f"Financial statements fetch failed for {symbol}: {e}")
+            return None
+    
+    def _get_earnings_data(self, symbol: str) -> Optional[Dict[str, Any]]:
+        """
+        获取盈利报告数据（Earnings）
+        
+        包括：历史盈利、盈利预测、盈利日期等
+        """
+        try:
+            ticker = yf.Ticker(symbol)
+            earnings_data = {}
+            
+            # 历史盈利数据
+            try:
+                earnings_history = ticker.earnings_history
+                if earnings_history is not None and not earnings_history.empty:
+                    # 获取最近4个季度
+                    recent_earnings = earnings_history.head(4)
+                    earnings_data['history'] = []
+                    for _, row in recent_earnings.iterrows():
+                        earnings_data['history'].append({
+                            'date': str(row.get('Date', '')),
+                            'eps_actual': float(row.get('EPS Actual', 0)) if row.get('EPS Actual') is not None else None,
+                            'eps_estimate': float(row.get('EPS Estimate', 0)) if row.get('EPS Estimate') is not None else None,
+                            'surprise': float(row.get('Surprise(%)', 0)) if row.get('Surprise(%)') is not None else None,
+                        })
+            except Exception as e:
+                logger.debug(f"Earnings history fetch failed for {symbol}: {e}")
+            
+            # 盈利日历（未来盈利日期）
+            try:
+                earnings_calendar = ticker.calendar
+                if earnings_calendar is not None and not earnings_calendar.empty:
+                    earnings_data['upcoming'] = {
+                        'next_earnings_date': str(earnings_calendar.index[0]) if len(earnings_calendar.index) > 0 else None,
+                        'eps_estimate': float(earnings_calendar.loc[earnings_calendar.index[0], 'Earnings Estimate']) if len(earnings_calendar.index) > 0 and 'Earnings Estimate' in earnings_calendar.columns else None,
+                        'revenue_estimate': float(earnings_calendar.loc[earnings_calendar.index[0], 'Revenue Estimate']) if len(earnings_calendar.index) > 0 and 'Revenue Estimate' in earnings_calendar.columns else None,
+                    }
+            except Exception as e:
+                logger.debug(f"Earnings calendar fetch failed for {symbol}: {e}")
+            
+            # 季度盈利数据
+            try:
+                quarterly_earnings = ticker.quarterly_earnings
+                if quarterly_earnings is not None and not quarterly_earnings.empty:
+                    latest_q = quarterly_earnings.index[0] if len(quarterly_earnings.index) > 0 else None
+                    if latest_q:
+                        earnings_data['quarterly'] = {
+                            'latest_quarter': str(latest_q),
+                            'revenue': float(quarterly_earnings.loc[latest_q, 'Revenue']) if 'Revenue' in quarterly_earnings.columns else None,
+                            'earnings': float(quarterly_earnings.loc[latest_q, 'Earnings']) if 'Earnings' in quarterly_earnings.columns else None,
+                        }
+            except Exception as e:
+                logger.debug(f"Quarterly earnings fetch failed for {symbol}: {e}")
+            
+            return earnings_data if earnings_data else None
+            
+        except Exception as e:
+            logger.debug(f"Earnings data fetch failed for {symbol}: {e}")
+            return None
+    
    def _get_crypto_info(self, symbol: str) -> Optional[Dict[str, Any]]:
        """加密货币信息 (固定描述为主)"""
        # 常见加密货币的描述
@@ -295,6 +295,32 @@ class PendingOrderWorker:
                            except Exception:
                                pass
                            exch_size.setdefault(hb_sym, {"long": 0.0, "short": 0.0})[side] = float(qty_base)
+                            
+                            # Extract entry price from OKX position data
+                            # OKX API returns avgPx (average price) or avgPxEp (average price in equity) for positions
+                            try:
+                                # Try avgPx first (average entry price)
+                                avg_px = p.get("avgPx")
+                                if avg_px:
+                                    entry_price = float(avg_px)
+                                else:
+                                    # Fallback to avgPxEp (average price in equity)
+                                    avg_px_ep = p.get("avgPxEp")
+                                    if avg_px_ep:
+                                        entry_price = float(avg_px_ep)
+                                    else:
+                                        # Fallback to last price if available
+                                        last_px = p.get("last")
+                                        entry_price = float(last_px) if last_px else 0.0
+                                
+                                if entry_price > 0:
+                                    exch_entry_price.setdefault(hb_sym, {"long": 0.0, "short": 0.0})[side] = entry_price
+                                    logger.debug(f"[PositionSync] OKX {hb_sym} {side}: entry_price={entry_price} from avgPx={p.get('avgPx')} or avgPxEp={p.get('avgPxEp')}")
+                                else:
+                                    logger.warning(f"[PositionSync] OKX {hb_sym} {side}: Could not extract entry price from position data: {p}")
+                            except Exception as e:
+                                logger.warning(f"[PositionSync] Failed to extract entry price for OKX {hb_sym} {side}: {e}")
+                                # Don't set entry_price, will remain 0.0

                elif isinstance(client, BitgetMixClient) and market_type == "swap":
                    product_type = str(exchange_config.get("product_type") or exchange_config.get("productType") or "USDT-FUTURES")