fix(visualization): 修复净值计算逻辑，与轮动策略结果对齐

问题：
- HTML报告错误地将所有品种持仓收益简单累加
- 没有考虑仓位占比权重
- 导致净值曲线和KPI指标与策略实际结果不一致

修复：
- 使用出场净值和仓位占比计算每日组合净值
- 净值 = sum(出场净值 * 仓位占比)
- 总收益 = (最终净值 - 初始净值) / 初始净值 * 100
- 月度收益使用净值变化率计算
- 最大回撤基于真实净值曲线计算
- 胜率基于每日净值涨跌计算
- 修复pandas FutureWarning警告

现在HTML报告的净值曲线、收益指标与轮动策略完全一致

visualization/report_generator/generate_report.py

@@ -54,49 +54,58 @@ class ReportGenerator:
         """计算关键指标"""
         df = trades_filtered if trades_filtered is not None else self.trades_df
         
-        # 转换持仓收益为数值（去除百分号）
-        if df['持仓收益'].dtype == 'object':
-            df = df.copy()
-            df['持仓收益_num'] = df['持仓收益'].str.rstrip('%').astype(float)
-        else:
-            df = df.copy()
-            df['持仓收益_num'] = df['持仓收益']
+        # 使用净值计算真实收益
+        # 按日期分组计算每日组合净值
+        daily_nav = df.groupby('出场日期').apply(
+            lambda x: (x['出场净值'].astype(float) * x['仓位占比'].str.rstrip('%').astype(float) / 100).sum(),
+            include_groups=False
+        ).reset_index()
+        daily_nav.columns = ['date', 'nav']
+        daily_nav = daily_nav.sort_values('date')
         
-        # 总收益
-        total_return = df['持仓收益_num'].sum()
+        # 总收益 = (最终净值 - 初始净值) / 初始净值 * 100
+        initial_nav = daily_nav['nav'].iloc[0]
+        final_nav = daily_nav['nav'].iloc[-1]
+        total_return = (final_nav - initial_nav) / initial_nav * 100
         
         # 年化收益
-        days = (df['出场日期'].max() - df['出场日期'].min()).days
+        days = (daily_nav['date'].iloc[-1] - daily_nav['date'].iloc[0]).days
         if days > 0:
             annual_return = total_return / (days / 365.0)
         else:
             annual_return = 0
         
-        # 胜率
-        win_count = (df['持仓收益_num'] > 0).sum()
-        total_count = len(df)
+        # 胜率 - 使用净值变化
+        daily_nav['nav_change'] = daily_nav['nav'].pct_change()
+        win_count = (daily_nav['nav_change'] > 0).sum()
+        total_count = len(daily_nav) - 1  # 减去第一天
         win_rate = (win_count / total_count * 100) if total_count > 0 else 0
         loss_count = total_count - win_count
         
         # 夏普比率
-        daily_returns = df.groupby('出场日期')['持仓收益_num'].sum()
-        if daily_returns.std() > 0:
-            sharpe_ratio = daily_returns.mean() / daily_returns.std() * np.sqrt(252)
+        if daily_nav['nav_change'].std() > 0:
+            sharpe_ratio = daily_nav['nav_change'].mean() / daily_nav['nav_change'].std() * np.sqrt(252)
         else:
             sharpe_ratio = 0
         
-        # 最大回撤（简化计算）
-        cum_returns = df['持仓收益_num'].cumsum()
-        running_max = cum_returns.cummax()
-        drawdown = (cum_returns - running_max)
-        max_drawdown = drawdown.min() if len(drawdown) > 0 else 0
+        # 最大回撤
+        running_max = daily_nav['nav'].cummax()
+        drawdown = (daily_nav['nav'] - running_max) / running_max * 100
+        max_drawdown = drawdown.min()
         
         # 调仓次数
         total_trades = len(df)
         
-        # 最佳品种
-        symbol_returns = df.groupby('品种代码')['持仓收益_num'].sum()
-        best_symbol = symbol_returns.idxmax() if len(symbol_returns) > 0 else 'N/A'
+        # 最佳品种 - 从 summary 获取
+        if self.summary_df is not None:
+            symbol_col = self.summary_df['累计收益']
+            if symbol_col.dtype == 'object':
+                symbol_col_num = symbol_col.str.rstrip('%').astype(float)
+            else:
+                symbol_col_num = symbol_col
+            best_symbol = self.summary_df.loc[symbol_col_num.idxmax(), '品种代码']
+        else:
+            best_symbol = 'N/A'
         
         # 平均持仓天数
         avg_holding_days = df['持仓天数'].mean() if len(df) > 0 else 0
@@ -128,21 +137,39 @@ class ReportGenerator:
         
         df_sorted = df.sort_values('出场日期')
         
-        # 净值曲线数据
-        df_sorted['累计收益'] = df_sorted['持仓收益_num'].cumsum()
-        nav_values = df_sorted['累计收益'].tolist()
-        nav_dates = df_sorted['出场日期'].dt.strftime('%Y-%m-%d').tolist()
+        # 净值曲线数据 - 使用出场净值（考虑仓位加权）
+        # 按日期分组，计算每日的加权平均净值
+        daily_nav = df_sorted.groupby('出场日期').apply(
+            lambda x: (x['出场净值'].astype(float) * x['仓位占比'].str.rstrip('%').astype(float) / 100).sum(),
+            include_groups=False
+        ).reset_index()
+        daily_nav.columns = ['date', 'nav']
+        daily_nav = daily_nav.sort_values('date')
         
-        # 月度收益数据
+        nav_values = daily_nav['nav'].round(4).tolist()
+        nav_dates = daily_nav['date'].dt.strftime('%Y-%m-%d').tolist()
+        
+        # 月度收益数据 - 使用净值变化计算
         df_copy = df_sorted.copy()
         df_copy['年月'] = df_copy['出场日期'].dt.to_period('M')
-        monthly = df_copy.groupby('年月')['持仓收益_num'].sum().reset_index()
-        monthly['年月_str'] = monthly['年月'].astype(str)
-        monthly_dates = monthly['年月_str'].tolist()
-        monthly_values = monthly['持仓收益_num'].round(2).tolist()
+        monthly_nav = df_copy.groupby('年月').apply(
+            lambda x: (x['出场净值'].astype(float) * x['仓位占比'].str.rstrip('%').astype(float) / 100).sum(),
+            include_groups=False
+        ).reset_index()
+        monthly_nav.columns = ['年月', 'nav']
+        monthly_nav = monthly_nav.sort_values('年月')
+        monthly_nav['nav_change'] = monthly_nav['nav'].pct_change() * 100
+        monthly_nav['nav_change'] = monthly_nav['nav_change'].fillna(0)
         
-        # 品种收益排行
-        symbol_returns = df.groupby('品种代码')['持仓收益_num'].sum().sort_values()
+        monthly_nav['年月_str'] = monthly_nav['年月'].astype(str)
+        monthly_dates = monthly_nav['年月_str'].tolist()
+        monthly_values = monthly_nav['nav_change'].round(2).tolist()
+        
+        # 品种收益排行 - 使用累计收益列
+        symbol_returns = self.summary_df.set_index('品种代码')['累计收益']
+        if symbol_returns.dtype == 'object':
+            symbol_returns = symbol_returns.str.rstrip('%').astype(float)
+        symbol_returns = symbol_returns.sort_values()
         symbol_names = []
         symbol_returns_list = []