diff --git a/auto_trade_polymarket.py b/auto_trade_polymarket.py
index ee90f57..a5ca99b 100644
--- a/auto_trade_polymarket.py
+++ b/auto_trade_polymarket.py
@@ -35,6 +35,7 @@ sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
 
 from data.fetcher import BinanceDataFetcher
 from data.trading_db import TradingDatabase
+from api.models import PredictionModel
 
 load_dotenv()
 
@@ -634,7 +635,11 @@ class PolymarketBTCTrader:
         logger.info(f"  上一K线时间: {prev_time} (用于预测)")
         logger.info(f"  上一收盘价: ${prev_close:,.2f}")
         logger.info(f"  当前K线时间: {current_time} (预测目标)")
-        logger.info(f"  当前收盘价: ${current_close:,.2f} (实际结果)")
+        logger.info(f"  当前收盘价: ${current_close:,.2f} (未收盘)")
+
+        # 回填上一小时的实际结果
+        # prev_time的K线已收盘，用它的收盘价(prev_close) vs 再前一根的收盘价来判断涨跌
+        self._update_previous_prediction(df)
         
         # 用除最后一根外的数据进行预测（无create_labels，与test逻辑一致）
         df_for_prediction = df.iloc[:-1]
@@ -659,9 +664,72 @@ class PolymarketBTCTrader:
         confidence = probability if prediction == 1 else 1 - probability
         logger.info(f"🤖 DLinear预测: {direction}")
         logger.info(f"   置信度: {confidence:.2%}")
-        
+
+        # 保存预测结果到 model_predictions 表
+        # 当前K线尚未收盘，actual_result 和 is_correct 留空，等收盘后更新
+        try:
+            et_tz = ZoneInfo('America/New_York')
+            # current_time 是目标K线的UTC时间戳（当前正在进行的K线）
+            current_time_utc = pd.Timestamp(current_time)
+            current_time_et = current_time_utc.tz_localize('UTC').tz_convert(et_tz)
+
+            prediction_data = {
+                "timestamp": current_time_utc.isoformat(),
+                "timestamp_et": current_time_et.isoformat(),
+                "date": current_time_et.strftime('%Y-%m-%d'),
+                "hour": current_time_et.hour,
+                "model_name": "DLinear",
+                "prediction": prediction,
+                "probability": probability,
+                "actual_result": None,
+                "is_correct": None,
+                "open_price": float(df['open'].iloc[-1]),
+                "close_price": None,
+            }
+            PredictionModel.save_prediction(prediction_data)
+            logger.info(f"💾 预测记录已保存 (ET {current_time_et.strftime('%Y-%m-%d %H:%M')})")
+        except Exception as e:
+            logger.error(f"保存预测记录失败: {e}")
+
         return prediction, probability
-    
+
+    def _update_previous_prediction(self, df: pd.DataFrame):
+        """回填上一小时预测的实际结果
+
+        df.iloc[-2] 是上一小时已收盘K线，收盘价 vs 开盘价即为实际涨跌。
+        """
+        if len(df) < 2:
+            return
+
+        try:
+            et_tz = ZoneInfo('America/New_York')
+            prev_row = df.iloc[-2]
+            prev_utc = pd.Timestamp(prev_row['timestamp'])
+            prev_et = prev_utc.tz_localize('UTC').tz_convert(et_tz)
+            prev_date = prev_et.strftime('%Y-%m-%d')
+            prev_hour = prev_et.hour
+
+            prev_open = float(prev_row['open'])
+            prev_close = float(prev_row['close'])
+            actual_result = 1 if prev_close > prev_open else 0
+
+            updated = PredictionModel.update_actual_result(
+                date=prev_date,
+                hour=prev_hour,
+                model_name="DLinear",
+                actual_result=actual_result,
+                close_price=prev_close
+            )
+            if updated:
+                record = PredictionModel.get_prediction(prev_date, prev_hour, "DLinear")
+                is_correct = record.get('is_correct') if record else None
+                correct_str = '✅ 正确' if is_correct == 1 else '❌ 错误' if is_correct == 0 else '未知'
+                logger.info(f"📝 回填上一小时结果: ET {prev_date} {prev_hour}:00, "
+                            f"开:{prev_open:.2f} 收:{prev_close:.2f} "
+                            f"实际{'涨' if actual_result == 1 else '跌'}, {correct_str}")
+        except Exception as e:
+            logger.error(f"回填上一小时结果失败: {e}")
+
     def run_once(self) -> Dict:
         """运行一次交易流程（不等待整点）"""
         logger.info("运行单次交易...")