diff --git a/freqtrade/optimize/backtesting.py b/freqtrade/optimize/backtesting.py
index a0f645f36..6e9840042 100644
--- a/freqtrade/optimize/backtesting.py
+++ b/freqtrade/optimize/backtesting.py
@@ -9,6 +9,7 @@ from copy import deepcopy
 from datetime import datetime, timedelta, timezone
 from typing import Any, Dict, List, Optional, Tuple
 
+from numpy import nan
 from pandas import DataFrame
 
 from freqtrade.configuration import TimeRange, validate_config_consistency
@@ -289,10 +290,12 @@ class Backtesting:
             # To avoid using data from future, we use buy/sell signals shifted
             # from the previous candle
             for col in headers[5:]:
+                tag_col = col in ('enter_tag', 'exit_tag')
                 if col in df_analyzed.columns:
-                    df_analyzed.loc[:, col] = df_analyzed.loc[:, col].shift(1)
+                    df_analyzed.loc[:, col] = df_analyzed.loc[:, col].replace(
+                        [nan], [0 if not tag_col else None]).shift(1)
                 else:
-                    df_analyzed.loc[:, col] = 0 if col not in ('enter_tag', 'exit_tag') else None
+                    df_analyzed.loc[:, col] = 0 if not tag_col else None
 
             # Update dataprovider cache
             self.dataprovider._set_cached_df(pair, self.timeframe, df_analyzed, CandleType.SPOT)
diff --git a/tests/optimize/__init__.py b/tests/optimize/__init__.py
index ce6f17f6e..e9827172f 100644
--- a/tests/optimize/__init__.py
+++ b/tests/optimize/__init__.py
@@ -57,10 +57,6 @@ def _build_backtest_dataframe(data):
     # Ensure floats are in place
     for column in ['open', 'high', 'low', 'close', 'volume']:
         frame[column] = frame[column].astype('float64')
-    if 'enter_tag' not in columns:
-        frame['enter_tag'] = None
-    if 'exit_tag' not in columns:
-        frame['exit_tag'] = None
 
     # Ensure all candles make kindof sense
     assert all(frame['low'] <= frame['close'])