diff --git a/freqtrade/data/converter.py b/freqtrade/data/converter.py
index c71f7795c..cd5719326 100644
--- a/freqtrade/data/converter.py
+++ b/freqtrade/data/converter.py
@@ -195,7 +195,18 @@ def order_book_to_dataframe(bids: list, asks: list) -> DataFrame:
     return frame
 
 
-def trades_remove_duplicates(trades: pd.DataFrame) -> pd.DataFrame:
+def trades_remove_duplicates(trades: List[List]) -> List[List]:
+    """
+    Removes duplicates from the trades list.
+    Uses itertools.groupby to avoid converting to pandas.
+    Tests show it as being pretty efficient on lists of 4M Lists.
+    :param trades: List of Lists with constants.DEFAULT_TRADES_COLUMNS as columns
+    :return: same format as above, but with duplicates removed
+    """
+    return [i for i, _ in itertools.groupby(sorted(trades, key=itemgetter(0)))]
+
+
+def trades_df_remove_duplicates(trades: pd.DataFrame) -> pd.DataFrame:
     """
     Removes duplicates from the trades DataFrame.
     Uses pandas.DataFrame.drop_duplicates to remove duplicates based on the 'timestamp' column.
diff --git a/freqtrade/data/history/idatahandler.py b/freqtrade/data/history/idatahandler.py
index 4618dfdeb..a11c79bd9 100644
--- a/freqtrade/data/history/idatahandler.py
+++ b/freqtrade/data/history/idatahandler.py
@@ -16,7 +16,8 @@ from pandas import DataFrame, to_datetime
 from freqtrade import misc
 from freqtrade.configuration import TimeRange
 from freqtrade.constants import ListPairsWithTimeframes, TradeList
-from freqtrade.data.converter import clean_ohlcv_dataframe, trades_remove_duplicates, trim_dataframe
+from freqtrade.data.converter import (clean_ohlcv_dataframe, trades_df_remove_duplicates,
+                                      trim_dataframe)
 from freqtrade.enums import CandleType, TradingMode
 from freqtrade.exchange import timeframe_to_seconds
 
@@ -216,12 +217,12 @@ class IDataHandler(ABC):
         :param timerange: Timerange to load trades for - currently not implemented
         :return: List of trades
         """
-        trades = trades_remove_duplicates(self._trades_load(pair, timerange=timerange))
+        trades = trades_df_remove_duplicates(self._trades_load(pair, timerange=timerange))
         trades['timestamp'] = to_datetime(trades['timestamp'], unit='ms', utc=True)
         return trades
 
     def trades_load_aslist(self, pair: str, timerange: Optional[TimeRange] = None) -> TradeList:
-        trades = trades_remove_duplicates(self._trades_load(pair, timerange=timerange))
+        trades = trades_df_remove_duplicates(self._trades_load(pair, timerange=timerange))
         return trades.values.tolist()
 
     @classmethod
diff --git a/tests/data/test_converter.py b/tests/data/test_converter.py
index a701e2f07..9f733f316 100644
--- a/tests/data/test_converter.py
+++ b/tests/data/test_converter.py
@@ -10,8 +10,9 @@ import pytest
 from freqtrade.configuration.timerange import TimeRange
 from freqtrade.data.converter import (convert_ohlcv_format, convert_trades_format,
                                       ohlcv_fill_up_missing_data, ohlcv_to_dataframe,
-                                      reduce_dataframe_footprint, trades_dict_to_list,
-                                      trades_remove_duplicates, trades_to_ohlcv, trim_dataframe)
+                                      reduce_dataframe_footprint, trades_df_remove_duplicates,
+                                      trades_dict_to_list, trades_remove_duplicates,
+                                      trades_to_ohlcv, trim_dataframe)
 from freqtrade.data.history import (get_timerange, load_data, load_pair_history,
                                     validate_backtest_data)
 from freqtrade.data.history.idatahandler import IDataHandler
@@ -298,11 +299,20 @@ def test_trim_dataframe(testdatadir) -> None:
     assert all(data_modify.iloc[0] == data.iloc[25])
 
 
-def test_trades_remove_duplicates(trades_history_df):
+def test_trades_remove_duplicates(trades_history):
+    trades_history1 = trades_history * 3
+    assert len(trades_history1) == len(trades_history) * 3
+    res = trades_remove_duplicates(trades_history1)
+    assert len(res) == len(trades_history)
+    for i, t in enumerate(res):
+        assert t == trades_history[i]
+
+
+def test_trades_df_remove_duplicates(trades_history_df):
     trades_history1 = pd.concat([trades_history_df, trades_history_df, trades_history_df]
                                 ).reset_index(drop=True)
     assert len(trades_history1) == len(trades_history_df) * 3
-    res = trades_remove_duplicates(trades_history1)
+    res = trades_df_remove_duplicates(trades_history1)
     assert len(res) == len(trades_history_df)
     assert res.equals(trades_history_df)