Fix checkpointing

Author: MDUYN

investing_algorithm_framework/app/app.py

@@ -1,4 +1,5 @@
 import gc
+import shutil
 import inspect
 import logging
 import os
@@ -1142,7 +1143,9 @@ def filter_function(
                             market=market,
                             trading_symbol=trading_symbol,
                             use_checkpoints=use_checkpoints,
-                            backtest_storage_directory=None,
+                            backtest_storage_directory=(
+                                backtest_storage_directory
+                            )
                         )
                     )
 
@@ -1200,6 +1203,9 @@ def load_backtest_filter_fn(bt: Backtest) -> bool:
                         backtests_ordered_by_strategy.setdefault(
                             backtest.metadata["id"], []
                         ).append(backtest)
+
+                    # Remove all temp storage directories
+                    shutil.rmtree(path)
             else:
                 # Remove all strategies that are not in the final selection
                 backtests_ordered_by_strategy = {
@@ -1214,6 +1220,13 @@ def load_backtest_filter_fn(bt: Backtest) -> bool:
                     combine_backtests(backtests_ordered_by_strategy[strategy])
                 )
 
+            if backtest_storage_directory is not None:
+                # Save final combined backtests to storage directory
+                save_backtests_to_directory(
+                    backtests=backtests,
+                    directory_path=backtest_storage_directory,
+                )
+
         return backtests
 
     def run_vector_backtest(
@@ -1339,11 +1352,12 @@ def run_vector_backtest(
             backtest_date_range=backtest_date_range,
             storage_directory=backtest_storage_directory,
         ):
-            backtest = backtest_service.load_backtest_by_strategy(
-                strategy=strategy,
-                backtest_date_range=backtest_date_range,
-                storage_directory=backtest_storage_directory,
-            )
+            backtest = backtest_service\
+                .load_backtest_by_strategy_and_backtest_date_range(
+                    strategy=strategy,
+                    backtest_date_range=backtest_date_range,
+                    storage_directory=backtest_storage_directory,
+                )
         else:
             try:
                 run = backtest_service.create_vector_backtest(
@@ -1671,7 +1685,8 @@ def run_permutation_test(
             snapshot_interval=SnapshotInterval.DAILY,
             risk_free_rate=risk_free_rate,
             market=market,
-            trading_symbol=trading_symbol
+            trading_symbol=trading_symbol,
+            use_checkpoints=False
         )
         backtest_metrics = backtest.get_backtest_metrics(backtest_date_range)
 
@@ -1753,7 +1768,8 @@ def run_permutation_test(
                 risk_free_rate=risk_free_rate,
                 skip_data_sources_initialization=True,
                 market=market,
-                trading_symbol=trading_symbol
+                trading_symbol=trading_symbol,
+                use_checkpoints=False
             )
 
             # Add the results of the permuted backtest to the main backtest

investing_algorithm_framework/domain/backtesting/backtest.py

@@ -467,6 +467,16 @@ def merge(self, other: 'Backtest') -> 'Backtest':
 
         return merged
 
+    def get_metadata(self) -> Dict[str, str]:
+        """
+        Get the metadata of the backtest.
+
+        Returns:
+            Dict[str, str]: A dictionary containing the metadata
+                of the backtest.
+        """
+        return self.metadata
+
     def get_backtest_date_ranges(self):
         """
         Get the date ranges for the backtest.

investing_algorithm_framework/services/backtesting/backtest_service.py

@@ -672,28 +672,30 @@ def backtest_exists(
         backtest_directory = os.path.join(storage_directory, strategy_id)
 
         if os.path.exists(backtest_directory):
-            backtest = Backtest.load(backtest_directory)
-            backtest_date_ranges = backtest\
-                .get_most_granular_ohlcv_data_source()
+            backtest = Backtest.open(backtest_directory)
+            backtest_date_ranges = backtest.get_backtest_date_ranges()
 
             for backtest_date_range_ref in backtest_date_ranges:
 
                 if backtest_date_range_ref.start_date \
-                        == backtest_date_range_ref.start_date and \
+                        == backtest_date_range.start_date and \
                         backtest_date_range_ref.end_date \
-                        == backtest_date_range_ref.end_date:
+                        == backtest_date_range.end_date:
                     return True
 
         return False
 
-    def load_backtest_by_strategy(
+    def load_backtest_by_strategy_and_backtest_date_range(
         self,
         strategy,
         backtest_date_range: BacktestDateRange,
         storage_directory: str
     ) -> Backtest:
         """
         Load a backtest for the given strategy and backtest date range.
+        If the backtest does not exist, an exception will be raised.
+        For the given backtest, only the run and metrics corresponding
+        to the backtest date range will be returned.
 
         Args:
             strategy: The strategy to load the backtest for.
@@ -709,14 +711,9 @@ def load_backtest_by_strategy(
         backtest_directory = os.path.join(storage_directory, strategy_id)
 
         if os.path.exists(backtest_directory):
-            backtest = Backtest.load(backtest_directory)
-            run = backtest.get_run(backtest_date_range)
-            metrics = backtest.get_metrics(backtest_date_range)
-            return Backtest(
-                backtest_runs=[run],
-                backtest_summary=generate_backtest_summary_metrics(
-                    [metrics]
-                )
-            )
+            backtest = Backtest.open(backtest_directory)
+            run = backtest.get_backtest_run(backtest_date_range)
+            metadata = backtest.get_metadata()
+            return Backtest(backtest_runs=[run], metadata=metadata)
         else:
             raise OperationalException("Backtest does not exist.")