implemented a parameter for skipping errors during model training in the backtest process

[xsa-dev]

Summary

In case of any data disruptions related to a coin, currently, a hard exception occurs, interrupting the backtesting and limiting the completion of backtesting for the FreqAI strategy with other coins. To bypass this, it was suggested to use an exception skipping parameter as a warning, during which the exception is ignored during backtesting, allowing the backtesting to continue. In a separate case (with another parameter), resolving this issue may involve additional data preprocessing, which is not included in this feature's development.

Solve the issue: #9671

Quick changelog

• new parameter warn_exceptions_on_backtest_only to config_examples/config_freqai.example.json in config.freqai
• modify backtesting logic in freqtrade/freqai/freqai_interface.py

What's new?

We added an exception skipping parameter to handle data disruptions during backtesting, letting the FreqAI strategy proceed across coins even when exceptions occur, and there's potential for further data preprocessing to tackle such issues later on.

[coveralls]

coverage: 94.719% (+0.06%) from 94.656%
when pulling 62c64e2 on xsa-dev:Error-Skipping
into 8e5c574 on freqtrade:develop.

[xsa-dev]

Suggested edit:

diff --git a/freqtrade/freqai/freqai_interface.py b/freqtrade/freqai/freqai_interface.py
index 224b9dbe1..ae17c5940 100644
--- a/freqtrade/freqai/freqai_interface.py
+++ b/freqtrade/freqai/freqai_interface.py
@@ -5,7 +5,7 @@ from abc import ABC, abstractmethod
 from collections import deque
 from datetime import datetime, timezone
 from pathlib import Path
-from typing import Any, Dict, List, Literal, Optional, Tuple, Union
+from typing import Any, Dict, List, Literal, Optional, Tuple
 
 import datasieve.transforms as ds
 import numpy as np
@@ -108,7 +108,7 @@ class IFreqaiModel(ABC):
         self.data_provider: Optional[DataProvider] = None
         self.max_system_threads = max(int(psutil.cpu_count() * 2 - 2), 1)
         self.can_short = True  # overridden in start() with strategy.can_short
-        self.model: Union[Any, None] = None
+        self.model: Any = None
         if self.ft_params.get('principal_component_analysis', False) and self.continual_learning:
             self.ft_params.update({'principal_component_analysis': False})
             logger.warning('User tried to use PCA with continual learning. Deactivating PCA.')
@@ -258,7 +258,7 @@ class IFreqaiModel(ABC):
                     self.dd.save_metric_tracker_to_disk()
 
 
-    def _train_model(self, dataframe_train, pair, dk, tr_backtest) -> Union[Any, None]:
+    def _train_model(self, dataframe_train, pair, dk, tr_backtest):
         try:
             self.tb_logger = get_tb_logger(self.dd.model_type, dk.data_path,
                                             self.activate_tensorboard)

[xsa-dev]

Done

[xsa-dev]

close #10145