CGPT5 proposed enhancements #1319
Conversation
| try: | ||
| value = func(*args, **kwargs) | ||
| if isinstance(value, pd.Series): | ||
| value = value.to_numpy() |
There was a problem hiding this comment.
This case is covered a few lines further below ... 🤔
backtesting.py/backtesting/backtesting.py
Lines 153 to 154 in cc1ebc8
| ('contingent', self.is_contingent), | ||
| ('tag', self.__tag), | ||
| ) if value is not None)) # noqa: E126 | ||
| ) if (value is not None and (not isinstance(value, bool) or value)))) # noqa: E126 |
There was a problem hiding this comment.
Covered by the use of try_, fixed in 8dd1e36.
| warnings.warn( | ||
| f'({data.index[self._i]}) broker canceled order {order} due to insufficient margin ' | ||
| f'(equity={self.equity:.2f}, margin_available={self.margin_available:.2f}).', | ||
| category=UserWarning) |
There was a problem hiding this comment.
This hunk is legit, I think. 👍 If it can be made to match the prior art in 5e72e7b ...
Indeed, CGPT went wild on it and changed the functional behavior of the backtest. I wanted to keep it simple and avoid going down a rabbit hole of changes so trimmed it to the minimum valid changes without breaking the current backtest behavior. |
2 participants
The patch corresponds to issue #1318, auto-AI generated and it does not change the backtesting result.