pandas FutureWarning: Series.__setitem__ treating keys as labels

[atharvajoshi01]

Several places in the codebase use integer-based indexing on Series via __setitem__ (e.g., series[i] = value) instead of .iloc[i]. With recent pandas versions this triggers FutureWarnings and will break in pandas 3.0 when integer keys are always treated as labels.

This is the same class of bug that ta library fixed in their PSARIndicator (bukosabino/ta#361).

Spotted while reviewing PR #1346. Happy to put together a PR if you want — would just need to grep for __setitem__ patterns on Series objects and replace with .iloc[] where the intent is positional.

[atharvajoshi01]

Dug into this a bit more. The main _equity tracking in backtesting.py uses np.tile which returns an ndarray, so the integer indexing there is fine. The warnings are more likely coming from the indicator/data wrapper classes where pandas Series objects get indexed by position with integer keys.

The fix is mechanical: replace series[i] = value with series.iloc[i] = value wherever the intent is positional. Same pattern as the ta library fix in bukosabino/ta#361. I can put up a PR if there's interest.