BUG: Preserve stop-loss value in stats._trades when SL is gapped through

[nyxst4ck]

Problem

When a trade is closed by its stop-loss, the SL value is sometimes missing (NaN) in stats._trades["SL"], even though an SL was explicitly set. This happens specifically when the price gaps through the stop: the bar opens beyond the stop price, so the fill price is worse than the stop.

This is the still-unfixed part of #1288 reported in #1340.

Root cause

When a stop order is hit, the stop becomes a market order and its stop_price is cleared:

order._replace(stop_price=None)

Since the trade reads its SL from trade._sl_order.stop, clearing it drops the value from the trades data frame. The previous fix (#1288) restored it, but only when the fill price equalled the stop exactly:

if price == stop_price:
    trade._sl_order._replace(stop_price=stop_price)

On a gap-through, the order fills at the (worse) open price, so price != stop_price and the SL is never restored.

Fix

Restore the recorded SL whenever the executed order is the trade's own stop-loss order, regardless of the fill price. The TP/trade.close() paths are unaffected (those orders carry no stop price).

Test

Added test_sl_value_in_trades_df_when_gapped_through, which enters a long position whose SL (99.5) is gapped through on the next bar's open (99.19). The exit price is below the stop, but the SL value must still be recorded.

• Red on master: AssertionError: None != 99.5
• Green with this change.

Existing SL/TP regression tests (test_sl_tp_values_in_trades_df, test_sl_always_before_tp, test_stop_entry_and_tp_in_same_bar) still pass.

Fixes #1340

[nyxst4ck]

Marked ready for review. I do not see any checks reported on this branch yet.

[kernc]

Many thanks! If you find any other similarly neat fixes, they are welcome!