The strategy has no stop-loss, which is the first thing anyone asks about, and reasonably so. So we tried to add one properly rather than defend its absence.
We swept the threshold across the entire range that could be justified, from tight to very wide, and measured the result each time on the full history.
Every level made things worse, and the pattern of how they made things worse was the informative part: the wider the stop, the smaller the damage, converging on the case where the stop is so far away it never fires. In other words, the best stop level available to us is the one equivalent to having no stop at all.
That is not a licence to run without risk control. It says something specific about this structure: the position is closed either at its target or by the clock, and a price-based exit placed between those two mostly converts recoverable drawdowns into realised losses.
What it does mean is that risk has to be controlled somewhere else, which is why size is measured rather than assumed and why capital is held with room to spare. A strategy that declines a stop-loss has taken on an obligation elsewhere, and should say where.