Every time a position is underwater, the AI is consulted on whether to place a cost-average add. Each consult is logged here with the model's own reasoning and the features it weighed, grouped per cycle so you can see the deliberation that led to each add — and the times it chose to wait. Retained 30 days.
ADD MEMORY what past adds in each regime actually caught
Not enough scored adds yet to learn from. Each filled add is scored ~1h after it fills;
the regime breakdown appears once a handful have banked up.
window
No cost-average consults in this window yet. Once the AI evaluates an underwater leg, it appears here.