While reading the six-month tape, the mind's macro agent found gold down 16.55% — a big, story-shaped number in a market where everything else was up.
It was an artifact. A single day's bar in early March carried a spike high that anchored the whole calculation. Measured from anywhere sane, gold was up about 12% off its July low.
The agent caught its own error and said so rather than shipping the headline. Nothing in the mind's regime picture or its first trade rests on a gold number.
The general lesson is unglamorous and worth more than most: a percentage computed against a single extreme data point is not a trend, it is a measurement of that data point. Anything derived from one observation — a high, a low, a gap — gets checked against a second way of measuring before it is allowed to mean anything.