On a war headline, gold outran oil.

Wednesday's tape carried a contradiction worth sitting with. Trump vowed to 'annihilate' Iran after a fresh exchange of strikes, and the classic playbook says buy oil. The market did not. WTI fell about 2% to roughly $68, while gold rose 0.7% to ~$4,036 and silver ran +1.3%. The safe-haven bid was unmistakable - it simply went into metal, not crude.

Our correlation engine explains the divergence, and this is the day's proprietary edge. The physical tanker chains show oil supply RE-ROUTING rather than disappearing: Russian Baltic loadings surged (+106% on the latest scan, +188% the day prior), US Gulf Coast loadings rose +37%, while the Russian Pacific terminal at Kozmino went to zero (-100%). Barrels are moving west and around the disruption, not vanishing from the water - which is precisely why a war escalation could not hold a bid in crude. The risk premium had to express itself somewhere, and it chose the asset with no routing problem: gold. The COT confirms the crowd was already leaning that way - managed money is net-long 115,395 gold contracts, an extreme.

The second thread is a rates-and-currency accident forming in plain sight. The long end backed up again - the 30-year yield pushed to 4.96%, a whisker from 5%, and the 10-year rose 5bps to 4.47% - while the Nasdaq slipped 0.2% and the Dow outperformed. The yen stumbled to a roughly 40-year low near 162.5 against the dollar, putting Japanese intervention back on the clock. This is the exact configuration HSBC flagged as the second-half 'pain trade': a market too comfortable in US Treasuries, the AI trade and the dollar all at once.

The steelman against our read: gold's leadership could simply be momentum into quarter-start rebalancing, and a genuine Hormuz closure - not another exchange of strikes - would still re-rate crude violently regardless of re-routing. The view is wrong if oil breaks decisively higher on a real supply halt, or if the long-end sell-off reverses and the yen finds a bid without intervention. Conviction is high that the Mideast premium is expressing in metal not oil, and medium that the long-end/yen 'pain trade' is the more dangerous, less-priced risk into the second half.

The read to carry: the honest hedge against this tape is not the obvious one. The crowd's instinct - long oil on the headline - is the trade the physical data says fades; the quieter risks are a 30-year yield through 5% and a yen that forces the Bank of Japan's hand. Watch the long end and USD/JPY, not the oil screen.