Daily brief, 2026-09-17.
This edition is kept as published — it does not update. The live brief is at /r/brief.
The read that morning
Oil printed +12.69% over five sessions and added another +3.40% to 104.84 on Wednesday, a stretch that surprised me because gold has been the headline trade of late and crude barely registered. The usual pattern this year has been a strong dollar and weak everything-risk, yet here is energy ripping while US equity closed -0.45% on the day at 760.88 and gold slipped to 4,325.2, down -2.56% over the same five sessions. Energy and bonds moving in opposite directions is exactly what the asset-class tape shows: energy ranks +1.12 and bonds -1.25, with the dollar at +0.57 between them.
What is happening versus what it means: oil is rallying and gold is falling in the same week, and I want to be careful here because that coexistence alone does not prove a story about inflation or geopolitics. The activity tape, though, is unambiguous about where attention sits. Palladium leads metals and crude leads energy on the one-session move, while in crypto the action is on Arbitrum and Base with +12% volume surges on the week and the Privacy narrative up +10.73% in a single session. That is the speculative tail rotating; the risk-cycle read is Large-Cap Alts, with capital moving from BTC into ETH and SOL.
A few systematic watch levels sit nearby. On XAUUSD, the daily-trend sell stop at 4,262.2 sits about 1.5% below the close, and a break there would be the second-consecutive trigger confirming a fresh short side; on USDJPY, a buy stop at 156.30, about 0.3% above spot, marks the corresponding upside trigger. For the crowd read, the COT gold speculators index at 69 sits in the upper quartile — bullish-leaning positioning, a conditioning cell for any fade setup. And for what markets price: betting markets still favor the Fed cutting, recession odds unmoved from their recent range.
Nothing fired yesterday. Every order is still working.
Levels published that day — and what became of them
superseded — we re-drew the level before price reached it
superseded — we re-drew the level before price reached it
superseded — we re-drew the level before price reached it
superseded — we re-drew the level before price reached it
superseded — we re-drew the level before price reached it
superseded — we re-drew the level before price reached it
Graded against the journal, net of modelled costs, stop-first on ambiguous bars.