Wall Street Couldn't Call Gold Right Either Direction, and Nobody's Owning It

Wall Street Couldn't Call Gold Right Either Direction, and Nobody's Owning It

Every major shop on the Street spent the second quarter revising gold targets in the wrong direction at the wrong moment, and now that the metal is ripping back toward $4,300 an ounce, the same desks that spent June and July cutting numbers are conspicuously quiet about it.

Walk through the timeline, because it's worth sitting with. In January, gold hit its record near $5,600 an ounce, and the sell-side consensus for year-end 2026 clustered somewhere between $4,900 and $6,300, built on the assumption that the Fed was about to start cutting. Reasonable enough call given the setup at the time. Then Warsh happened, the "sound money" doctrine happened, and gold proceeded to have its worst quarter since 2013, bleeding from that January peak down through the $4,000 level by early summer. Goldman capitulated on June 19th, slicing its year-end target from $5,400 to $4,900 and pushing its first expected cut all the way out to 2027. JPMorgan followed on July 3rd, tearing up a $6,000 glide path in favor of a far more modest $4,300 for Q3 and $4,500 for Q4. HSBC joined on July 9th, trimming its 2026 average from $4,864 to $4,560. Three of the largest research shops in the world, cutting numbers in the same three-week window, all chasing the same falling knife downward.

Here's the part that should embarrass everybody involved: JPMorgan's freshly lowered Q3 target of $4,300 was basically hit by Friday afternoon, a little over a month after the note went out. Gold climbed toward that exact level on the back of a July jobs report that cut 23,000 positions nobody expected to disappear, with 103,000 in downward revisions layered on top for good measure. The bank downgraded its number because the old one looked embarrassing next to a collapsing chart, and the metal turned around and closed the gap almost immediately once the macro data gave it a reason to. That's not forecasting. That's marking your homework to the test after you've already seen the answer key, and still getting the timing wrong.

None of this is really about gold. It's about what happens to institutional conviction the moment a trade stops cooperating. Every one of those revisions came dressed up in careful language about "recalibrating the policy path" and "reassessing the rate-cut timeline," but strip out the hedging and what actually happened is simpler: the price fell, so the target fell with it, three separate times, from three separate buildings on nearly identical calendars. Nobody at Goldman, JPMorgan, or HSBC was making an independent judgment about where gold was going. They were watching the tape and writing it down afterward with a research logo stapled to the top.

Compare that to who actually held their nerve through the drawdown. European ETF investors added 17.3 tonnes in July alone, worth roughly $2 billion, the second-best inflow month of the year for that region, led by UK buyers adding $875 million and Swiss buyers adding $657 million — both markets now sitting above $2 billion in net inflows for the year. That's money moving in size while the sell-side was writing capitulation notes, not chasing a bottom after the fact. China told a similar story from a different angle: with the CSI 300 posting its sharpest monthly decline since January 2016, institutional investors there leaned harder into gold-backed assets as a hedge against the equity rout, layering on top of central banks that never stopped accumulating through any of this. The dumb money, if you want to use that phrase, bought the correction. The smart money, or what passes for it on a research floor, spent the correction downgrading targets that the market would go on to invalidate within weeks.

There's an obvious lesson buried in here about position sizing and about not outsourcing conviction to a research note with a six-month shelf life, but the more interesting question is what this says about the current information environment more broadly. The Fed under Warsh has deliberately reduced how much it tells markets about its own thinking, which means every desk on the Street is now pricing gold, rates, and everything downstream of them off a thinner data diet than they're used to. When the inputs get noisier and the forward guidance disappears, you'd expect institutional forecasters to widen their error bars and hold positions with more patience, not less. Instead the last quarter showed the opposite: convictions got shorter, revisions got more frequent, and every downgrade arrived just in time to look foolish a few weeks later.

Gold's rebound toward $4,300 isn't a vindication of anyone's macro thesis. It's a jobs report that broke badly enough to drag hike odds back out of the market, full stop. But the fact that the metal is sitting almost exactly where JPMorgan grudgingly lowered its target to, mere weeks after making the call, is a decent proxy for how much of "research" right now is just narration with a delay built in. Watch the flows, not the notes. The flows didn't blink.