GOLD: The Seller Runs Out. The Buyer Doesn'tOANDA:XAUUSD
GOLD: A RATES REGIME PRICED OFF DATA THAT ALREADY TURNED
Regime analysis and the July book | XAUUSD W-D-4H-1H
====================================================================
THESIS
Gold's 28% drawdown from January is not a fundamental repricing. It is
a change in who sets the marginal price. Understanding that distinction
is the whole trade, because the two marginal actors have completely
different exhaustion profiles - and only one of them is running out.
====================================================================
I. REGIME IDENTIFICATION: WHAT IS ACTUALLY PRICING GOLD
Gold is not one asset. It is a function of four inputs: real yields,
the dollar, official-sector flow, and crisis premium. Which of those
dominates is not constant. That is the regime, and almost nobody
trading this chart has named which one they are in.
2022 - 2025: OFFICIAL-SECTOR REGIME. Central banks bought ~1,000t
per year. Gold decoupled from real yields to a degree that broke
most models. Rate-based frameworks stopped working, and everyone
quietly stopped using them.
2026: RATES REGIME. Gold recoupled. This drawdown IS the recoupling.
The mechanism is not mysterious. Iran blockaded the Strait of Hormuz
in late February. Energy repriced. US inflation hit 4.2% y/y - a
three-year high. That forced a hawkish repricing of the Fed path. Real
yields rose, and gold - which yields nothing - repriced against them
mechanically. Every 10bp of real yield increase raises the carrying
cost of a non-yielding asset.
That is a clean, complete, unsentimental explanation for a 28% decline.
No conspiracy, no manipulation, no broken market. Just a factor
regime that changed and a lot of positioning that hadn't noticed.
====================================================================
II. THE REGIME IS RUNNING ON STALE INPUTS
Here is where it gets interesting.
The hawkish repricing was driven by an ENERGY-DRIVEN inflation shock.
That shock has already reversed:
- WTI has collapsed below $69.
- June payrolls printed 57,000 against a 110,000 forecast - a
catastrophic miss that roughly HALVED September hike odds.
- The June FOMC minutes revealed a committee split 9-to-8 on hikes.
Not a hawkish committee. A deadlocked one.
Gold is currently priced off an inflation impulse whose source has
already deflated, and off a hawkish Fed path that a nearly-tied
committee is visibly struggling to justify.
This is the setup that matters: the REGIME is intact, but the INPUT
driving the regime has turned, and price hasn't.
A rates regime does not end because someone declares it over. It ends
when the rate path that sustains it stops being credible.
====================================================================
III. THE FLOW ASYMMETRY - THE ACTUAL EDGE
This is the part I would build a book around.
THE MARGINAL SELLER IS EXHAUSTIBLE.
Gold ETF outflows have run since May. Rolling 90-day flows peaked near
+$30bn in late February and now sit at -$5bn to -$10bn. That seller is
rate-sensitive, mark-to-market, and finite. When the rate view flips,
the selling doesn't slow - it reverses. And ETF holdings remain well
BELOW their pandemic-era peak, meaning positioning is not stretched
and inflow capacity is fully intact.
THE MARGINAL BUYER IS NOT EXHAUSTIBLE.
Central banks never stopped. Not for one month of this drawdown. The
PBoC has now run a 20-month buying streak, ramping from ~1t/month
through February to 5t in March, 8t in April, 14.93t in June - total
holdings 2,346t. Chinese net imports hit 317t in Q1, nearly triple the
prior quarter.
And in May, Goldman found that UK trade data had understated London
vault outflows since August 2025, forcing an upward revision of
sovereign demand to 60 TONNES PER MONTH from 29.
Read that again. The price-insensitive bid was roughly DOUBLE what the
market believed. That is not a forecast. That is a measurement error
that has now been corrected.
Central banks do not respond to FOMC meetings. They accumulate on
decade-long reserve mandates. They are the definition of a price-
insensitive buyer, and there are more of them coming: an OMFIF survey
of 90 central banks and sovereign funds on June 30 found - for the
first time ever - more institutions planning to CUT dollar allocations
than raise them, with a net 30% intending to add gold within two years.
The asymmetry is structural: a finite, rate-sensitive seller against an
infinite, mandate-driven buyer. The seller sets the price today. The
seller does not set the price forever.
====================================================================
IV. THE VOLATILITY REGIME
Realised vol spiked above 50% during the decline. It has since
compressed below 30%. The 20-year average is 17%.
So: vol is elevated versus its own history, but COMPRESSING - and
compressing directly into a scheduled binary event. Gold vol spikes
historically mean-revert.
Compression into a catalyst is not indecision. It is a market waiting.
====================================================================
V. THE EVENT
FOMC: July 28-29. Rate decision Wednesday July 29, 2:00pm ET.
Current target range: 3.50% - 3.75%. Held in June. No SEP at this
meeting - no dot plot to anchor the reaction.
A 9-8 committee split, a no-projection meeting, and compressed vol.
That is close to a definitional coin flip with a fat tail on each side.
SCENARIO TREE (my subjective weights, argue with them):
HAWKISH HOLD (~45%)
Rates unchanged, aggressive forward guidance. Sustains pressure
without new downside momentum. Gold grinds. Range persists.
3,900 - 4,100.
NEUTRAL / DOVISH HOLD (~35%)
Rates unchanged, softer guidance acknowledging the payrolls miss and
energy deflation. This is the regime crack. Real yields fall, the
rate-sensitive seller stops, and there is no supply above.
Target 4,300 - 4,400 (former support, now resistance).
SURPRISE HIKE (~20%)
The deepest downside. Consensus places gold in the 3,895 - 4,000
band on this. Deutsche Bank flags 3,800 on a three-to-four hike path.
Rates regime confirmed and extended.
Note what the tree says: the modal outcome is NOTHING. The distribution
is fat-tailed, not directional. That has a specific implication, and
it is not "pick a side."
====================================================================
VI. THE STRUCTURE ON THE CHART
Descending trendline from the May 11 high converges with the 3,960
shelf around JULY 28.
The apex is the FOMC. That is a calendar fact, not an interpretation.
Levels:
Support shelf 3,955 - 3,995 (tested three times)
Trendline ~4,010, falling ~2 pts per 4H bar
Resistance 4,190, then 4,300 - 4,400
Downside 3,895 - 3,900, then 3,800
One correction to what you will read elsewhere: the "support held three
times therefore support is strong" reasoning is backwards. Each test
CONSUMES resting bids. It does not replenish them. Every stop on this
chart sits in the same place, and it has been advertised for a month.
A sweep through 3,960 that recovers inside the bar is the single most
likely shape of the resolution, and it is engineered to look exactly
like a breakdown.
I want the CLOSE. Not the wick.
Also: do not measure the full triangle. It gives ~3,130 and it is
nonsense. A four-month pattern does not generate a credible 4H target.
====================================================================
VII. HOW TO TRADE IT THIS MONTH
The most useful thing I can say: SEPARATE THE BOOKS. The regime thesis
and the July P&L are different questions, and conflating them is how
people with a correct view still lose money.
STRUCTURAL BOOK (6-18 months)
The rates regime has an expiry date because its input already turned
and its seller is finite. Positioning is not stretched. The official
bid is double what was modelled. This is an accumulation thesis, sized
to survive 3,800. Institutional targets cluster 4,400 - 5,500 with
Deutsche the lone 3,800 dissent - a ~65% dispersion, which is itself
the signal: the payoff is bimodal and nobody has an edge on direction.
JULY BOOK (13 days)
You are not trading a view. You are trading an event with compressed
vol into a deadlocked committee. That argues for CONVEXITY, not
direction: own optionality across the apex rather than picking a side
before the information exists. If options are not in your toolkit,
the honest equivalent is: do nothing, and be ready.
THE BREAK IS NOT THE TRADE. THE BREAK IS THE INFORMATION.
Which way this resolves tells you which regime governs H2. That is
worth more than the 50 points you would make guessing it.
====================================================================
VIII. WHAT KILLS THIS THESIS
- CPI reaccelerates despite crude sub-$69. The energy-deflation leg
of the argument dies, and the hawkish path becomes justified rather
than stale.
- China's SAFE reserve data breaks the 20-month streak. Watch the
first week of the month. That would be the first interruption since
November 2024 and would gut the price-insensitive-buyer argument
outright. This is the single highest-information datapoint in the
entire thesis.
- A sustained break below 3,955 with real yields RISING. Then the
regime is not expiring, it is extending, and I am early - which in
this business is the same as wrong.
- Gold spends late July at 3,970 with vol under 20%. Then this is a
range, not a coil, there is no catalyst resolution, and I have
over-read a calendar coincidence.
====================================================================
IX. WHAT I DO NOT KNOW
Stated plainly, because most posts hide this:
- I do not have live positioning data. CFTC managed-money net length
would materially sharpen the crowding argument and I have not seen
it.
- I do not have the implied vol surface. Whether to own or sell
convexity into July 29 depends entirely on IV vs RV, and I am
reasoning from realised vol reported second-hand.
- I do not have the current 10y TIPS yield. The recoupling claim in
Section I is the load-bearing wall of this entire piece, and it
should be tested with a rolling gold/real-yield beta rather than
asserted from narrative.
- Bank price targets are close to worthless as timing tools. I cite
them for dispersion, not direction.
If you have a terminal, those four inputs would confirm or destroy this
thesis in about twenty minutes. I would rather tell you what would
falsify it than pretend I've already checked.
====================================================================
This is analysis of market conditions and factor regimes. It is not
financial advice, not a recommendation, and not a signal. I am not a
financial advisor. Trading involves substantial risk of loss.
