A plain-English guide to gold as a macro signal.

Gold closed below 4,000 dollars an ounce today, the first sub-four-thousand print in three months and an 18-percent retreat from the early-April high. The analysis piece that ships alongside this one reads the break as a four-driver pile-up: a positioning unwind, a real-yield path that did not deliver the relief the surface read implied, and a leaking geopolitical premium. None of those is obvious unless you know what gold is actually pricing in the first place. This is that explainer.

Gold is the asset macro traders watch most and the asset they explain worst. The standard explanations (it is an inflation hedge, it is a fear gauge, it is a currency) are all partly true and all individually wrong. Gold prices into four distinct drivers, and what matters in practice is not which one is "the" explanation but which ones agree and which ones disagree on a given day. When they all push the same way, the move is informative; when they disagree, the disagreement is the signal.

Driver one: the real-yield identity

Gold pays no coupon. Holding an ounce instead of a one-year Treasury bill costs the holder the real yield on that bill. When real yields rise, gold's relative cost of carry rises and the metal weakens; when real yields fall, the carry argument flips and the metal firms. This is the cleanest single relationship in the gold market, and the one most desks anchor on first.

The mechanic uses the real yield, not the nominal yield. Real yield is nominal minus expected inflation, and the standard market measure is the 10-year TIPS yield (the break-even framework we covered last month does the decomposition). When you read commentary that says "the 10-year fell, so gold should rally," verify the break-even before treating the move as a signal. A nominal drop driven by collapsing inflation expectations does not move the real yield, and gold does not get the relief the headline suggests.

Driver two: the dollar inverse

Gold is priced in dollars on the headline benchmark (the COMEX-derived spot, what we cite as XAU/USD). When the dollar strengthens against the rest of the basket, gold gets cheaper for non-dollar buyers everywhere, all else equal. That makes the metal quoted in their own currency more attractive, which would in isolation lift dollar-quoted gold via cross-currency arbitrage. The inverse correlation between DXY and gold is one of the more reliable medium-frequency relationships in macro.

The relationship breaks down in two specific regimes. The first is a global risk-off event where the bid for the dollar (the world's reserve currency) and the bid for gold (the world's reserve store of value) move together. The second is a US-specific stress episode (a debt-ceiling impasse, an institutional shock) where the dollar weakens but the gold bid does not show up because the fear premium is being priced into Treasuries and the yen instead. Both regimes are recognisable in real time. Most days, neither is in play, and the dollar inverse holds.

Driver three: the central-bank bid

Central banks have been net buyers of gold every year since 2010 and the marginal buyer in the 2024-2026 stretch. The People's Bank of China, the Reserve Bank of India, Turkey, Poland, the Central Bank of Russia (where positions can be verified), and a long tail of emerging-market sovereigns have together purchased enough physical metal to set the floor under price during periods when speculative positioning was light. The IMF World Gold Council composite estimates these flows at over 1,000 tonnes per year for three consecutive years, an unprecedented run.

The flow is slow. Central banks do not buy on intraday weakness and do not chase intraday rallies. They report quarterly with a lag. On a daily-close horizon they are invisible. On a 90-day or annual horizon they are why the metal does not retrace the way the real-yield model says it should. When you see a gold print that the real-yield model "shouldn't" support, central-bank flow is usually the missing factor. The print is correct; the model is incomplete.

Driver four: the risk-off impulse

Gold rallies on shocks. The 2008 financial crisis, the 2020 pandemic, the 2022 Russia-Ukraine outbreak, the 2024 Hamas-Israel sequence and the 2025-2026 Iran flare-up all produced one-to-three-week gold spikes that decayed once the immediate risk faded. The shock-and-decay pattern is the geopolitical premium component of price, and it sits on top of the real-yield, dollar, and central-bank legs as a distinct layer.

Reading geopolitical premium in real time is the hard part. A market that is pricing roughly a 15-dollar-per-ounce premium on top of the macro-implied fair value will give that premium back over weeks when the headline driver fades. The standard tell is the term structure of gold ETF flows: when the front-of-curve flows fade while the back end stays bid, the premium is being unwound but the strategic base is intact. When both fade together, the bid is gone.

Putting the drivers together

The four drivers do not weigh equally on every horizon.

• Intraday. Real yields and the dollar dominate. Central-bank flow and geopolitical premium are slow-moving and do not move the tape inside a session unless an outright purchase announcement or a war headline crosses.
• Weekly to monthly. All four drivers compete. The most useful single decomposition is to compute the implied gold price from real yields and the dollar (a two-factor regression on the trailing year) and read the residual as the central-bank + geopolitical leg. A widening positive residual is the bid strengthening; a narrowing residual is the bid leaking.
• Multi-year. Central-bank flow dominates. The 2024-2026 run from 2,000 to 4,800 is not explained by real yields (which have not moved enough) or the dollar (which barely moved on net). It is explained by structural sovereign buying.

How to use this when reading a gold print

The workflow on any major gold print: list the four drivers' state at that moment, mark which are pushing the same direction as the print and which are pushing against it. The print is informative when at least three agree. When the print disagrees with at least two drivers (the way today's break of 4,000 disagrees with the real-yield and dollar legs), the move is being driven by the residual factors (positioning unwind, flow rotation, premium leakage) rather than by macro fundamentals. That is not a less-real signal. It is just a different signal.

Where this shows up in practice

Recent analysis pieces that lean on parts of this framework:

• Gold broke 4,000. USD/JPY pressed to 162. The driver decomposition for today's print.
• Brent breaks $75, bonds rally. The Iran-de-escalation flow that compressed the gold geopolitical premium leg by collapsing the energy premium first.
• A plain-English guide to real yields. The first driver in detail, including the nominal-vs-real distinction the casual read keeps getting wrong.

The framework is incomplete in two specific ways. It does not handle the producer-hedging book (gold miners' forward sales, which produce predictable supply that adds to the cap during high-price periods). It does not handle the jewellery-and-craft physical demand, which is a real but slow-moving floor. Both are structural and worth understanding for a complete picture, but neither moves the daily tape in a way the four-driver model misses.