What really happened to gold through the 2020 crisis

The panic phase came first

When the shutdowns began, the first thing that happened to gold was not a rally. Price dropped over a short run of sessions, in the same window that equity indices were falling and credit markets were seizing up. Anyone holding metal as insurance watched the insurance lose value at exactly the moment it was meant to pay out. That is not a data glitch and it is not unique to that year. It is what happens when the thing in short supply is cash rather than safety.

The move was fast in a specific way. Spreads widened, quotes thinned, and ranges that normally take most of a week appeared inside a single session. Working in that state is a different job from reading an orderly trend, which is why a separate approach to fast markets is worth having ready before you need it. The detail to hold onto is the order of events: gold fell first and recovered later, so the one line summary of the year quietly hides the part that would have hurt a position.

Why a haven gets sold in a funding squeeze

The mechanism is boring and it is mechanical. When leveraged positions move against their holders, margin is called. When funds face redemptions, cash has to be raised by the end of the day. In both cases the seller does not get to choose what they would prefer to sell. They sell what can be sold immediately at a known price, in size, without destroying the quote. Gold qualifies on every count, and after a long rise a lot of gold positions were sitting in profit, which makes them even easier to liquidate.

So the asset gets hit because it is liquid, not because the market has changed its mind about it. This is also why correlation tables mislead during stress. The relationships you measured in calm conditions are the relationships between opinions. In a squeeze you are measuring the relationship between funding needs, and nearly everything correlates to one. The practical reading is that gold is a haven against monetary outcomes, not against a shortage of dollars. Those are two different fears and they do not arrive on the same day.

The turn, and what was actually driving it

Once the funding panic was met with central bank liquidity, the second phase began, and it had a cleaner driver. Policy rates were cut to the floor and large scale asset purchases pinned down nominal yields. Inflation expectations, which had collapsed in the panic, recovered. Nominal yield minus expected inflation is the real return on holding cash and short bonds, and that number fell sharply.

Gold pays no coupon and generates no earnings, so its competition is precisely that real return. When the real return on safe cash is negative, the cost of holding an asset that yields nothing goes away. This is the most reliable single thread in gold pricing, and it is worth studying on its own in the relationship between real yields and gold. The honest caveat is that the link is loose over short horizons. Real yields can fall for a week while gold does nothing, and gold can rally while they are flat.

The plumbing broke in the middle

Between the two phases something unusual happened that most retrospectives skip. Grounded flights and closed refining capacity made it slow and expensive to move metal between vaults and to recast bars into the formats an exchange will accept for delivery. The futures price and the cash price are normally tied together by the cost and certainty of making delivery. When delivery becomes uncertain, that tie stretches, and the gap between the two prices widened far beyond its usual range.

For anyone watching a chart, the lesson is specific. A futures chart and a spot chart are not two pictures of the same instrument, and in stress they can disagree by an amount that matters to a stop. If your levels came from one and your fills come from the other, that is a silent problem. The differences are set out in the comparison of spot and futures gold. It also explains a quirk you will meet later in any volume discussion: futures carry real traded volume, while spot feeds publish price updates rather than contracts.

Investment demand arrived through a wrapper

The buying in the second phase did not look like the buying of earlier decades. A large part of it came through listed funds and futures rather than through coins and small bars, because that is the route a pension fund or a discretionary manager can actually use. Retail demand for physical metal was strong as well, but the retail supply chain was the part of the market that had been disrupted, so small bars and coins often traded at unusually wide premiums to the quoted price.

Two things follow from that. First, the price you see on a screen is the wholesale price, and it is not the price at which a small quantity of metal changes hands in a shop. Second, flow driven demand is reversible in a way that a coin in a drawer is not. A fund share can be sold in a morning. That cuts both ways and it showed up clearly a few years earlier, in the long decline that followed the previous peak. Flow is a description of what has already happened, never a forecast of the next move.

The rest of the year was an argument

After the summer high, gold spent months going nowhere in particular. The policy surprise had been delivered and priced, growth expectations started to recover, and the real yield picture stopped falling. What followed was a broad, choppy range that punished anyone still trading the breakout playbook that had worked earlier in the year. Trend followers gave back, mean reversion traders did better, and the market generally rewarded patience over conviction.

This is the part of 2020 that is most useful and least discussed. The same instrument can require two different approaches within one calendar year, and the switch is not announced. Recognising the change is a skill of its own, covered in how range conditions change the job. On the live XAUUSD chart the Market Regime label exists for exactly this reason. It tells you what kind of market you are in so that you know when a model you like does not apply, and its most valuable output is often the instruction to do nothing.

What a chart reader should keep

Stripped of narrative, the year leaves a short list of durable points.

• Correlations are conditional. The relationship between gold and risk assets depends on whether the market is short of cash or short of confidence.
• A crisis has phases. The first reaction and the considered reaction can run in opposite directions, days or weeks apart.
• The instrument matters. Cash, futures and funds are linked by arbitrage that can stretch when the physical world gets in the way.
• Flows explain, they do not predict. Holdings data arrives after the move it describes.

None of that tells you where price goes next, and nothing here should be read as a view. It is a set of questions to ask when the tape turns disorderly again: what is scarce right now, who is being forced to act, and which version of the price am I actually looking at.

The caveat worth repeating

One crisis is one observation. It is tempting to turn the 2020 sequence into a rule, something like sell the first panic and buy the policy response, but the sample behind that rule is a single episode with a single kind of shock and a single kind of official reaction. A squeeze funded differently, or met with a slower response, would leave a different shape on the chart. Earlier stress events in gold did not all follow this pattern either.

The more defensible takeaway is about preparation rather than prediction. Know in advance how your plan behaves when spreads triple and when your usual levels get skipped entirely. Decide beforehand whether you reduce size, widen stops, or simply stop trading, because that decision is much harder to make honestly while it is happening. Anchoring your habits around a risk framework you wrote in calm conditions, and sizing to survive the worst hour rather than the average one, is the part of this that carries forward to any future event.