Why Did Gold Fall When Inflation Was Still High?

The case: a 26% drop with inflation still running hot

Start with what actually happened, using MacroGold's own tracked data. Gold (COMEX futures, which trades essentially in step with spot gold) peaked near $5,405 an ounce over the past twelve months and closed at $3,974.95 on October 30, 2026 — a drawdown of about 26% from that high, with the chart now in what our system flags as a confirmed downtrend. Over the same window, US headline CPI (the Consumer Price Index, the government's main measure of retail price inflation) was still rising 3.72% year-on-year as of its September 2026 reading, and core PCE inflation (the Fed's preferred gauge, which strips out food and energy) was up 3.12% year-on-year. Both readings sit near the top of their own one-year range — this was not a brief inflation spike that faded. Inflation stayed uncomfortably high while gold fell hard. That sequence is the opposite of the simple story many readers have absorbed: "inflation up, gold up."

Read this table left to right: gold roughly tracked a stable real yield into its February peak, then gave back most of its gain as the real yield climbed by nearly a full percentage point into October — while inflation itself barely moved.

Why "inflation is high" is the wrong lens

The popular shorthand treats gold as a direct hedge against the inflation rate. But gold pays no interest or dividend, so the question an investor actually weighs isn't "is inflation high," it's "what am I giving up by holding a zero-yield asset instead of a bond." That opportunity cost is measured by the real yield: the nominal yield on a Treasury bond (what the bond pays in dollars) minus the market's expectation of future inflation, which is captured by the "breakeven" rate (the inflation rate baked into the pricing difference between a regular Treasury bond and an inflation-protected one, called a TIPS, or Treasury Inflation-Protected Security).

Here is the part that breaks the simple narrative. The 10-year breakeven rate was 2.35% on October 29, 2026 — up only 5 basis points (a basis point is one-hundredth of a percentage point) over the entire past year. Meanwhile the nominal 10-year Treasury yield climbed about 106 basis points, to 5.05%. Almost the entire rise in nominal yields flowed straight into the real yield, not into inflation expectations. In other words, bond markets didn't get more worried about inflation over the past year — they demanded more compensation in real terms, likely reflecting heavier government borrowing, a firmer growth outlook, or a central bank seen as staying restrictive for longer. Whatever the cause, the effect on gold is the same: holding it got more expensive relative to holding a bond, regardless of what CPI printed that month.

What the correlation data actually shows — and where it misleads

An honest reader checking this story might pull up the daily correlation between gold and the real yield and find it unconvincing: over the most recent 21 trading days, it is just 0.05, essentially zero. Taken at face value, that number looks like it disproves the whole real-yield explanation.

It doesn't, and this is worth sitting with because it's a common trap. Day-to-day, gold is dominated by many short-term flows — momentum, options hedging, currency moves — that drown out a slower-moving variable like the real yield, which itself changes by only a few basis points on a typical day. When the same relationship is measured over 60-day changes instead of daily moves, the correlation is -0.52 — a clear, intuitive, inverse link between rising real yields and falling gold. The data provider's own note on this series is blunt about it: a near-zero daily correlation "doesn't contradict" the real-yield channel, it's simply the wrong timeframe to detect it. The honest limitation here is that no single correlation number tells the whole story; the horizon you measure over changes the answer, and anyone quoting a correlation should say which one.

Ruling out the other usual suspects

Two other explanations deserve a direct check rather than a wave of the hand. The first is speculative positioning: the idea that futures traders got overcrowded on the long side and the subsequent unwind did the damage. The CFTC's Commitments of Traders report, as of October 20, 2026, shows net non-commercial (speculative) long positioning at 61.5% of its one-year range — our own data labels this "neutral, no contrarian signal." Positioning did shrink by about 26,000 contracts in the prior four weeks, consistent with a market cooling off, but it started from an unremarkable level, not an extreme crowded long. That rules out "forced liquidation from an overcrowded trade" as the primary driver.

The second is the US dollar. A broad dollar index has trended higher over the past three months, and gold's correlation to the dollar has been consistently negative (around -0.51 over the past year in our data) because gold is priced in dollars internationally — a stronger dollar makes it costlier in other currencies, which can weigh on demand. This is a real, secondary headwind, but it is smaller and less consistent than the real-yield move, and it doesn't explain why the decline tracks the real yield's path so closely on a multi-week basis.

The takeaway, and its limits

The mechanism that survives scrutiny here is the real interest rate, not the headline inflation rate. Gold fell through 2026 not because inflation surprised to the downside — it didn't — but because nominal bond yields rose faster than inflation expectations did, lifting the real cost of holding a non-yielding asset. Investors tracking this relationship over time, including around scheduled CPI and Fed decisions on a calendar like https://macrogold.ca/calendar, are really watching for how nominal yields and breakevens move relative to each other, not just whether the inflation number itself was hot or cool.

This explanation has real limits, and they're worth stating plainly. A one-year window with one dominant case study is not a proof of a universal rule — other years have shown periods where gold and real yields moved together rather than inversely, and our own backtesting has not found a reliable, tradeable edge in this relationship at any fixed horizon. Correlation, even measured on the "right" timeframe, is not causation, and real yields are themselves driven by forces — fiscal policy, growth expectations, central bank credibility — that can also affect gold directly through other channels. What the data supports is narrower and more useful than a prediction: when gold and inflation diverge, the real yield is the first place to look before assuming the relationship is broken.