Central Bank Gold Buying Explained: Why It Moves the Price

What "central bank buying" actually means

When a central bank "buys gold," it is adding to its foreign exchange reserves — the pool of assets a country holds to settle international obligations, defend its currency, or simply store wealth outside its own banking system. Historically that pool was dominated by US Treasury bonds and dollar deposits. Since the 2008 financial crisis, and especially since 2022, a growing number of central banks have shifted a slice of those reserves into physical gold instead.

This is slow, deliberate, balance-sheet management — not a trading position. A central bank does not buy gold because it expects a 90-day rally; it buys because gold has no counterparty risk (nobody can freeze it or default on it) and because it does not depend on the goodwill of another government. Our news feed's physical-demand channel is full of this logic in real time: on October 5, 2026 it carried comments attributed to Bundesbank president Joachim Nagel at the LBMA conference arguing the "case for central banks to keep buying gold remains strong," framed explicitly around geopolitical risk to reserves. The same day, a headline flagged a large reported Chinese gold import pipeline supporting bullion demand. On October 1, coverage noted the Bank of Korea was preparing a further purchase from domestic producers. A separate piece on October 2 was titled, tellingly, "Central Bank Gold Reserves: One Sold, Most Are Still Buying" — a reminder that this is a broad, multi-country trend with occasional exceptions, not a unanimous bloc.

Why it changed how the price behaves, not just the demand number

The more interesting signal isn't the headlines — it's what happens when you try to explain the gold price using the two variables that used to explain most of it: real interest rates (the yield on bonds after subtracting expected inflation, which sets the opportunity cost of holding a zero-yield metal) and the US dollar (gold is priced in dollars, so a stronger dollar makes it costlier for non-dollar buyers).

We ran our standard fair-value regression of gold's level against real rates and the dollar over the last 756 trading sessions (November 2023 to October 6, 2026). It fails. The model's own diagnostic flags it as invalid: the real-rate coefficient comes out with the wrong sign (positive instead of negative), and it explains only about a quarter of the price level's variation (R² = 0.244). In plain terms: on a multi-year view, real rates and the dollar no longer account for where gold's price sits. That gap is exactly where structural, non-price-sensitive buying — central banks among the main candidates, alongside sovereign distrust of dollar assets and term-premium effects in bond markets — would show up. We can't attribute the gap to central banks specifically from this model; we can only say the usual explanation has stopped working, which is itself the tell.

Day to day, the old mechanism is still visible, just smaller than it once was. Measured on daily price moves (where the regression is statistically sound, R² = 0.164), a 10 basis-point rise in the 10-year real yield is associated with a 0.148% move down in gold same-day, and a 1% rise in the dollar with a 1.70% move down. The 21-day rolling correlation between gold and the 10-year real yield sits at -0.38; stretched to 60-day overlapping changes it strengthens to -0.51, which our data confirms is the more reliable window for this relationship — a near-zero daily reading does not mean the real-rate channel is broken, only that daily noise obscures it.

The current snapshot

Read the top half of the table as the fast market: a confirmed downtrend, oversold momentum, and futures speculators still holding a historically large long position even after a correction, which leaves room for further unwinding if sentiment sours. The central-bank story does not appear anywhere in this table, because official-sector reserve data is reported quarterly by the World Gold Council and the IMF, not daily — it operates on an entirely different clock than the price action above it. One trade-press piece from October 2 captured the tension directly, describing positioning as nearing "capitulation" while arguing central bank buying was emerging as the key support — two true observations about different timeframes, bundled into one headline.

What the data does not show

Three gaps are worth naming plainly. First, our feed does not carry a tonnage series — we can report that purchases are being discussed and reported, with dates, but we cannot hand you a verified monthly or quarterly tonnes-bought figure from this dataset, and you should treat any specific tonnage number quoted elsewhere as coming from that third-party source, not from us. Second, official reserve data is published with a lag measured in months; by the time a quarter's purchases are confirmed, the market has already moved on. Even our own futures-positioning data (COT) is released Friday for a Tuesday snapshot — up to three days stale before you read it. Third, and most important: "central banks are buying" describes a multi-year flow, not a forecast for the next six months. A central bank that bought steadily through 2023, 2024 and 2025 can pause, slow, or rebalance without warning, and reserve decisions are rarely announced in advance. Our own backtests have not found a measurable, tradeable edge from overlaying official-sector demand onto short-term price calls, and this article should not be read as one either.

The takeaway

Central bank buying is best understood as a change in the composition of structural demand for gold — a multi-year reallocation away from dollar-denominated reserves, driven by countries wanting an asset with no counterparty risk, not a signal timed to any particular quarter. It plausibly helps explain why gold's price level has drifted away from what real rates and the dollar alone would predict, which is a genuinely different regime from a decade ago. What it cannot tell you is whether the current oversold, historically crowded futures market and the 22.8% drawdown from the 2026 high resolve higher or lower from here — those two forces live on different calendars, and conflating them is the single most common mistake in how this topic gets reported. If you want to track when the next official reserve figures or key macro releases land rather than guess at their outcome, the data calendar at https://macrogold.ca/calendar lists the actual publication dates.