Gold is finishing August with its biggest monthly gain since 1999. Depending on the day you measured, the month-to-date move ran between roughly 13% and 15% — it was up 14.69% through 24 August and still near 14% on the 28th, after a 3.18% single-session drop that day took spot to $4,454.08 an ounce.
Twenty-seven years is a long time between records of that kind, and the reporting has largely treated it as a triumph. The context makes it something more interesting than that.
Gold’s all-time high was $5,589.38, set on 28 January this year, after a 23% run in four weeks. The session immediately after that peak, it fell 11.4%. It ground lower for months, dipped under $4,000 an ounce in late June, and has spent August climbing back. Even after the best month in a generation, it remains roughly 20% below that January close, and close to where it began 2026 — while still up around 29% against the same point last year.
The driver is the issuer, not the rate
The proximate cause is the US Treasury. On 19 August it announced it would at least double the maximum size of its liquidity support buyback operations in longer-dated nominal coupons — the 10-to-20-year and 20-to-30-year sectors — from $2 billion per operation to at least $4 billion, effective 9 September and running through 4 November. Long-dated Treasuries rallied on the news: the 10-year fell 6 basis points to 4.647% and the 30-year fell 9 basis points to 5.196%.
Gold went up anyway, and kept going up as yields subsequently backed up. The 10-year has since climbed to around 4.73% after Fed Chair Kevin Warsh warned that inflation has not meaningfully slowed, with market-implied odds of a hike by December running above 70%.
That combination is the tell. A zero-coupon asset is supposed to struggle when real yields rise, because the opportunity cost of holding it goes up. When gold rallies into higher yields and a central bank talking about hiking, the bid is not expressing a view on the path of rates. It is expressing a view on the balance sheet of the entity issuing the alternative — which is precisely what a buyback programme aimed at “liquidity support” in the long end invites people to think about.
Our take: A headline that reads “best month since 1999” and a position that is flat on the year are describing the same eight months. The January episode is the discipline here: a 23% four-week move that ended with an 11.4% drawdown in a single session. Debasement is a slow argument being expressed through a fast, thin market, and those two speeds do not reconcile gently. The August move is real, and it is also a recovery from a blowoff rather than a fresh leg. Treat the monthly record as information about volatility, not about direction.
The near-term calendar is unusually loaded. The first enlarged buyback operation runs on 9 September. The FOMC blackout covers 5–17 September, which puts the 11 September inflation print inside the silence, with nobody available to steer the reaction before the 16 September decision and dot plot. Whatever gold does across those two weeks, it will be doing it without official commentary to lean on.
What to watch
- 9 September. The first operation at the doubled size, and whether the long end takes it as support or as a signal that support was needed.
- The 30-year at 5.2%. It rallied 9 basis points on the announcement. Whether it holds that level through the blackout is the cleanest read on how the buyback is landing.
- 11 September CPI, inside the blackout. A hot print with no Fed speaker to frame it is exactly the setup that produces outsized moves in both bonds and metal.
- Volatility, not level. January showed what this market does when a crowded move unwinds. Position sizing is the variable that mattered then.
Nothing here is a forecast or a recommendation. It is a description of what moved, and of what the calendar puts in front of it next.