A gold-and-money rabbit hole · Sep 2026
A widely-shared thread says gold is a record 171% of the money supply, and that the last two times this line ran hot the world got a monetary reset. The metric is real. The record isn’t — and at its extremes it has marked gold tops, not launchpads.
The thread, from @bravosresearch: gold has climbed to a record 171% of total money supply, and the only two prior times this line ran this hot — the 1930s and the 1970s — both ended in a major monetary reset. Stitch in China halting retail “paper gold” trading, an estimated 133 paper claims per ounce of real metal, and a debased dollar, and you get a tidy conclusion: gold’s price is artificially suppressed and must reprice higher.
Credit where it’s due, because a lot of this is real. The gold-coverage ratio — the market value of a country’s gold against the money it has printed — is a genuine, century-old gauge of how “backed” a currency is. The monetary history checks out: the 1934 revaluation from $20.67 to $35, the 1944 Bretton Woods dollar-for-gold promise, Nixon closing the gold window in 1971. China really did suspend retail paper-gold trading at its big banks in July 2026. And gold really did spike near $5,600 in January 2026 before shedding about 30%. None of that is invented.
But a chart that’s meant to time something has to be tested against what actually happens next. So I rebuilt the ratio from scratch — the United States’ constant ~261.5 million ounces of gold times the monthly gold price, divided by the monetary base and by M2, back to the month gold began to float in 1971 — and asked it three plain questions. Are we really at a record? Does a high reading foretell gold repricing higher? And does it warn of trouble in stocks, the way “major warning sign” implies?
Their headline: a record 171%
What gold actually is as a share of the US monetary base today — sitting right on its 54-year median. The metric’s real record was ~113%, in January 1980.
After a 1980-style extreme
Average next-12-month gold return from the ratio’s top decile. High readings have marked gold tops (rank corr. −0.20 to −0.35), not launchpads.
The ratio as an equity warning
Correlation between the ratio and the next year’s S&P 500 return. As a stock-market warning sign it’s statistical noise.
Here is the whole series since 1971. On the standard construction — gold’s market value over the US monetary base — the line peaked near 113% in January 1980, at the blow-off top of the last great gold mania. Today it sits around 19%, essentially dead on its half-century median, well down the left-hand slope of that mountain. Measured against the broader M2 it’s about 4.6%, versus a 1980 peak near 12% — again, mid-pack, not extreme. There is no denominator I can build from actual US money and actual US gold that puts today at 171%, or anywhere near a record. That figure comes from a smaller, undisclosed base; the honest versions say something much more ordinary.
market value of ~261.5M oz ÷ US monetary base · year-end points
Now the part the thread gets backwards. Sort every month since 1971 into ten buckets by how high the coverage ratio was, then look at what gold did over the following year. The relationship is negative at every horizon — a rank correlation of −0.20 at one year, −0.35 at three. The cheap end (a low ratio) is where forward returns cluster high. The richest bucket — the 1980-style extreme the thread romanticizes — delivered a negative average over the next one, three, and five years. That is exactly the history: gold rose into a high ratio, and once the ratio got stretched, it fell for two decades. A high reading is a mean-reversion warning on gold itself, not a coiled spring.
gold vs. S&P 500 · decile 1 = cheapest ratio, decile 10 = richest (1980-style)
The thread frames the ratio as a “major warning sign” for markets. It isn’t. The correlation between the ratio and the next year’s S&P 500 return is +0.02 — and it’s faintly positive, not negative, out at longer horizons. In the top decile, the very readings that supposedly precede a reset, stocks went on to return roughly +11% over the next year, +43% over three, +84% over five. Whatever the gold-coverage ratio measures, it is a statement about gold’s valuation against money, not a forecast of equities. Using it to brace for a stock-market crash has no historical support.
An indicator earns its keep in a portfolio, so I wired the ratio into the simplest honest rule: hold the asset only when the coverage ratio is below its own expanding median (“cheap”), otherwise sit in cash, with the signal lagged a month so nothing peeks at data it couldn’t have had. It does contain the one real nugget — the cheap-side rule sidesteps the worst of gold’s post-1980 collapse and lifts gold’s risk-adjusted return slightly. But the thread’s implied trade — buy because the ratio is high — is the losing side of every column below.
| Rule | Gold CAGR | Gold MaxDD | Gold Sharpe | S&P CAGR | S&P MaxDD | S&P Sharpe |
|---|---|---|---|---|---|---|
| Buy & hold | 6.9% | −62% | 0.49 | 9.1% | −53% | 0.66 |
| In when ratio CHEAP | 5.5% | −41% | 0.56 | 7.2% | −50% | 0.62 |
| In when ratio RICH | 1.3% | −57% | 0.16 | 1.8% | −30% | 0.26 |
Being long gold because the coverage ratio is high — the thread’s whole thesis — earned 1.3% a year with a 57% drawdown. The ratio is a mild overvaluation gauge, not a reset alarm.
A high reading has said “gold is expensive,” not “gold is about to explode.”So where does that leave the thread? Its instinct — that money-printing and gold belong on the same page — is sound, and its facts are mostly right. What doesn’t hold is the load-bearing chart. We’re at the median of the coverage ratio, not a record; the metric’s genuine extremes have preceded gold declines, not repricings; and it carries no information about equities at all. The two-episode “it precedes resets” pattern is also a tiny sample dominated entirely by 1980 — the sign of the effect is clear, but you can’t build a base rate from one mountain.
BOGMBASE) and M2 (M2SL). Monthly, Jan 1971–Jul 2026.None of this makes gold a bad thing to own, or the worry about debt and debasement invalid — reasonable people hold gold for exactly those reasons. It just means this particular chart isn’t the timing device it’s presented as. “Record 171%” is a headline; “median, and historically a top signal” is what the data says. One rabbit hole deeper, the reset keeps not arriving on cue.