Is It Too Late to Buy Gold? What the Data Says
As of August 25, 2026, The Gold Barometer reads 29 out of 100. Across the 164 months since 1971 when gold sat within 5% of its own record, the median five-year return after inflation was 28.7%. The other 503 months returned a median 7.7%.
The alert
One email when the zone changes.
Nobody checks a barometer every day. If the score climbs out of this zone, the email finds you.
What the record says
On this record, buying near a record high was not the mistake it feels like. The medians ran the other way.
In money: $10,000 put in within 5% of a record was worth about $12,870 of shopping five years later. The same $10,000 put in away from records was worth about $10,770.
The record highs cluster inside two or three long rises, so these medians lean on a few eras. They knock down one reflex only: that a record price must be a bad moment to buy.
The two records, side by side
| Months since 1971 | Count | 1 year later | 5 years later | 5 years later, after inflation | Worst fall within 5 years |
|---|---|---|---|---|---|
| Within 5% of the record | 164 | 28.8% | 71.8% | 28.7% | -3.5% |
| Everywhere else | 503 | 1.8% | 24.8% | 7.7% | -10.5% |
All cells are medians of overlapping stretches. In practice they rest on a few long eras, not hundreds of separate bets. The full record, month by month, is on the history page.
Why the fear points the wrong way
The fear of a record price treats the record as a ceiling. On this record it behaved more like a milestone inside a longer climb. Momentum, the tendency of a strong year to be followed by a decent one, has carried gold through most of its record highs.
The opposite force is also real. A price far above its long-run, inflation-adjusted average has tended to mean thinner returns over many years. One force looks a year ahead, the other looks a decade ahead, and the score carries both in the entry price signal.
What this cannot say
It cannot say that the next record behaves like the past ones. The months near records sit inside a few long rises, and one more era could rewrite these medians. It also cannot know how long you plan to hold. Over ten years, buying at record highs has paid less than it did over five.
The instrument behind these numbers
The dial below is the one from the home page. It measures seven conditions for a buyer, and the distance to a record is not one of them. Both forces are measured inside the entry price signal, which has its own page.
- Buying conditions today
- Unfavorable
- Months in today's zone
- 151
- The related signal
- Entry price, explained
Common questions
Does a record high mean a crash is coming?
Not on this record. The worst fall within five years was a median -3.5% for months near a record, against -10.5% for the rest. Records mostly appeared inside long rises, not at their ends.
Why did months near record highs do better?
Because record highs cluster inside long rises, and long rises tended to continue for a while. The same force is measured inside the score as the trend part of the entry price signal.
Is gold overvalued right now?
The entry price part of the score compares the real, inflation-adjusted price with its long-run average every day. Its current reading, with its evidence and its limits, is on the entry price signal page.
So the highs were fine and yet today scores low. How do both hold?
The score weighs seven conditions, not the distance to a record. High prices pull down the part that asks if gold is expensive against its own past, even while the trend part reads strong. The blend, and the reasoning behind it, is on the methodology page.
Today's reading, with the buyer's verdict and the evidence behind it, is on the home page. One email goes out the day the reading moves to a new zone, in either direction. The sign-up is just below.
One email. Only when the needle moves to a new zone.
No newsletter. No noise. We write to you when the reading changes zone.