MEASURED · 1 OCTOBER 2026

Is gold a good investment? What happened to everyone who bought since 1971

Most answers to this question are a list of pros and cons. We took a different route: we treated every month since August 1971, 661 of them, as a day someone bought gold, and followed what happened to their money. How often were they up after one year, five, twenty? How long did the unlucky ones wait to get their money back? And did buying at a moment like this one turn out better or worse than average?

The short version

• Time helped a lot with the odds of being up, and much less with the odds of beating inflation: 56.9% over ten years, 63.2% over twenty.
• The one thing that stood out from chance was momentum over the next year: after gold rose 30% or more in a year, the following year was up 71.7% of the time, against 61.0% for all months. It did not last beyond that year.
• Gold’s price after inflation is near its highest ever. That has not been a reliable warning sign: random price paths produce a gap as large about half the time.

How often has gold made money?

The table takes every month from August 1971 as a purchase date and checks the price a set number of years later. A one-year hold was up 61.0% of the time; a twenty-year hold 87.6%. Beating inflation is a higher bar, and time cleared it far less often: 54.7% over one year, 63.2% over twenty.

Monthly average prices, 661 months from August 1971 to August 2026. “Typical year” is the median yearly rate; “worst outcome” is the total change over the whole hold. Before costs, storage and tax.

The worst outcomes all start in 1980, when gold had just made its biggest run on record. Bought in September 1980 and held twenty years, gold was still 59% down. How that peak formed and unwound is in gold price history.

How long might you wait to break even?

For most buyers, not long. From 54.5% of start months the price was at or above the purchase price the very next month, from 85.2% within a year, and from 96.7% within ten years. The exceptions were long. Every buyer who waited more than ten years bought between January 1980 and February 1990, and the buyer of January 1980 waited until May 2006: 26 years.

After inflation the waits were longer. The January 1980 buyer did not get their purchasing power back until February 2025, 45 years later, and 35 start months, the latest in February 2012, took more than ten years. Four recent start months were still below their purchase price in August 2026.

Is now a good time to buy gold?

Nobody can tell you that, and this page does not try. What the record can do is show what followed months that looked like this one. In August 2026 gold averaged $4,411 an ounce: +31% on a year earlier, 12% below its monthly record of $5,020 in February 2026, and, after inflation, higher than in 98.9% of months since 1971.

For each kind of month below, we compared the next year with the average of all months. Then we built 2,000 imaginary price histories from gold’s own monthly moves, shuffled in year-long blocks, and counted how often pure chance produced a gap at least as large. Long runs of similar months count as one stretch, because they are one piece of history, not many.

“Chance matches it” is for the next-year result: how often random histories produced a gap from the baseline at least as large in the same direction; under 5% is “rare by chance”. A result can be rare by chance and still rest on very few stretches of history. “Real price” fifths are against the whole record since 1971.

After a big year, the next year was usually good too. Following a 12-month rise of 30% or more, gold was up a year later 71.7% of the time, with a typical gain of +22%. Random histories matched that only 0.4% of the time, across ten separate stretches. Months at or near a record show the same tilt, and so, in reverse, do months far below it: over the next year, gold’s trend has tended to continue. It faded quickly. Over five years, buyers after a big year did no better than chance would explain (23.5% of random histories matched it). This condition is true now.

The stretches below show where the edge came from. Most of it is the long bull runs; after the short bursts in 1983, 2008, 2011 and 2020 the next year was much less reliable. You only find out which kind of run you are in afterwards.

A high price after inflation was not a reliable warning. When gold’s real price was in the top fifth of its history, as it is now, the next five years were up only 33.3% of the time against 69.2% for all months. That looks like a clear signal, but it comes from three stretches, and random histories produced a gap at least that large 49.7% of the time. The mirror case, buying when gold was cheapest in real terms, looked even better in hindsight and rests on just three stretches too.

After a 20% fall, the next year was weaker, not stronger. Months at least 20% below the record were followed by a rise a year later only 52.4% of the time, and random histories rarely produced a gap that large (3.7%). But 302 of the 401 months sit in one stretch, Apr 1980 to Dec 2005, so this is mostly one long slump rather than a rule. What buying the first day of a 20% fall did is in every gold price crash since 1971.

Should you buy gold at an all-time high?

Record months were not a bad time to buy over the following year. In the 92 months that set a new monthly record, gold was higher a year later 86.0% of the time, something random histories almost never matched (0.0%). But all of them fall into six stretches, over five years record buyers did no better than chance explains, and the purchase that took longest to recover, January 1980, was a record month too. A record says the trend has been up; it says nothing about how far it can fall when the trend turns. Every record close since 1971, and how long the old ones stood, is in gold’s all-time high.

Lump sum or buy a little each month?

We compared putting money in all at once with spreading it over twelve equal monthly purchases, judged on the same date. Buying all at once came out ahead 59.9% of the time over one year and 57.1% over ten, because gold rose more often than it fell. Monthly buying’s value was in the bad cases: it cut the worst one-year loss from -36% to -26%.

Each row starts from every month with enough data. Results are the total change at the end date, before costs. Whether any month of the year is better for buying is tested in gold seasonality: no month’s direction held up as more than chance.

What buying costs do to your odds

Coins and bars are bought above the market price and sold below it, and funds charge a yearly fee. A one-off cost matters most on short holds. With a total buying-and-selling cost of 6% of the money, the chance of being up after one year falls from 61.0% to 49.2%, worse than a coin flip. Over ten years the same cost moves it only from 75.6% to 71.0%.

2% and 6% are examples of a cheap and an expensive round trip, not quotes. Check your dealer’s buy and sell prices, or your fund’s fee, before you buy.

What this means if you’re deciding

• Decide how long you’ll hold before you buy. After a 2% to 6% cost, a one-year hold was up only 49.2% to 56.4% of the time, close to a coin flip. Over ten years or more it has usually been up, though not always ahead of inflation.
• Size it so a long slump can’t force you out. Of the 541 start months with ten years of history after them, 18 (about one in 30) waited longer than that to break even. Money you may need sooner should not be in gold.
• Don’t read a high price as a sell signal, or a big year as a guarantee. The real-price warning failed the chance test. Momentum after big years passed it, but only for the following year, and mostly thanks to a few long bull runs.
• Monthly buying is insurance, not a return booster. It lost to buying at once 59.9% of the time over a year but softened the worst outcomes.
• Count costs and tax. The IRS taxes gains on collectibles, coins included, at up to 28%, and dealer spreads come off before that.

How gold compared with the stock market over the same years is in gold vs the S&P 500, how it did against inflation in is gold a good hedge against inflation?, and whether silver would have been the better buy in the gold-silver ratio, tested. What real yields, the dollar and the Fed explain of the price is in why is gold going up? This page is history, not personal advice.

How we measured it

Prices: the World Bank’s monthly gold price (the “Pink Sheet”), the London price averaged over each month, from August 1971 to August 2026, 661 months. It is the same series as our price history and inflation studies, checked there against LBMA prices and our own broker’s. Inflation: the US consumer price index (CPI-U, not seasonally adjusted); October 2025 was never published and is filled between September and November.

Every start month with a full holding period counts, so neighbouring periods overlap and the 649 one-year periods are not 649 independent results. That is why the conditions table counts separate stretches and checks each result against 2,000 random histories, rebuilt from gold’s own monthly moves after inflation in 12-month blocks. Monthly averages hide single days, so the worst daily drawdowns were deeper than shown here. None of the figures include storage, fund fees or tax.

Common questions