Advertiser disclosure
Gold $4,363.10 +27.2% on the year Silver $64.50 +66.2% on the year

11 Sept 2026 live spot

Market data · Gold vs S&P 500 since 1968

Gold vs the S&P 500

Gold and the S&P 500 have returned 8.5% and 7.8% a year respectively since April 1968 on price alone, and the index returns 10.8% a year once its dividends are reinvested. One unit of the index cost 1.68 ounces of gold at the August 2026 close. A scheduled job rebuilds both series each month from the LBMA fixes and the Shiller and FRED index data, and it publishes the dividend adjustment rather than hiding it.

Both figures from the same month end, 2026-08-31 701 monthly observations since April 1968

Ounces of gold to buy one unit of the index

1.68

August 2026 close. Gold $4,562.75, index 7,686.14

Where that sits on 58 years

1.68
0.17January 1980 low 1.6558-year average 5.40July 1999 high

Read this before the chart

The index series here counts no dividends, which is worth 3.0 percentage points a year and reverses the answer. Every comparison below prints the price figure and the dividends-reinvested figure together, and the method is set out in full further down.

The numbers

Gold and the S&P 500 side by side

Both closes come from August 2026, the last month in the record. There is no live feed for the index, and pairing a live gold quote against a four-month-old index level would flatter whichever side happened to have the better four months. The bottom row is the same 58-year race run three ways.

Gold, monthly close

$4,562.75/oz

+33.1% since August 2025

The LBMA afternoon fix on the last trading day of August 2026. Gold pays no income, so its price change is its whole return.

S&P 500 close, price only

7,686

+19.0% since August 2025

The index level, which excludes dividends. With dividends the same twelve months returned 20.3%.

Ounces of gold per index unit

1.68

-10.6% since August 2025

The index divided by the gold price, both from the same month end. A falling number means gold is gaining on stocks, and the 701-month average is 1.65.

Gold since April 1968

8.5% a year

$10,000 became $1,166,944

Annualised over 58 years of LBMA monthly fixes. The sparkline is the rolling ten-year annualised return, which is how much that average moves around.

S&P 500 since 1968, price only

7.8% a year

$10,000 became $803,401

The figure most gold sites compare against, and the reason they conclude gold won. It counts no dividends at all.

S&P 500 since 1968, dividends reinvested

10.8% a year

$10,000 became $3,936,080

Measured rather than assumed. Built from Shiller's own dividend column reinvested month by month, which covers 699 of the 701 months here.

Sources: LBMA Gold Price (PM London auction, USD/oz). Shiller (ie_data.xls) S&P Composite for the deep history, FRED SP500 month-end close from the handover month. Dividend yield from Robert Shiller's ie_data.xls, retrieved 2026-09-09

The long record

Gold vs S&P 500 chart since 1968

$10,000 put into each in April 1968, plotted on a log axis so the first thirty years are readable rather than flat against the baseline. The gold line and the dividend-adjusted index line cross four times, and where they finish is the answer to the question this page exists for.

Growth of $10,000, April 1968 to August 2026

Gold against the S&P 500 with dividends reinvested

Index, dividends in

$3,936,080

Gold $1,166,944

The dashed rule is the $10,000 you started with. Gold spent from 1980 to 2002 below the index line and from 1974 to 1980 far above it, which is why the start date does more work in this comparison than the assets do.

$1,000$10,000$100,000$1,000,000$10,000,000Gold's 1980 peak$167,008The index's 1999 peak$65,371Starting stake $10,000
197019801990200020102020
GoldS&P 500, dividends reinvested
A $10,000 stake made in April 1968 grew to $1,166,944 in gold and $3,936,080 in the S&P 500 with dividends reinvested month by month from Shiller's dividend column. The same stake in the price index alone, with no dividends, reached $803,401. Log axis. Sources: LBMA monthly fixes, Shiller and FRED for the index.

Source: LBMA Gold Price (PM London auction, USD/oz) and Shiller (ie_data.xls) S&P Composite for the deep history, FRED SP500 month-end close from the handover month. Retrieved 2026-09-09

Download the full dataset as CSV

Gold, 58 years

$1,166,944

8.5% a year

Index, price only

$803,401

7.8% a year, gold ahead by +0.69

Index, dividends reinvested

$3,936,080

10.8% a year, gold behind by −2.29

Method, and the correction most sites skip

Why most gold vs stocks comparisons are wrong

The S&P 500 series behind every chart on this page is a price index, which records what the index level did and leaves out every dividend the 500 companies paid across 58 years. Dividends are the largest single component of long-run US equity return, so running the comparison that way hands gold a win it did not earn.

Measured on price alone, gold beat the index by +0.69 percentage points a year over the full record. Put the dividends back and the index wins by 2.29 points a year instead, and the $10,000 that grew to $1,166,944 in gold grew to $3,936,080 in the index. That is the same 58 years and the same two assets, and the only thing that changed is whether the dividends were counted.

Robert Shiller's spreadsheet carries a monthly dividend column alongside the price column this chart is built from, and this page now reads it. Each month's dividend is one twelfth of the annualised figure, reinvested at that month's index level, which is the standard construction for this dataset. That covers 699 of the 701 months here, because the dividend column stops in June 2026 while the price index runs to August 2026. For the 2 months after it we carry the last observed yield of 1.09% forward against the actual index level, so the estimate tracks the market rather than going stale.

Until September 2026 this page did something weaker, and it is worth saying what, because a correction nobody can see is indistinguishable from an edit. It compounded a flat 2.81% average yield on the price index, which was the best available answer while no reinvested series existed anywhere in our data. Measuring instead of assuming moved the index's annualised return by 0.05 points, from 10.8% to 10.8%, and gold's shortfall from 2.34 points a year to 2.29.

The interesting part is where the flat rate went wrong, because across the whole record it barely did. It was out by 0.05 points a year over 58 years and by 109,634 dollars on the $10,000 stake. Inside a single decade it was out by far more, because a yield that averaged 4.20% through the 1980s and 1.47% so far in the 2020s does not average usefully. Applying one rate to both understated the 1980s by roughly 1.6 percentage points a year, which over that decade is most of the argument.

Average S&P 500 dividend yield by decade, from Shiller's dividend column
Decade Average yield
1960s 3.11%
1970s 4.00%
1980s 4.20%
1990s 2.44%
2000s 1.79%
2010s 1.98%
2020s 1.47%

Computed from Robert Shiller, ie_data.xls, Yale University, over the months this page covers. The 1960s row starts April 1968 and the 2020s row ends June 2023

What replacing the estimate with the measurement moved
Figure Estimated Measured
Index return a year 10.8% 10.8%
$10,000 became $4,045,714 $3,936,080
Gold's gap a year −2.34 pts −2.29 pts
Gold wins, 581 windows 180 185

Estimated compounds a flat 2.81% yield on the price index. Measured reinvests Shiller's monthly dividend at each month-end level

Dividends are not the only flattering omission in a long-run gold comparison. Our inflation-adjusted gold price tracker sorts all 687 twelve-month windows in the record by the inflation inside them, and gold's median real return in the highest-inflation quarter was -1.0% against 3.5% across the calmer three quarters. That is the same species of correction as this one, applied to the hedging claim rather than to the return claim.

One more thing the chart cannot show. The index figures up to September 2016 are Shiller's monthly averages of daily closes rather than month-end closes, while the gold figure beside them is a single fix on the last trading day of the month. From October 2016 onward both sides are month-end. The mismatch is worth a few tenths of a percent in any single month and washes out over 58 years. It is still the reason a reading here can differ slightly from one taken off a chart built entirely on closes.

Drivers

What decides whether gold or stocks wins

The monthly returns of these two assets have a correlation of 0.04 over 58 years, which is close enough to zero that neither one predicts the other. What they do instead is take turns, and the turns are long enough that the start date of any comparison does more work than either asset does.

Every fall of 19% or more in the index since 1968, and gold over the same months

Gold during S&P 500 bear markets

Gold rose through 5 of the 8 times the index fell 19% or more, and the largest gap came in the 1973 to 1974 oil shock, when the index lost 43.4% over 23 months while gold gained 182.6%. The three gold lost are in the table underneath the chart.

S&P 500, priceGold
-29.0%%-15.3%%-43.4%%182.6%%-19.4%%-44.7%%-26.8%%6.8%%-43.7%%25.4%%-50.8%%16.1%%-20.0%%6.2%%-24.8%%-7.4%%
19681973198019872000200720192021
Peak-to-trough change in the S&P 500 on monthly closes and gold's change over the identical months, for all 8 falls of 19% or more since 1968. Gold rose through 5 of them. Measured on price for both, so the equity bars understate a dividend-reinvesting holder's outcome by roughly a fifth of a percentage point a month.
The index's peak, its trough, gold over the same months, and when the index got back
Peak Trough Months S&P 500 Gold Index recovered
December 1968 June 1970 18 −29.0% −15.3% March 1972
January 1973 December 1974 23 −43.4% +182.6% July 1980
November 1980 July 1982 20 −19.4% −44.7% November 1982
August 1987 December 1987 4 −26.8% +6.8% July 1989
August 2000 February 2003 30 −43.7% +25.4% May 2007
October 2007 March 2009 17 −50.8% +16.1% March 2013
December 2019 March 2020 3 −20.0% +6.2% July 2020
December 2021 September 2022 9 −24.8% −7.4% December 2023

Sources: LBMA monthly fixes and Shiller and FRED index levels, measured on price for both assets. Past performance does not predict future results

Gold rose through 5 of those 8 falls, which is the diversification case and it is a real one. The three it lost are worth as much attention as the five it won, because they share a cause. Gold fell 44.7% alongside the index between November 1980 and July 1982, while Paul Volcker held the federal funds rate near 20%. It fell 9.2% alongside it again in 2022, when the Federal Reserve raised rates at the fastest pace since then. An asset that pays nothing gets repriced when cash starts paying something, and that repricing hits gold and equities at the same time rather than at opposite times.

That table measures the index's falls rather than gold's. Gold has had declines of its own on the same scale and has taken longer to climb out of several of them, and the full record of every drawdown over 20% since 1968 sits on our gold price history tracker, which is the nominal series underneath the rolling windows further down this page.

Volatility over the full record

Gold has run at 19.0% annualised against the index's 13.2%, so the asset most often described as the safe half of a portfolio has been the more volatile half of this pair for 58 years. Over the last ten years the two have converged to within 0.0 points of each other.

What the correlation buys you

A correlation of 0.04 means gold has diversified an equity portfolio genuinely rather than nominally, since a hedge that moves with the thing it hedges is not a hedge. What it does not buy is return, and the two studies below separate those two claims because the sales pitch usually runs them together.

One number for the whole comparison

The S&P 500 priced in ounces of gold

Dividing the index level by the gold price gives the ounces of gold it takes to buy one unit of the index, which collapses both assets into a single line. It ran from 0.17 in January 1980 to 5.40 in July 1999, a range of 32 times. It is also the only figure on this page that dividends do not touch, because it compares two prices.

Index level divided by the gold price, monthly

Ounces of gold to buy one unit of the S&P 500

August 2026

1.68

The dashed rule is the 1.65 average across all 701 months. Turn on the shading to see the 8 bear markets in the index, which show up here as sharp drops because the index is the numerator.

0.1110January 1980 low0.17July 1999 high5.4058-year average 1.65
197019801990200020102020
Ounces of gold required to buy one unit of the S&P 500, from April 1968 to August 2026. The record low of 0.17 came in January 1980 at gold's peak, and the record high of 5.40 came in July 1999 at the top of the dot-com market. Log axis, because a linear one renders the whole of the 1980s as a flat line. Sources: LBMA monthly fixes, Shiller and FRED.

Where the 701 monthly readings landed

Distribution of the gold to S&P 500 ratio

8816594102137582037
0.0-0.50.5-1.01.0-1.51.5-2.02.0-2.52.5-3.03.0-4.04+
Count of monthly readings in each band of the ratio, April 1968 to August 2026. The dark bar is the band the August 2026 reading of 1.68 falls into.

The busiest band sits between half an ounce and one, which covers 165 of the 701 months and almost all of the 1980s and 1990s. Today's reading of 1.68 sits in the 1.5 to 2.0 band and at the 55th percentile of the whole record, which is another way of saying it is telling nobody anything at the moment.

What we found in the data

What happened after gold's 1980 peak and the index's 1999 peak

This ratio has reached a genuine extreme twice in 58 years, and both times it called the next decade correctly. In January 1980 a unit of the index cost 0.17 ounces, and over the ten years that followed gold lost 4.4% a year while the index returned 16.6% a year with dividends reinvested. In July 1999 a unit cost 5.40 ounces, and the following decade handed gold 13.9% a year against the index's -2.1%.

Two observations is not a strategy and we are not offering it as one. What the record supports is narrower, which is that the reading carried information only at the ends of its range. Those ends lasted about a month each across 58 years, which puts today's reading a long way from either of them.

Ten years after January 1980, gold behind by

21.1 pts a year

Ten years after July 1999, gold ahead by

16.0 pts a year

Gold's annualised lead over the index across the ten years after each extreme

Against the price indexAgainst the index with dividends
-16.3 pts pts-21.1 pts pts17.7 pts pts16.0 pts pts-14.3 pts pts-16.6 pts pts
January 1980July 1999August 2011
Gold's annualised return minus the S&P 500's, over the ten years following each of the three turning points, shown against the price index and against the index with dividends reinvested. Dividends move every bar in the same direction and by roughly 3 points a year, and they never change the sign.
Annualised returns over the five and ten years after each turning point
From Ratio Window Gold Index, price Index, dividends in Gold's lead
January 1980 0.17 5 years −14.0% +9.1% +14.6% −28.7 pts
10 years −4.4% +11.9% +16.6% −21.1 pts
July 1999 5.40 5 years +8.9% −4.3% −2.9% +11.8 pts
10 years +13.9% −3.8% −2.1% +16.0 pts
August 2011 0.65 5 years −6.3% +12.9% +15.2% −21.5 pts
10 years +0.0% +14.3% +16.6% −16.6 pts

Annualised. The dividends-in column reinvests Shiller's monthly dividend at each month-end index level. No dealer premium, storage or tax applied to the gold column, so a real gold IRA outcome is worse than the figure shown

Every window, not a chosen one

How often gold beat the S&P 500 over 10 years

Gold beat the index in 185 of the 581 rolling ten-year windows since April 1968, which is 32% of them, with a median shortfall of 6.5 percentage points a year. Measured against the price index instead, gold wins 238 of them, so 53 windows give a different answer depending on whether anyone counted the dividends.

Share of ten-year windows gold won, by the decade the window started in

Gold's win rate against the S&P 500, by starting decade

The decade a ten-year window began in decided the answer far more reliably than anything about gold did. Windows that opened in the 1980s went to the index every single time, and windows that opened in the 2000s went to gold in 81 cases out of 120 even after dividends.

Against the price indexAgainst the index with dividends
100%%100%%66%%42%%0%%0%%30%%25%%78%%68%%11%%4%%
1960s1970s1980s1990s2000s2010s
The share of rolling ten-year windows in which gold's annualised return beat the S&P 500's, grouped by the decade each window began in. The 1960s group covers only the 21 months from April 1968 and the 2010s group ends where a full ten-year window runs out of data. Every window that started in the 1980s went to the index, and every window that started in the 2010s goes to the index once dividends are counted.
All 581 rolling ten-year windows, grouped by the decade they started in
Windows starting in Windows Gold wins, price Gold wins, dividends in Median gold Median gap
1960s 21 21 21 +18.9% +15.5 pts
1970s 120 79 50 +11.6% −3.0 pts
1980s 120 0 0 −1.9% −17.7 pts
1990s 120 36 30 +1.1% −9.3 pts
2000s 120 93 81 +10.8% +2.9 pts
2010s 80 9 3 +3.9% −8.5 pts
All windows 581 238 185 −6.5 pts

Overlapping windows, so these are not independent observations. Past performance does not predict future results

Two rows carry most of the information. Every one of the 120 windows that started in the 1980s went to the index, and it was not close, with a median gap of 17.7 percentage points a year. The 2010s look almost as one-sided in the other direction once dividends are counted, because gold won 9 of those windows on price and 3 of them on total return.

Stretching the window makes the dividend correction matter more rather than less. Over five-year windows gold wins 263 of 641 on total return against 287 on price, a difference of 24 windows. Over twenty-year windows the same comparison runs 115 of 461 against 216, because twenty years of reinvested dividends compound into a gap that no plausible move in the gold price closes. Anyone quoting a long-run figure without saying which index they used is quoting a number that could go either way.

Long history

Gold vs the S&P 500 by decade, 1968 to today

Gold has won 3 of the 7 decades in the record, and lost the other 4. Both of the decades gold won before this one were followed immediately by a decade it lost badly, which is the pattern anyone extrapolating from the 2020s should have in front of them.

GoldS&P 500 with dividends
-6.1%%0.2%%31.1%%6.0%%-4.9%%17.0%%-3.5%%18.4%%14.5%%-0.7%%3.5%%13.5%%17.4%%15.8%%
1960s1970s1980s1990s2000s2010s2020s
Annualised return for each calendar decade, gold against the S&P 500 with dividends reinvested. The 1960s bar covers only April 1968 to December 1969 and the 2020s bar covers January 2020 to August 2026, so both are partial. Gold won the 1970s, the 2000s and so far the 2020s.

The 1970s and the 2000s are the two decades that every gold advertisement is built from, and the returns in them were genuinely enormous, at 31.1% a year and 14.5% a year. Both were followed by a decade in which gold returned -4.9% and 3.5% a year while the index compounded at 17.0% and 13.5%. Nobody knows whether the 2020s will rhyme, and this table cannot tell you, but it does say that a ten-year gold run has never once been followed by another one in this record.

Timeline

Gold vs stocks timeline, with the ratio at every turn

Every entry carries the two prices at that month and the ratio they made, plus where the ratio stood twelve months later, so the story and the numbers cannot drift apart.

  1. August 1971

    Nixon closes the gold window

    The dollar stops converting into gold at $35 an ounce, so for the first time since 1934 both assets on this page are priced by buyers and sellers rather than by statute. A unit of the index cost 2.39 ounces that month and 1.66 ounces a year later, which was the start of the fastest re-rating anywhere in the record.

    Ounces per unit 2.39 Gold $40.65 Index 97.24 Twelve months on 1.66

  2. December 1974

    The oil shock bear market bottoms

    The index had fallen 43.4% from its January 1973 peak while gold rose 182.6% over the same 23 months. That is the widest gap between the two anywhere in this record. The index did not close back above its 1973 level until July 1980, and gold was the better asset for every month in between.

    Ounces per unit 0.36 Gold $186.50 Index 67.07 Twelve months on 0.63

  3. January 1980

    Gold peaks and the index costs 0.17 ounces

    Gold closed January at $653 after a year in which US consumer prices rose 13.3%, and a unit of the index cost 0.17 ounces, which is the cheapest reading in 58 years. Over the following decade gold lost 4.4% a year while the index returned 16.6% a year with dividends reinvested. That is a gap of 21.1 percentage points a year against anyone who read the extreme as a reason to hold metal.

    Ounces per unit 0.17 Gold $653.00 Index 110.90 Twelve months on 0.26

  4. October 1987

    Black Monday

    The index fell 26.8% between its August reading and its December low while gold gained 6.8% over those same four months, which is the behaviour that puts gold in a portfolio in the first place. The ratio dropped from 0.73 in August to 0.60 in October, then spent the next twelve years going the other way until it reached 5.40.

    Ounces per unit 0.60 Gold $468.80 Index 280.20 Twelve months on 0.67

  5. July 1999

    The index reaches its record price in gold

    A unit of the index cost 5.40 ounces, which is 32 times what it cost in January 1980 and the highest reading in the record. Gold sat at $255.60, two months before fifteen European central banks signed the Washington Agreement and capped their own sales. The ten years that followed handed gold 13.9% a year against the index's minus 2.1% with dividends counted.

    Ounces per unit 5.40 Gold $255.60 Index 1,380.99 Twelve months on 5.32

  6. September 2008

    Lehman fails and everything liquid is sold

    Gold and the index both fell in the scramble for dollars that autumn, though the index fell further and kept falling. A unit of it cost 1.38 ounces in September and 0.83 ounces by March 2009. Gold rose 16.1% between the index's October 2007 peak and its March 2009 trough, which is the clearest case in the record of the two assets doing genuinely different things inside one crisis.

    Ounces per unit 1.38 Gold $884.50 Index 1,216.95 Twelve months on 1.05

  7. August 2011

    Gold's second peak

    Gold closed August at $1,813.50 against an index at 1,185.31, so a unit cost 0.65 ounces, the cheapest reading since November 1988. Gold then returned nothing at all over the following ten years, while the index returned 16.6% a year with dividends reinvested. That is the second time this ratio marked a top in gold rather than a bargain in stocks.

    Ounces per unit 0.65 Gold $1,813.50 Index 1,185.31 Twelve months on 0.85

  8. March 2020

    The Covid liquidation

    The index gave up 20.0% between its December 2019 close and the end of March while gold held near $1,608.95, and the ratio bottomed at 1.61 before the recovery in equities ran faster than the recovery in anything else. A year later a unit of the index cost 2.35 ounces, which is a 46% swing inside twelve months and a reminder of how quickly this number moves.

    Ounces per unit 1.61 Gold $1,608.95 Index 2,584.59 Twelve months on 2.35

  9. January 2026

    Gold's run takes the ratio to 1.39

    Gold reached $4,981.85 in January against an index at 6,939.03, so a unit cost 1.39 ounces, which is the lowest reading since February 2014. By the April 2026 close gold had given back 7.4% while the index added 3.9%, which put the ratio at 1.56 and within a rounding error of its 58-year median at that point.

    Ounces per unit 1.39 Gold $4,981.85 Index 6,939.03 Twelve months on not yet

Our view

Gold IRA Digest's perspective

Every entry carries the day it was written, because an undated opinion sitting under a data table reads as current no matter how old it has become. An old note here stays visible as an old note rather than quietly passing for this month's thinking.

· reading the April 2026 close · latest

Almost every page that compares these two assets over the long run puts a price index on the stock side. That drops the S&P's compound return by 3.0 percentage points a year and hands gold a 58-year win it did not earn. Until today we filled the missing half with an estimate, and it now reads Shiller's own dividend column instead, which moved the index's annualised return by 0.04 points and gold's shortfall from 2.19 points to 2.15. The estimate was close across the whole record and wrong by as much as 1.6 points inside a single decade, which is the case for measuring rather than assuming even when the assumption looks harmless.

· reading the April 2026 close

At 1.56 ounces to buy a unit of the index, this ratio is sitting less than a tenth of a standard deviation below its 58-year mean, which means it is not telling anyone anything at the moment. The two readings worth acting on were 0.17 in January 1980 and 5.40 in July 1999. The second of those sat more than three standard deviations from the mean, against less than a tenth of one today.

· reading the April 2026 close

Gold has returned 18.6% a year so far this decade against 15.4% for the index with dividends, so the 2020s are on track to be only the third decade in the record that gold has won. Both of the previous two reversed in the decade that followed, which is worth holding in mind before treating six good years as a trend.

Written by the Gold IRA Digest Editorial Team. Opinion, labelled as opinion, and never a recommendation to buy or sell either asset. Each note names the month its figures came from, so a later data refresh cannot quietly restate an old view under its original date.

What is scheduled

What to watch next

Windows are months rather than exact dates, because the later release dates are not published yet and inventing one would be the same mistake as inventing a verification date.

  • Monthly, mid-month US CPI release Real yields, and through them the opportunity cost of holding an asset that pays nothing against one that pays a dividend Bureau of Labor Statistics
  • Eight times a year FOMC rate decision The same channel but faster, because gold reprices on the projections more than on the decision itself Federal Reserve
  • Quarterly, from mid-January S&P 500 earnings season The index half of this ratio, since the index price is a multiple of earnings that get revised four times a year S&P Dow Jones Indices
  • Sep 2026 Our monthly data refresh Ours Every figure on this page, including the 58-year averages and the rolling window counts Gold IRA Digest
  • Shipped 9 Sep 2026 Our total-return rebuild, done Ours Nothing further. This page used to estimate the S&P's dividends from a flat 2.81% average yield, and it now reads Shiller's own dividend column for 663 of the 697 months, which moved gold's shortfall from 2.18 points a year to 2.15 Gold IRA Digest
  • When Shiller next publishes Shiller dividend column extended past June 2023 The 34 months at the end of this record that currently carry the last observed yield forward rather than a measured dividend Robert Shiller, Yale University
  • Jan 2027 S&P Dow Jones Indices annual dividend summary The yield assumption sitting underneath every adjusted figure on this page S&P Dow Jones Indices
  • Feb 2027 World Gold Council full-year demand trends Central bank buying, which has been the largest single change in gold demand since 2022 World Gold Council

What this means if you are weighing a gold IRA

What gold vs stocks means for a gold IRA

Every figure above prices the metal alone. A gold IRA adds a setup fee, an annual custodian fee and a depository storage fee on top, and it adds a dealer premium over spot at purchase that an index fund has no equivalent for. On a $50,000 purchase a premium of 5% against one of 12% is a $3,500 difference before the account has done anything. That is more than the 2.29 points a year this whole comparison turns on would produce over a short holding period. Decide the allocation on the diversification argument in the drawdown table, then decide the provider on the fee schedule, because those are two separate decisions and only one of them is settled by a 58-year chart.

The comparison above also cannot settle which form of gold you hold, and that choice changes the cost more than the chart does. A gold ETF and a mining stock both sit inside an ordinary brokerage IRA at brokerage cost, while bullion needs a self-directed IRA, a custodian and a depository, and the metal has to clear the 99.5% purity rule before a custodian will accept it. Everything on this page prices the metal at the spot price, so an ETF holder tracks that line minus an expense ratio and a mining stock does not track it at all.

Gold's shortfall, 58 years

2.29 pts

A year, against the index with dividends reinvested

Bear markets gold rose through

5 of 8

Falls of 19% or more in the index since 1968

Typical dealer premium

5% to 12%

Charged over spot on IRA-eligible gold coins at purchase

Common questions

Gold vs stocks questions

Has gold beaten the S&P 500 since 1968?
No, not once the dividends the index paid are counted. The S&P returned 10.8% a year against gold's 8.5%, and $10,000 invested in April 1968 grew to $3,936,080 in the index against $1,166,944 in gold. On price alone gold does win, by +0.69 percentage points a year, and that is the comparison most gold sites publish. Dividends are the whole difference between those two answers.
Why does the dividend adjustment matter so much?
Dividends compound, which is what turns a modest yield into most of the difference. The S&P's dividend yield averaged 2.8% a year across this record in Robert Shiller's data, and reinvesting it month by month lifts the annualised return by 3.0 percentage points rather than 2.8, since the shares those dividends buy go on to pay dividends of their own. Over 58 years that turns $803,401 into $3,936,080.
How often has gold beaten stocks over ten years?
In 185 of the 581 rolling ten-year windows since April 1968, which is 32% of them, measured against the index with dividends reinvested. Measured against the price index alone gold wins 238 of them, so 53 windows change their answer depending on whether dividends are counted. Over twenty-year windows gold wins 115 of 461.
Does gold protect a portfolio when stocks fall?
Yes, usually, though not in every case. The index has had 8 falls of 19% or more on monthly closes since 1968, and gold rose through 5 of them. It fell alongside stocks in 1968 to 1970, in 1980 to 1982 when Paul Volcker raised rates to 20%, and again in 2022. The monthly returns of the two assets correlate at 0.04 over the full record. That is close enough to zero that neither one tells you much about the other.
What is the gold to S&P 500 ratio right now?
1.68 ounces of gold to buy one unit of the index, using the August 2026 closes for both. That sits just under the 701-month average of 1.65 and almost exactly on the median of 1.52. The record low is 0.17 in January 1980 and the record high is 5.40 in July 1999.
How should gold and stocks be split inside an IRA?
That is an allocation question rather than a winner question, and the record above answers only the second one. Gold lost to the index over the full 58 years on total return and beat it inside three of the seven decades, so the case for holding any is diversification rather than growth. A gold IRA also charges a setup fee, an annual custodian fee and a storage fee that an index fund does not, and those come out of the gold side of the comparison before you start.
How often does this page update?
The monthly series are rebuilt from the LBMA fixes and the Shiller and FRED index data, and every figure here is computed at build time from those files rather than written by hand. The last observation in both series is August 2026. There is no live quote for the index, so this page carries no live number at all and dates everything to the month it came from.

How this page is built

Gold is the LBMA afternoon fix in US dollars per troy ounce at each monthly close, 701 observations from April 1968. The S&P 500 is Robert Shiller's S&P Composite spliced to the FRED SP500 series, matched to gold on the same month-end dates. Every figure here is computed at build time from those two files, so the numbers are in the HTML before any script runs and every chart works with JavaScript turned off.

The index series is price only, so total return comes from a third file. Shiller's ie_data.xls also publishes a monthly dividend per index unit, and we reinvest one twelfth of that annualised figure at each month-end index level to build the total-return series every adjusted number here reads. It is measured through June 2026, which covers 699 of the 701 months on this page. For the 2 months after it, where Shiller has published prices but not yet dividends, we carry the last observed yield of 1.09% forward against the actual index level.

Two further limits are worth knowing before anyone reuses these figures. The index values up to September 2016 are Shiller's monthly averages of daily closes rather than month-end closes. Gold beside them is a single fix on the last trading day, so the two sides only line up properly from October 2016. Returns everywhere on this page carry no fees, no dealer premium, no storage cost and no tax. A real gold IRA outcome is worse than the gold column shown by the size of all four.

Sources

Full dataset available as CSV, including the dividend-adjusted column, so anyone who disagrees with our yield assumption can rebuild it. Corrections go to the editorial team and are logged publicly.

Gold IRA Digest is an independent publisher. This page is information, not investment advice, and nothing on it is a recommendation to buy or sell gold, stocks or any fund. Past performance does not predict future results. Some links on this site earn us a commission, which never changes what we publish about a company.