The short answer
From 30 July 2002 to 29 October 2026, gold far outperformed US Treasuries: COMEX gold futures rose 1,227% (11.3% a year), against 130% (3.5% a year) for long-term Treasuries (TLT, 20+ year) and 128% (3.5% a year) for intermediate Treasuries (IEF, 7–10 year), both with interest reinvested. Gold won 16 of the 23 full calendar years against each.
Bonds were not always the calmer choice. Intermediate Treasuries had about a third of gold’s volatility (6.4% vs 17.5% a year), but long Treasuries fell 48.4% from 2020 to 2023 and had not recovered by October 2026. The real yield link is real but loose: week to week, a 10 basis point rise in the 10-year real yield has gone with about a 0.5–0.65% fall in gold, yet since 2022 gold has risen sharply while real yields stayed high.
Returns, volatility and drawdowns
| Measure (30 Jul 2002 – 29 Oct 2026) | Gold (COMEX futures) | Long Treasuries (TLT, 20+ yr) | Intermediate Treasuries (IEF, 7–10 yr) |
|---|---|---|---|
| Total change | +1,227% | +130% | +128% |
| Annualised return | +11.3% | +3.5% | +3.5% |
| Annualised volatility (weekly) | 17.5% | 13.4% | 6.4% |
| Worst peak-to-trough fall | −44.4% | −48.4% | −23.9% |
| Peak → trough | 22 Aug 2011 → 17 Dec 2015 | 4 Aug 2020 → 19 Oct 2023 | 4 Aug 2020 → 19 Oct 2023 |
| Recovered by 29 Oct 2026? | Yes, 23 Jul 2020 | No | No |
| Correlation of weekly returns with gold | – | 0.15 | 0.21 |
The two bond funds ended with almost the same total return by very different paths. IEF’s 7–10 year bonds moved gently; TLT’s long bonds swung almost as much as gold and suffered the deepest fall in the table when yields surged from 2020 to 2023. Gold’s own worst fall (2011–2015) took almost nine years to recover.
Year by year
| Year | Gold | TLT (20+ yr) | IEF (7–10 yr) |
|---|---|---|---|
| 2003 | +19.6% | +1.6% | +5.3% |
| 2004 | +5.2% | +8.0% | +4.0% |
| 2005 | +18.2% | +9.3% | +2.8% |
| 2006 | +22.8% | +0.7% | +2.5% |
| 2007 | +31.4% | +10.3% | +10.4% |
| 2008 | +5.8% | +34.0% | +17.9% |
| 2009 | +23.9% | −21.8% | −6.6% |
| 2010 | +29.8% | +9.0% | +9.4% |
| 2011 | +10.2% | +34.0% | +15.6% |
| 2012 | +7.0% | +2.4% | +3.7% |
| 2013 | −28.2% | −13.4% | −6.1% |
| 2014 | −1.5% | +27.3% | +9.1% |
| 2015 | −10.4% | −1.8% | +1.5% |
| 2016 | +8.5% | +1.2% | +1.0% |
| 2017 | +13.6% | +9.2% | +2.6% |
| 2018 | −2.1% | −1.6% | +1.0% |
| 2019 | +18.9% | +14.1% | +8.0% |
| 2020 | +24.7% | +18.2% | +10.0% |
| 2021 | −3.5% | −4.6% | −3.3% |
| 2022 | −0.1% | −31.2% | −15.2% |
| 2023 | +13.4% | +2.8% | +3.6% |
| 2024 | +27.5% | −8.1% | −0.6% |
| 2025 | +64.4% | +4.2% | +8.0% |
| 2026 (to 29 Oct) | −7.1% | −3.2% | −2.2% |
The years that stand out:
- 2008: long Treasuries returned +34.0% and gold +5.8%. In the acute phase of a financial crisis, Treasuries were the stronger haven.
- 2013: gold fell 28.2%, its worst year in the table, and bonds also lost money as yields rose.
- 2022: long Treasuries lost 31.2% and intermediate ones 15.2%, while gold was flat (−0.1%). In a year of rapidly rising rates and high inflation, gold held its value much better than bonds.
- 2024–2025: gold returned +27.5% and +64.4%, while long Treasuries lost 8.1% in 2024 and the two bond funds gained 4.2% and 8.0% in 2025.
Gold and intermediate bonds both lost money in the same year only three times (2013, 2021, 2022). In the other three losing years for gold (2014, 2015 and 2018), IEF rose, so intermediate bonds offset half of gold’s losing years.
Correlation: weak and positive
| Period | Weeks | Gold vs IEF: correlation of weekly returns |
|---|---|---|
| 2002–2007 | 282 | 0.14 |
| 2008–2012 | 261 | 0.04 |
| 2013–2019 | 365 | 0.44 |
| 2020–2021 | 105 | 0.34 |
| 2022–2023 | 104 | 0.39 |
| 2024–2026 | 148 | 0.13 |
On average gold and Treasuries moved slightly together (weekly correlation 0.21 with IEF, 0.15 with TLT), and in 2013–2019 the link was strongest (0.44). That surprises people who think of gold as the “opposite” of bonds. Both tend to gain when yields fall, because a lower yield raises bond prices and lowers the cost of holding a non-yielding asset like gold.
The real-yield link
The real yield is the return on an inflation-protected Treasury (TIPS), roughly the nominal yield minus expected inflation. It is the opportunity cost of holding gold, and our gold vs real yields and the dollar tracker charts it every day.
Across 1,243 weeks from January 2003 to October 2026, the correlation between gold’s weekly return and the weekly change in the 10-year real yield was −0.28: when real yields rose, gold tended to fall. It was negative in 22 of 24 years, with 2008 roughly zero and 2025 the clear exception (+0.31).
The size of the effect, estimated by regression on weekly data:
| Period | Weeks | Gold move per +10 bp in the 10-year real yield | Correlation |
|---|---|---|---|
| 2003–2021 | 991 | about −0.65% | −0.29 |
| 2022–Oct 2026 | 252 | about −0.51% | −0.25 |
So the week-to-week link survived after 2022, only a little weaker. What broke was the level relationship:
| Date | 10-year real yield (TIPS) | Gold (GC=F close) |
|---|---|---|
| 2 Jan 2003 | 2.43% | $346.1 |
| 31 Dec 2012 | −0.67% | $1,674.8 |
| 6 Aug 2020 | −1.08% | $2,069.4 |
| 31 Dec 2021 | −1.04% | $1,828.6 |
| 21 Oct 2022 | 1.69% | $1,656.3 |
| 19 Oct 2023 | 2.49% | $1,980.5 |
| 31 Dec 2024 | 2.24% | $2,641.0 |
| 31 Dec 2025 | 1.93% | $4,341.1 |
| 28 Oct 2026 | 2.67% | $4,051.6 |
In October 2022 a 1.69% real yield went with gold at $1,656. Four years later a higher real yield, 2.67%, went with gold at $4,052. Something other than real yields lifted gold’s trend; our data cannot say what, and we do not want to guess. For traders, the lesson is to use real yields for direction of change (a sharp weekly rise is a headwind) rather than as a fair-value model for the price.
How we built this comparison
- Gold: COMEX gold futures (GC=F) daily closes from Yahoo Finance.
- Treasuries: two exchange-traded funds that hold US Treasury bonds, TLT (20+ year maturities) and IEF (7–10 year), using Yahoo Finance’s adjusted closes, which reinvest interest distributions and are after fund fees. Both start on 30 July 2002, so that is our start date.
- Real yield: the 10-year Treasury inflation-indexed yield, FRED series DFII10, daily from January 2003.
- Method: total and annualised returns from daily closes; volatility from weekly returns × √52; drawdowns from daily closes; correlations and regression slopes on weekly (last close of the week) returns and yield changes.
Limitations
- Funds, not bonds. ETFs hold a rolling basket of bonds and charge a fee; a single bond held to maturity returns its yield regardless of price swings.
- Gold futures include roll effects and exclude the interest a futures holder earns on collateral.
- One period: 2002–2026 includes the end of a 40-year bond bull market and a large rise in yields from 2020; another period could look very different.
- Correlation is not causation, and a weekly regression slope is an average across very different weeks.
What traders can do with this
- Watch the weekly change in real yields, not the level. A fast rise is still a headwind for gold, even in 2026. The real yields tracker shows it.
- Don’t assume bonds and gold hedge each other. They often move together because both like falling yields.
- Mark data days. Real yields move most on CPI and jobs-report days; our studies of gold on CPI days and gold on NFP days show how big those days get, and the economic calendar lists the next ones.
- Size from volatility. Gold’s volatility is close to the stock market’s and nearly three times intermediate bonds’; the position size calculator keeps the risk per trade fixed.
More comparisons: gold vs the S&P 500, gold vs real estate and is gold a good inflation hedge?, all on our gold data studies hub.
Follow the macro with other traders
Inside the XAUUSD Global community, members track yields, the dollar and the data calendar together each week. No signals. Your first month is free, then $19.99 a month.
Sources
- FRED: 10-Year Treasury Inflation-Indexed Security, Constant Maturity (DFII10)
- COMEX gold futures (GC=F) and Treasury bond ETFs TLT and IEF (adjusted closes) via Yahoo Finance, downloaded 30 October 2026; all returns, drawdowns, correlations and regressions are our own calculations
Educational content only, not financial advice. Past performance does not predict future returns. Trading gold with leverage carries a high risk of loss.
Frequently asked questions
Has gold beaten bonds?
Since July 2002, by a wide margin. To 29 October 2026, COMEX gold futures rose 1,227% (11.3% a year), while long-term Treasuries (TLT, 20+ year bonds, interest reinvested) returned 130% (3.5% a year) and intermediate Treasuries (IEF, 7–10 year) 128% (3.5% a year). Gold beat each of them in 16 of the 23 full years from 2003 to 2025.
Are bonds safer than gold?
Intermediate Treasuries were much calmer: annualised volatility of 6.4% against 17.5% for gold. Long Treasuries were not: TLT’s worst fall was −48.4%, from August 2020 to October 2023, deeper than gold’s worst fall of −44.4% (August 2011 to December 2015), and by 29 October 2026 neither bond fund had regained its 2020 peak.
Do gold and bonds move together?
Slightly, on average. The correlation of weekly returns since 2002 was 0.21 with IEF and 0.15 with TLT. It ranged from 0.04 (2008–2012) to 0.44 (2013–2019). Both tend to gain when yields fall, which is why they often rise together rather than in opposite directions.
How do real yields affect gold?
Week to week, gold has tended to fall when the 10-year real yield rises. From 2003 to 2021, a 10 basis point weekly rise in the 10-year TIPS yield went with a gold move of about −0.65% on average (correlation −0.29); from 2022 to October 2026 the figure was about −0.51% (correlation −0.25). The link was negative in 22 of the 24 years, but positive in 2025.
Why did gold rise in 2025 while real yields stayed high?
The data show the level link broke down: the 10-year real yield was 1.69% in October 2022 with gold at $1,656 and 2.67% in October 2026 with gold at $4,052. Our data cannot say why; commonly cited factors such as central-bank buying are outside this study. What the data do show is that weekly changes in real yields still mattered, just against a rising trend driven by something else.