Real yields and gold:how to read the 10-year TIPS yield

Every macro argument about gold eventually lands on one number: the US 10-year TIPS real yield. It is harder than the dollar index because it measures the opportunity cost of holding gold directly — the real return you give up is either plus 2.6% or minus 1.2%, and that is a number, not a narrative.

Hard does not mean precise. This article skips the storytelling and answers four practical questions: how to read these three rates, why TIPS is the most direct of them, how weak the negative correlation with gold really is on daily data, and whether it is worth watching on 60-second to 5-minute trades. Every number is reproducible: site quotes carry the sampling time, macro series carry the FRED code and reading date.

1. Three rates, one formula

Many people treat "the interest rate" as one thing. There are at least three readings here, and mixing them gives you opposite conclusions:

The 2bp gap between 2.34% and the calculated 2.32% is not an error: T10YIE is a fitted difference between the nominal and TIPS curves, while DFII10 is the quoted yield on one specific TIPS; and the two series are a day apart (10-23 versus 10-22). The formula is good enough in practice, but do not expect two decimal places of precision.

The relationship is one sentence:

Why not use nominal yield minus CPI?

Because CPI is inflation that has already happened, while gold is priced on the future. A real yield built from CPI always lags the market, and in a phase of fast-rising inflation like 2022 it can even give a signal with the wrong sign. The advantage of TIPS is that the principal is adjusted with CPI, so the yield to maturity is exactly the ten-year real return bond traders bid with real money — and you do not have to guess inflation expectations yourself.

That is also why what gold does on CPI day and this article are separate: CPI is about the expectation gap at the moment of release, TIPS is about the ten-year price the market sets. One is event trading, the other is a background variable.

2. The mechanism behind the negative correlation: one opportunity-cost chain

Gold pays nothing. Hold it and the cash flow on your statement is always zero; hold TIPS and on 2026-10-22 you can lock in a 2.64% real return. On a USD 1 million position that is USD 26,400 of real purchasing power a year — that is the opportunity cost of holding gold, and it is an opportunity cost you can put a number on.

The transmission has three links, and each can be broken:

• Relative return. Real yields rise → the risk-free real return rises → non-yielding gold becomes relatively less attractive → the price comes under pressure. This is the main mechanism.
• The dollar. Rising real yields usually come with a firmer dollar (rate differentials pull money in), and gold is priced in dollars, so a dearer dollar mechanically presses on the gold price. This is the other half of the chain described in the dollar index and Treasury yields.
• Risk appetite and discounting. The real yield is the risk-free anchor. When it moves, the relative valuation of every non-yielding asset has to be recalculated — gold, silver, bitcoin included.

Quantify the chain: regressing 247 trading days over the past year, a 1bp rise in the real yield corresponds to an average -0.068% move in gold that day, or in plain terms +10bp ≈ -0.68%. The direction matches the textbook, but R² is only 1.97% — meaning the real yield explains 2% of the daily variance in gold. That number matters, and the rest of this article circles it.

3. How far real yields have travelled since 2021

To understand where 2.64% sits today you need to see where it came from. All five dates below are from FRED and can be checked line by line:

Source: FRED (DFII10 / DGS10), retrieved 2026-10-24. Implied inflation expectation = nominal − real, calculated here, normally within 2–5bp of the T10YIE definition.

From -1.04% at the end of 2021 to 2.95% on 2026-10-05 is a 399 basis point rise; the period low was -1.19% (2021-08-03) and the high 2.95%, a 414bp range. This is the most violent repricing of real yields in forty years.

By the textbook, gold should have been crushed. Reality: over the past year (2025-10-24 to 2026-10-22) the COMEX gold futures close went from 4,137.80 to 4,060.40, only -1.9%, while the 10-year TIPS yield rose from 1.73% to 2.64%, up 91bp. Yields up 91bp, gold essentially unchanged — that is the intuitive version of "the correlation exists but the strength is weak".

4. Measured: how weak the daily correlation really is

These numbers are computed here, with the sample, definitions and instruments all stated, so you can reproduce them:

Gold series: Yahoo Finance · COMEX gold futures continuous contract (GC=F) daily close; real yield: FRED · DFII10. The two series are date-aligned to their intersection with missing values dropped. Retrieved 2026-10-24 05:13 UTC.

Three conclusions, each more counter-intuitive than the last:

• On daily data it is essentially noise. A correlation of -0.140 with R² under 2% means "TIPS rose, gold fell" holds only 57.9% of the time (it went the other way on 42.1% of days). After spread costs that hit rate cannot support any short-term strategy.
• Stretch the horizon and the correlation appears. In the same window, the 5-day cumulative change correlation is -0.418 and the 20-day is -0.611. Once noise is averaged out, the opportunity-cost chain becomes visible. The real yield is therefore naturally a variable for holding days to weeks, not for minutes.
• It is weaker on big-move days. Across the 39 days when TIPS moved more than 5bp, the correlation was only -0.087. The reason is not mysterious: large jumps in real yields are usually driven by risk-off or liquidity events (a Treasury rally, a blowout inflation print), and those events push gold at the same time, so the two forces partly cancel. That is the setup for the section on when the relationship fails.

One definitional note: the companion piece measured correlation between the dollar index and real yields uses 252 trading days and a spot gold series and gets a daily correlation of -0.118; this article uses 247 days and COMEX futures closes and gets -0.140. The difference comes from the window and the instrument (futures carry roll and basis), not from an error. Both reach the same conclusion: daily explanatory power is a few percent — trust the sign, not the size.

5. Rolling 60 days: a stable sign, an unstable strength

Correlation is not a constant. Using a rolling 60-day window inside the sample above gives 187 windows:

• 94.7% of them are negative — the sign is stable enough to use as a background assumption.
• The median is -0.171, the weakest reading -0.039 (almost zero) and the strongest -0.354.
• The latest window (to 2026-10-22) reads -0.170, right around the median.

Practical implication: treat "negative correlation" as the default world view, but never use a specific correlation value as a threshold. When the rolling value drops to around -0.05, macro is not the dominant variable in that period (usually central bank buying or a risk-off move has overwhelmed the rate logic), and explaining gold with real yields has almost no power at all.

6. When this relationship fails

Three situations are typical, and each has an observable warning sign:

1. Risk-off and liquidity crunch

When the market is genuinely frightened it wants assets that convert to cash immediately, not real return. Gold and Treasuries get bought together and the real yield decouples from the gold price. Worse, the TIPS market is far shallower than nominal Treasuries, so in a squeeze a liquidity premium enters the yield and the reading itself is distorted. Warning signs: gold and Treasuries rising together, an abnormal widening between TIPS and nominal yields, a fast jump in VIX.

2. Central bank buying

After 2022, when some countries had reserves frozen, moving reserves into assets no other government can control went from theory to policy option. Such buying is price-insensitive, ignores real yields and takes metal directly out of the physical market. Warning signs: gold rising at the same time as real yields, gold ETFs seeing persistent outflows without the price falling, and the rolling correlation drifting towards zero.

3. Dollar squeeze and funding stress

When dollar funding tightens (repo rates spike, basis trades are forced to unwind), institutions sell whatever is easiest to sell for dollars, and TIPS is on that list. A rising TIPS yield then does not mean a better real return, it means a more expensive liquidity premium, and using it to infer gold gives you the answer backwards. Warning signs: abnormal overnight repo rates, the dollar index and gold rising together.

7. From 60 seconds to 5 minutes: how to use it, and why not to rely on it

Bluntly: on the measurements above, with a daily correlation of -0.140 and -0.611 only at 20 days, the real yield is not suitable as a minute-level entry signal. What it can do is act as a filter — deciding whether today favours a direction at all, not telling you when to enter.

A four-step filter, each step executable

1. Read it once before the open, not tick by tick. DFII10 is daily data, updated in the afternoon US Eastern time the next day; there is one new value a day. Refreshing it intraday adds no information for short-term trading.
2. Decide whether it is a macro-dominated day. Use |ΔTIPS| ≥ 5bp: only 39 of the last 247 days (15.8%) qualified. On days that do not, strike the real yield from your decision process.
3. Add weight only when it agrees with the dollar index. If real yields rise and the dollar also firms, both variables point the same way and confidence is highest; if one rises while the other falls, one of them is telling a different story, so cut size rather than doubling the bet.
4. Work out the cost hurdle first. Measured at 2026-10-24 13:13 Beijing time, the spread is USD 0.658 (bid 3,997.561 / ask 3,998.219); over the past year the standard deviation of daily gold returns is 1.758% with annualised volatility of 27.9%. At 1-minute resolution a round-trip spread consumes a serious share of the available range, and any directional edge has to clear that bar first — the arithmetic is in the gold spread comparison.

If you are trading minutes, what decides P&L is session and cost, not macro. For when liquidity is best and the spread narrowest, see the best hours to trade gold.

8. How to check it yourself: three codes and one habit

All of this data is free and FRED offers direct CSV download without an API key:

• DFII10 — the 10-year TIPS real yield, the core indicator on this page.
• DGS10 — the 10-year nominal Treasury yield.
• T10YIE — the 10-year breakeven inflation rate.

The download URL pattern is fred.stlouisfed.org/graph/fredgraph.csv?id=CODE; just swap the code. Updates land in the afternoon US Eastern time each business day (early morning Beijing time the next day) and slip on US holidays; what you see on Monday is usually Friday reading, so do not treat it as the value for today.

Keep your own table with four fields: date / DFII10 / DGS10 / note. The note records one thing only — whether there was a major event that day (FOMC, CPI, payrolls). Look back in three months and it becomes obvious which days were rate-logic days and which were event days that overwhelmed the rate logic.

Three pitfalls: TIPS is less liquid than nominal Treasuries, so closing quotes can be a basis point off; month-end index rebalancing can make TIPS yields jump for reasons unrelated to fundamentals; and inflation data get revised, so historical breakevens change, which means long-run comparisons need data pulled at the same point in time.

9. Three common misconceptions

• Treating the nominal yield as the real yield. 4.96% is nominal, 2.64% is real, and the 2.32 point gap is inflation expectations. Get that one number wrong and the conclusion can invert.
• Using CPI in place of inflation expectations. CPI is the past; TIPS prices the next ten years. A real yield built on CPI lags systematically around inflation turning points.
• Treating the daily correlation as a trading signal. -0.140 with R² of 1.97% is a relationship only slightly better than a coin flip. Its value is explanation and filtering, not forecasting the next candle.

For the everyday reading of the dollar index and Treasury yields, see the dollar index and Treasury yields: how they affect gold; to quantify the two variables together, see the dollar index and real yields: measured correlation. This article covers how to read the real-yield line, where to find it, and which holding periods it actually works for — the three do not overlap.

10. Frequently asked questions

Q. Why read the TIPS yield for the real rate instead of nominal minus CPI?
A. Because CPI is inflation that has already happened while gold is priced on the future. TIPS principal is adjusted with CPI, so the yield to maturity is the ten-year real return the market prices with real money. On 2026-10-22: nominal 4.96%, TIPS 2.64%, and the 2.32 point gap is inflation expectations, broadly in line with FRED T10YIE at 2.34% on 10-23.

Q. If real yields rise, must gold fall?
A. No. Over 247 trading days in the past year the daily correlation was only -0.140, R² 1.97%, and the two moved the same way on 42.1% of days. Over the same period the real yield rose from 1.73% to 2.64% (+91bp) while gold fell just 1.9%. The sign is negative, the strength is weak.

Q. Why is the daily correlation only -0.14 when the 20-day is -0.61?
A. Because daily TIPS changes are sticky: on 33 of 247 days (13.4%) there was no change at all, so the daily signal-to-noise ratio is very low; at 5 and 20 days the noise averages out and the opportunity-cost chain shows through. The real yield is therefore a variable that matches holding periods of days to weeks.

Q. Where can I find the 10-year TIPS real yield for free?
A. FRED series DFII10, direct CSV with no key at fred.stlouisfed.org/graph/fredgraph.csv?id=DFII10. It updates each business afternoon, and what you see on Monday is usually Friday reading.

Q. How should a 60-second to 5-minute trader use the real yield?
A. As a background filter only, read once a day. Use |ΔTIPS| ≥ 5bp to decide whether today is macro-dominated (only 15.8% of days in the past year qualified), weight it against the dollar index direction, and work out the spread hurdle first — USD 0.658 measured at 2026-10-24 13:13 Beijing time. Minute-level outcomes depend on cost and session, not on macro.