The dollar index and Treasury yields:how they actually move the gold price

Anyone who trades gold has heard both lines: "the dollar up, gold down" and "Treasury yields up, gold down". They are half right, and the wrong half is where accounts get hurt — because they describe a long-run tendency, not a signal you can place an order on.

No mysticism here. We take 224 real daily closes and measure how strong the statistical relationship between gold, the dollar index and the 10-year Treasury yield actually is, then unpack the variable that really drives it — the real yield — and finally switch to renminbi to see how much of an overseas Chinese investor's account was damaged by the exchange rate.

1. Today's board: the numbers side by side

The table below is the foundation for every calculation in this article. Gold comes from the XAUAD live quote feed; everything else is public closing data.

Basis: gold is the live quote at 2026-10-21 09:53 Beijing time; yields, index and FX are the previous session's close; CPI is the US Bureau of Labor Statistics figure for September 2026.

One detail matters: the fed funds rate at 3.88% is below CPI at 4.02%. That means the short-end real rate is negative, about -0.14%, while the 10-year real rate is positive at +1.12%. That steep curve — negative at the front, positive at the back — is the key to understanding the 2026 gold market, and section four unpacks it.

2. Three transmission chains, not one

Chain one: the unit-of-account effect (dollar index)

Gold is priced in dollars. As the dollar strengthens, the same gold becomes more expensive for holders of other currencies, demand falls and the price comes under pressure. This chain is the shortest and most visible, which is why so many people treat it as the only variable.

But check the magnitude: the dollar index went from 98.59 to 101.73, up 3.4%. If the unit-of-account effect were the only factor, gold should have fallen roughly 3%–4% — in fact it fell 4.6%, which looks like a perfect fit. That is a coincidence: the real-yield chain, a much thicker one, was working at the same time and the two partly offset each other.

Chain two: opportunity cost (real yields)

Gold pays no interest and no dividend. The cost of holding it is the risk-free interest you give up. With the 10-year at 5.14%, every ounce you hold (USD 4,117) forgoes about USD 211 of interest a year. Gold has to rise more than 5.14% for you to match a Treasury account doing nothing.

But the nominal yield is not the point — the real yield is. Inflation erodes that 5.14% too. The true competitor is the TIPS yield, the real rate the market actually trades with real money.

Chain three: credit and safe-haven demand (the one you cannot quantify)

The first two chains can be calculated. The third cannot: when the market doubts sovereign credit, geopolitical stability or the financial system, money flows into gold regardless of the interest cost. Gold's run to 5,318 in January 2026 was this chain briefly overpowering the other two.

The implication for traders is direct: the first two chains set gold's centre of gravity, the third sets the deviation. Trade direction off the centre of gravity and place stops off the deviation — that is the more robust way to use them.

3. Running the correlations: the counter-intuitive part

Using the 224 common trading days from 2025-10-20 to 2026-10-19, we compute the Pearson correlation of daily gold returns against daily changes in the 10-year yield and daily dollar index returns:

Read it plainly. -0.36 is a moderate-to-weak negative correlation: when the dollar rises gold is more likely to fall, but 87% of the intraday variance has nothing to do with the dollar index. -0.19 is weaker still — daily changes in the 10-year yield explain almost none of gold's movement.

That contradicts most people's intuition, and the reason is the time scale. On monthly and annual horizons real yields and gold are tightly linked; on daily bars the noise dwarfs the signal. So use them like this:

• For direction (holding days to weeks): watch the trend in real yields and rate-cut expectations — that works;
• For intraday: ignore the dollar index and yields, watch the order book, the session and the spread; macro data is just event risk;
• The most dangerous habit: shorting gold the moment the dollar index ticks up 0.3% — statistically that has almost no edge.

For a more granular treatment, read "The dollar index and real yields: how do they really move gold?" — it breaks out the weights of the six DXY currencies, the exact data definitions for the 10-year TIPS real yield (FRED series DFII10), 60-day rolling correlations, and how the relationship breaks down in risk-off episodes, during central bank buying and in dollar squeezes. Its sample is 252 trading days from 2025-10-23 to 2026-10-23 using Yahoo Finance GC=F, and it gets -0.374; this article uses 224 trading days from 2025-10-20 to 2026-10-19 and gets -0.36. Different windows and different sources, very close numbers — the two articles cross-check each other.

4. Real yields: subtracting 4.02% from 5.14%

This is the one piece of arithmetic worth memorising, because it explains the last nine months of the gold market.

One positive, one negative — what does that tell you? That the front end is still loose while the long end has already tightened. The market is saying: no rush now, but long-run inflation and debt supply will force long bonds to demand more compensation. This kind of bear steepening is one of the least friendly combinations for gold —

• A negative real rate at the short end: theoretically bullish for gold, and it stopped the price from collapsing;
• A positive real rate of +1.12% at the long end: it compressed the valuation, taking gold from 5,318 in January all the way back to 4,117, a 22.6% drawdown.

Look at monthly averages and the path is clear: April averaged 4,735, June 4,262, August bounced to 4,459, and October is back at 4,165. Every rally came in a window of rising cut expectations plus falling long-end yields; every decline matched yields turning back up.

5. Switching to renminbi: the exchange rate doubled the loss

The layer overseas Chinese investors most often miss: if your account is denominated in renminbi, you are carrying two variables — gold and the exchange rate.

The conversion is fixed: CNY/g = USD/oz ÷ 31.1035 × FX. Attributing the 19 October numbers:

• USD gold fell from 4,359.4 to 4,160.1, down 4.6%;
• CNY gold fell from 997.85 to 894.91, down 10.3%;
• Had the exchange rate stayed at 7.1195, the 19 October price would have converted to CNY 952.21/g;
• The actual figure was 894.91, a gap of -57.30 per gram, which is what renminbi appreciation took out.

The reverse also holds. When gold peaked at 5,318 on 29 January 2026, that converted to about CNY 1,189/g at the then exchange rate; at 884 today the drawdown in renminbi terms is 25.7%, deeper than the 22.6% in dollars. If you bought by the gram in a domestic jewellery shop during those days, add the shop's premium (typically 3%–8%) on top of the real loss.

6. The rate-cut trade: expectations matter more than delivery

Many people treat a Fed cut as the buy button for gold. The data says that button misfires regularly, because the market trades the expectation, not the decision.

The current setup: fed funds 3.88%, CPI 4.02%, 10-year 5.14%. Negative real rate at the front, inflation still above target, tight pricing at the long end — a combination of "room to cut but no willingness to cut fast". Gold typically behaves like this in such phases:

1. Data weakens → cut expectations rise → gold rallies (the trigger for every rally of the past year);
2. Inflation rebounds → cut expectations fade → gold retreats (the June and October pullbacks);
3. The cut actually lands → often "good news exhausted", unless the cut is driven by recession rather than by inflation cooling.

In practice, rather than guessing the decision, watch three things: the direction of CPI year-on-year, the slope of the 10-year yield, and the probability the market prices for a cut at the next meeting. The first two are hard data, the third is sentiment; sentiment drives short-term noise, hard data drives the medium-term direction.

7. A five-minute daily checklist

8. The cost ledger: the part short-term traders skip

Today's live spread is USD 0.662/oz, or 0.016% of the price. On its own that looks tiny, but leverage and frequency multiply it:

That is why most intraday traders lose: not because they get direction wrong, but because the cost structure is rigged against frequency. If you must trade short-term, either cut the frequency (once a day at most) or only trade sessions with enough volatility that your target profit is at least five times the cost.

Holding overnight adds swap financing, charged as an annualised percentage of notional, more expensive the higher your leverage — check your platform's published rates. On a trend trade it is negligible; on a long hold against the market it quietly becomes part of the loss.

9. Four common mistakes

1. Treating the dollar index as the only variable. The correlation is -0.36, explaining only 13% of variance. Use it as an intraday signal and your hit rate is close to a coin flip.
2. Using nominal yields instead of real yields. 5.14% sounds high, but with inflation at 4.02% the real rate is only 1.12%. Shorting gold every time you see "yields at a new high" gets you punished in the months when inflation runs hotter.
3. Ignoring the FX layer. CNY gold = USD gold × FX. Over the past year the exchange rate contributed four-tenths of the decline, so watching only the dollar price is watching half your account.
4. Carrying a heavy position into a data release. Around CPI or payrolls the spread instantly widens 3–10 times and slippage can take out your stop outright. Either go flat into the number or halve your usual size.

10. Frequently asked questions

Q1. If the dollar index rises, must gold fall?

A. No. Over the past 224 trading days the correlation between daily gold returns and daily dollar index returns was -0.36 — a moderate-to-weak negative relationship explaining about 13% of variance. A stronger dollar usually weighs on gold, but safe-haven demand, central bank buying and changes in real yields can all overwhelm the FX effect, so the dollar index alone cannot tell you which way today goes.

Q2. Why do higher Treasury yields weigh on gold?

A. Because gold pays no interest. With the 10-year at 5.14%, holding one ounce of gold (USD 4,117) means giving up about USD 211 of interest a year. The higher the yield, the more interest you forgo, and the less attractive gold is by comparison. What actually matters is the real yield after inflation, not the nominal one.

Q3. How is the real yield calculated, and where is it now?

A. The simplest version is nominal yield minus inflation. On the October 2026 numbers: 10-year at 5.14% minus CPI at 4.02% gives a real yield of about 1.12%, now positive. That is one of the main reasons gold has retraced 22.6% from its January high of 5,318 this year.

Q4. If the Fed cuts rates, does gold always rise?

A. No — what matters is whether the cut was already priced. The market usually lifts gold during the phase when cut expectations build, and then sells the fact once the cut lands. The reason for the cut matters too: a cut driven by cooling inflation is bullish for gold; a forced cut in a recession with real yields still high usually sees gold fall first and rally later.

Q5. Why did the CNY gold price fall more than the USD price?

A. Because CNY gold equals USD gold multiplied by the exchange rate. Over the past year USD gold fell 4.6% while USD/CNY went from 7.1195 to 6.6777, a renminbi appreciation of about 6%; combined, the CNY price went from 997.85 to 894.91 per gram, down 10.3%, with the FX factor alone accounting for roughly -57 per gram.

Related tools