Skip to content

Gold and the dollar moved opposite on 66% of days since 2011

A daily candlestick chart of gold, OANDA XAUUSD, through 2025, with the US dollar index drawn over it as a blue line on its own scale, read on the left axis, and no indicators. Gold starts near 2,660, rises to about 3,500 on 22 April, trades between about 3,120 and 3,450 until August, climbs to about 4,380 on 20 October, dips to about 3,890 at the end of October, reaches about 4,550 in late December and ends near 4,320. The dollar index starts near 109.3, peaks near 110.2 in mid January, falls to about 98 in late April and about 96 in July and September, rises to about 100 in November and ends near 98.3. A white caption reads: Gold, OANDA XAUUSD, daily, 2025. Blue line: the US dollar index, TVC DXY, on the left axis. A large orange headline reads: Dollar down 1%: gold rose 85% of the time. A text box asks whether gold moves against the dollar: gold and the US dollar index, daily, 2011 to 2025, 3,796 days; they moved in opposite directions on 65.6% of days; every year more than half, from 58.9% in 2024 to 77.8% in 2018; 66.8% from 2011 to 2018 and 64.2% from 2019 to 2025; luck alone never matched it, 0 of 3,757 tries; on the 65 days the dollar fell over 1%, gold rose on 85%; on the 71 days it rose over 1%, gold fell on 76%; and in 2025 gold rose 64.7% and the dollar index fell 9.4% over the year. A blue label reads: Dollar index (blue).
Gold, daily, 2025, with the US dollar index in blue on the left axis: from 2011 to 2025 they moved in opposite directions on 65.6% of days.

Gold is priced in dollars, so a weaker dollar should mean dearer gold. We measured how often gold and the US dollar index moved in opposite directions from one close to the next, from 2011 to 2025.

What we measured

TradingView daily closes for gold (OANDA XAUUSD) and the US dollar index (TVC DXY), on the days both traded: 3,796 days.

What we found

They moved in opposite directions on 65.6% of days, about 2 in 3, with a correlation of -0.41. Every one of the 15 years was above half: the lowest was 2024 at 58.9%, the highest 2018 at 77.8%.

The big dollar days show it most. On the 65 days the dollar index fell more than 1%, gold rose on 55, 85%. On its 71 days up more than 1%, gold fell on 54, 76%.

YearOpposite directions (%)CorrelationDaysGold over the year (%)Dollar index over the year (%)
201159.4-0.23252+10.1+1.5
201268.3-0.48252+7.1-0.5
201364.4-0.41252-28.1+0.3
201468.5-0.45251-1.9+12.8
201561.9-0.36252-10.3+9.3
201665.7-0.39252+8.5+3.6
201768.3-0.56251+13.2-9.9
201877.8-0.64252-1.6+4.4
201964.7-0.51252+18.3+0.2
202060.5-0.36253+25.1-6.7
202166.7-0.51252-3.7+6.3
202266.4-0.50251-0.3+8.3
202367.8-0.54257+13.1-2.1
202458.9-0.39259+27.2+7.1
202564.7-0.41258+64.7-9.4

Is it luck?

The two halves agree: 66.8% from 2011 to 2018, 64.2% from 2019 to 2025. Shifting gold’s days against the dollar’s, 3,757 ways, never gave a share as high; the highest was 53.3%.

Opposite days, not always opposite years

Over whole years they went opposite ways in 11 of the 15. In 2011, 2016, 2019 and 2024 both rose; in 2024 gold gained 27.2% and the dollar index 7.1%, while they moved opposite on 58.9% of days.

What to do with it

On a big dollar day, expect gold to move the other way more often than not. Over a year, look at both. Our post on gold, silver and copper measured how far each moves in a normal week: gold 5%, silver 11%, copper 5%.

On the chart

Gold, daily, 2025, with the dollar index drawn over it in blue on its own scale: read it on the left axis. Over 2025 gold rose 64.7% and the dollar index fell 9.4%.

How to check two markets yourself

  1. Take both markets’ daily closes on the days both traded.
  2. Count the days they moved in opposite directions, year by year.
  3. Then look at the biggest days of one market and see what the other did.

What this does not tell you

Which way either will go, or which one moves first. Past data.

Observations, not recommendations.

We’re Signal Pilot Labs. We build the Elite Seven, seven indicators for TradingView, and we post what we measure, including the parts that didn’t work.

See what we build at signalpilot.io.


Past measurement described after the fact. Nothing here is a recommendation or an indication of future results.