FX Claims, Tested · #5

Do currencies follow bond yields?

“Watch the gap between two countries' 2-year bond yields and you'll know where their currency is heading.”

Verdict: mixed

It describes each month well, but it doesn't predict the next one.

Five major pairs · 2004–2026 · 2-year government yields · last run 1 Oct 2026
← All claims tested

The claim

Professional traders watch the gap between two countries' 2-year government bond yields. The 2-year yield is very sensitive to what markets expect central banks to do, so a widening gap is read as a sign that money will flow to the higher-yielding currency. Charts that lay a yield gap over a currency pair often look convincing, and we wanted to know whether the link is something you could trade. It builds on our central bank rates page, but uses market yields instead of the official rates.

The rules, set before we looked at any results

The gapThe base currency's 2-year yield minus the quote currency's, read at each month-end
PairsEUR/USD, GBP/USD, USD/JPY, USD/CAD, USD/CHF from late 2004
Question 1: does it describe?Compare the change in the gap during a month with the pair's move over the same month
Question 2: does it predict?Compare the change in the gap with the pair's move over the next month
The tradeGap widened over the month: buy the pair. Narrowed: sell it. Hold one month, starting the second trading day after month-end
Costs0.02% on every change of position (about two pips). Price moves only: interest is left out, because carry has its own test

Yields come from official sources: the US Treasury (via FRED), the ECB, the Bank of England, Japan's Ministry of Finance, the Bank of Canada, the Swiss National Bank and the Reserve Bank of Australia. Definitions differ slightly between them, which is fine for direction but not for exact basis points. AUD/USD is shown for information only, because Australian data starts in 2013, and the NZ dollar is left out for lack of free daily data.

Describing versus predicting

A correlation of +1 means two things move in perfect step, and 0 means no link. Gold bars are the month the gap changed in. Blue bars are the month after.

-0.2+0.0+0.2+0.4+0.6Same monthNext monthEUR/USD, same month: +0.44+0.44EUR/USD, next month: +0.05+0.05EUR/USDGBP/USD, same month: +0.22+0.22GBP/USD, next month: +0.04+0.04GBP/USDUSD/JPY, same month: +0.54+0.54USD/JPY, next month: +0.05+0.05USD/JPYUSD/CAD, same month: +0.38+0.38USD/CAD, next month: -0.09-0.09USD/CADUSD/CHF, same month: +0.28+0.28USD/CHF, next month: +0.06+0.06USD/CHFAUD/USD, same month: +0.28+0.28AUD/USD, next month: +0.01+0.01AUD/USD
PairMonthsSame-month correlationShare of the move explainedSame directionNext-month correlation
EUR/USD264+0.4419%67%+0.05
GBP/USD264+0.225%65%+0.04
USD/JPY264+0.5429%68%+0.05
USD/CAD264+0.3815%65%−0.09
USD/CHF263+0.288%59%+0.06
AUD/USD (from 2013)156+0.288%61%+0.01

It describes

In the same month, every pair moved with its yield gap: correlations ran from +0.22 to +0.54, and every one is statistically solid. The gap explains about 29% of USD/JPY's monthly moves and 19% of EUR/USD's, and noticeably less for the others. So the chart overlays aren't lying: yields and currencies really do move together.

It doesn't predict

Move one month along and the link disappears. Next-month correlations run from −0.09 to +0.06, and none is distinguishable from zero. Knowing the gap widened this month told us almost nothing about what the pair would do next month.

A likely reason. Bond yields and currencies react to the same news at the same moment, such as a surprise inflation figure or a central bank speech. By the time a month-end reading shows the gap has widened, the currency has usually already moved. That's an explanation, not something this test proves.

Would trading it have made money?

The simple rule: buy a pair when its gap widened over the month, sell it when the gap narrowed. Equal weight across the five pairs, after costs.

-20%-10%+0%+10%+20%20082012201620202024
Yearly return
−0.47%
Worst drawdown
-31%
Winning years
8 of 23
Best year
2008: +23.2%

Year by year

One big year (2008, when the financial crisis sent yields and currencies on violent one-way runs) and then mostly small losses.

2004: -3.1%042005: -6.5%2006: +1.7%062007: +0.6%2008: +23.2%08+23%2009: -4.9%2010: -7.8%102011: -4.4%2012: -3.4%122013: -7.2%2014: +2.1%142015: +8.7%2016: +8.6%162017: -6.1%2018: -0.0%182019: -3.7%2020: -2.6%202021: -5.8%2022: -0.3%222023: +0.3%2024: -0.5%242025: -1.3%2026: +3.4%26

Did the exact rule matter?

Two variations, reported as a check rather than a hunt for the best one. The highlighted row is the main rule.

RuleYearly returnSharpeWorst drawdownSharpe 2004–14 / 2015–now
Main rule: gap widened over the last month−0.47%-0.09-31%-0.19 / +0.01
Variation: over the last 3 months+1.60%+0.30-17%+0.45 / +0.14
Variation: hold the higher-yielding currency−0.35%-0.06-24%-0.19 / +0.05

The main rule lost money, and no single pair made any: results by pair ran from −1.57% (USD/JPY) to −0.03% (GBP/USD) a year. The 3-month version made +1.60% a year, positive in both halves, but with a t-statistic of only 1.4, which isn't distinguishable from zero. When three versions are tried, one of them looking mildly good is what chance alone regularly produces, so it's something to watch, not something to trade. Holding whichever currency yields more (the last row) didn't help either, which fits with what the carry test found.

Where the gaps sit right now

For interest: each pair's current 2-year gap, how it changed over roughly the last month, and what the pair did. Given the results above, treat this as context, not a signal. Figures to Thu 1 Oct; Australian yields are published a few days late.

PairGap (base minus quote)Change in the gapPair moved
EUR/USD−1.68−0.15−2.49%Agreed
GBP/USD−0.21−0.18−2.06%Agreed
USD/JPY+2.94+0.34−1.13%Diverged
USD/CAD+1.52+0.19+2.50%Agreed
USD/CHF+4.60+0.34+2.72%Agreed
AUD/USD+0.06−0.15−2.87%Agreed

This month the gap and the pair moved the same way on 5 of 6; USD/JPY was the exception.

What this means for you.
  • The overlay charts flatter the idea. A yield gap and a currency line up well within the same month, which is what makes the picture convincing. A trade needs the next move, and there the link is gone.
  • The gap is a good explainer, not a forecast. If a pair jumps and the yield gap hasn't moved, or the gap moves and the pair doesn't, that's worth a look. It helps you understand a move, but it won't tell you when one is coming.
  • Don't expect yields to lead. Both respond to the same news, so by the time the gap has visibly changed, the currency has usually moved with it.
What this test doesn't prove.
  • It tests a monthly horizon. Much shorter timeframes, such as the minutes after a data release, are a different question.
  • It uses 2-year government yields. The expected path of central bank rates, as priced in interest-rate futures or swaps, is a related but separate signal.
  • It measures price moves only. A trader holding the higher-yielding currency also earns the interest, which the carry test covers.
  • The New Zealand dollar isn't covered, and the Australian dollar only from 2013.
Risk warning: CFDs and spread bets are complex instruments and come with a high risk of losing money rapidly due to leverage. Most retail accounts lose money when trading them. You should consider whether you understand how they work and whether you can afford to take the high risk of losing your money. Satdish provides education, not financial advice.
Data: 2-year government bond yields from FRED (US Treasury), ECB (AAA euro-area curve), Bank of England, Japan Ministry of Finance, Bank of Canada, Swiss National Bank and the Reserve Bank of Australia; Yahoo Finance daily closes. Re-run monthly.