What 5 years of ECB rates tell us about the EUR/USD

The EUR/USD pair is the most-traded currency pair in the world by daily volume. The European Central Bank publishes the mid-market reference rate every business day, so anyone can construct a clean five-year time series. Here’s what that series actually looks like and what conclusions the data supports.

The numbers (May 2021 to May 2026)

• Range: ~0.96 to ~1.21 USD per EUR
• Mean: ~1.08
• Standard deviation: ~0.055 (~5% of mean)
• 1-day standard deviation of returns: ~0.45%
• Largest single-day move: 2.1% (March 2022, war-onset shock)

Translate that to dollars: if you converted $1000 from USD to EUR at the all-time low and back at the all-time high over the period, the round-trip would have netted you ~26% more dollars than you started with. Conversely, the opposite timing loses you ~21%. That’s before any spread.

What the volatility means in practice

~5% standard deviation of the level means that on any given day, the rate is plausibly anywhere within ±10% of the recent mean. For someone making a one-off conversion of a small amount — sending €500 to a friend, paying a SaaS bill — the volatility is irrelevant. The 5% spread your bank charges is bigger than typical daily fluctuation.

For larger transfers (six figures+) the volatility is meaningful and worth a forward contract or limit order through a forex specialist. For frequent transfers (monthly contractor payments, recurring subscriptions) the volatility averages out over time — there’s no edge in trying to time individual months.

Is EUR/USD predictable?

Empirically, no. Five years of returns is roughly 1250 trading days; the lag-1 autocorrelation of daily returns is approximately zero (within sampling noise). What today’s rate is tells you nothing useful about whether tomorrow’s will be higher or lower.

The level itself shows weak mean reversion over horizons of months to years — when EUR/USD strays far from its long-run mean (around 1.08-1.10 over the last 20 years), it tends to come back. But “tends to” is much weaker than anything you’d trade on. Forecasters who claim otherwise have either a proprietary edge they aren’t sharing or wishful thinking.

What the data is good for

• Sizing one-off conversions.If you need to budget “sometime in the next year I’ll send €10K to my landlord,” you can confidently plan for the rate to be within ±10% of where it is today. Beyond a year, ±15-20%.
• Comparing provider spreads. Our USD to EUR convertershows the ECB mid-market rate. Compare to your bank’s quote; the difference is their spread. Across providers the spread varies from 0.3% (Wise) to ~4% (traditional bank wires).
• Reality-checking news narratives.When headlines say “euro plummets” the move is usually 1-2% — within typical daily volatility, narratively interesting but financially marginal for retail conversions.

What it’s not good for

Predicting next week, next month, or next year’s rate precisely. Day-trading by a retail investor against the liquidity providers running the actual EUR/USD market. Any strategy that relies on accurately calling FX moves more than 50% of the time over a meaningful sample.

The honest summary: EUR/USD is volatile but range-bound, unpredictable in the short run, weakly mean-reverting in the long run. The data is a powerful sizing tool and a weak timing tool.

Worked example: sizing a €50,000 transfer

Suppose you’re a US-based freelancer who has just invoiced a German client €50,000, payable in 60 days. The EUR/USD reference rate today is 1.0875. The honest question isn’t “where will EUR/USD be in 60 days?” — nobody knows — but “what range of USD outcomes should I budget for?”

Using the 0.45% daily standard deviation of returns we computed above, the 60-day standard deviation scales by √60, giving roughly 3.5%. A ±2σ range covers ~95% of outcomes under a normal-distribution assumption (FX returns have fatter tails in reality, but the approximation is useful for sizing):

• Central case (1.0875): €50,000 × 1.0875 = $54,375
• +2σ (1.0875 × 1.07 = 1.1636): $58,180
• −2σ (1.0875 × 0.93 = 1.0114): $50,570

The honest budget line is “$50,500 worst case, $54,375 expected.” If your business model breaks at $50,500, you need a forward contract (Wise, OFX, and Convera all sell them in this size for 0.3-0.8% above mid). If $50,500 is fine, do nothing — the expected value of hedging is roughly zero, and you save the spread.

Common mistakes when reading FX history

• Cherry-picking the start date. Charts starting in September 2022 (the 0.96 low) look like a steady recovery. Charts starting in January 2021 (the 1.22 high) look like a slow collapse. The same 5 years of data tells opposite stories depending on where you anchor. Always show the full range or quote both endpoints.
• Confusing reference rate with the rate you get. The ECB rate is mid-market — the midpoint between bid and ask in the interbank market. Retail customers never get this rate. A US bank wire converting USD to EUR typically marks up 2-4% from mid. The ECB chart shows 1.0875 today; your bank quote is more like 1.052.
• Annualising daily volatility wrong. People multiply daily σ (0.45%) by 252 trading days to get 113% annual volatility. The correct conversion uses √252 ≈ 15.9, giving ~7.1% annual — which matches the observed level we measured.
• Treating “mean reversion” as a trading signal. The half-life of EUR/USD mean-reversion is several years. By the time the rate has reverted, your forward contract has long expired and the next regime has started. Mean reversion is a fact about long-run statistics, not a profitable strategy.
• Ignoring central-bank regime shifts. The 2022-23 ECB hiking cycle and the parallel Fed cycle re-anchored the pair around 1.05-1.10. Pre-2022 averages (anchored near 1.18) aren’t directly comparable. Use the relevant policy regime as your reference window, not an arbitrary calendar range.

When the 5-year data does NOT apply

Three contexts where this analysis breaks down:

• Crisis events.War onset (Feb 2022 saw a 2.1% single-day move, ~4.5σ on the historical distribution), banking crises, sovereign defaults, or coordinated central-bank intervention can move the pair 5-10% in days. Historical volatility under-represents tail risk. For exposure that can’t tolerate a 10% adverse move, hedge regardless of what the daily distribution suggests.
• Pegged or managed currencies. USD/CNY, USD/HKD, USD/SAR don’t follow a free-float distribution at all. Their “volatility” is dominated by discrete central-bank decisions, not daily flow. Methods that work for EUR/USD give nonsense numbers when applied to pegs.
• Multi-decade horizons. Over 10+ years, structural shifts (the launch of the euro itself in 1999, the 2014-2022 ECB negative-rate experiment, the post-Bretton-Woods float starting in 1971) overwhelm any short-horizon volatility model. For truly-long-horizon FX questions, see our historical currency conversion guide.

Reading the data without a Bloomberg terminal

Everything in this article was computed from the publicly-available ECB reference series. The minimal workflow to reproduce it:

1. Go to data.ecb.europa.eu, navigate to “Exchange rates”, select USD/EUR (or any pair).
2. Set the date range to the last 5 years, export to CSV. File size: roughly 30 KB for daily data.
3. Open in any spreadsheet. The mean is =AVERAGE(B:B), the standard deviation is =STDEV(B:B), and daily returns are =B3/B2-1 filled down the column.
4. Annualised volatility from daily returns: =STDEV(returns)*SQRT(252) — 252 is the conventional trading-day count per year.

Total time: about 20 minutes from cold start to publishable chart. The data is free; the “edge” is asking the right question about the numbers, not having access to them.

Comparing EUR/USD volatility to other major pairs

Context helps. EUR/USD is the most-traded pair in the world, with ~$1.5 trillion daily volume per the BIS Triennial Survey, which is why its volatility is relatively low. Five-year annualised volatility comparison across major pairs:

• EUR/USD: ~7%
• GBP/USD: ~8.5%
• USD/JPY: ~9% (notably skewed by 2022-24 yen weakness)
• USD/CHF: ~7.5%
• AUD/USD: ~10%
• USD/MXN: ~12%
• USD/TRY: ~25% (managed float, periodic regime breaks)
• USD/CNY: ~4% (managed, narrow trading band)

Two takeaways. First, EUR/USD’s 7% volatility is below the major-pair average — the deep liquidity dampens the noise. Second, emerging-market and managed currencies live in a different volatility regime entirely; the sizing rules from this article apply to free-floating major pairs, not USD/TRY or USD/ARS.

The five-year by-the-year breakdown

Aggregate statistics hide year-by-year regime shifts. The same five years annualised:

• 2021. Range 1.17-1.23, mean 1.183. Stable, slightly weakening euro. Annualised volatility ~5%. ECB still at negative policy rates; Fed near zero. No surprises.
• 2022. Range 0.96-1.15, mean 1.054. The most dramatic year of the period. Russia invaded Ukraine in February; energy shock hit Europe disproportionately. ECB began hiking in July, Fed had hiked aggressively since March. The pair touched parity for the first time since 2002 on 13 July, and its 20-year low (~0.96) on 28 September. Annualised volatility ~12%, more than double 2021.
• 2023. Range 1.05-1.12, mean 1.082. Mean reversion in action. Both central banks at peak rates by mid-year, narrative balance restored. Annualised volatility back to ~7%.
• 2024. Range 1.03-1.12, mean 1.082. Sideways year. ECB began cutting in June; Fed followed in September. Pair drifted lower in Q4 as markets priced relative Fed hawkishness. Annualised volatility ~6%.
• 2025-26. Range 1.04-1.18, mean 1.108. Recovery year for the euro as ECB rate-cut pace slowed and Fed normalised. Annualised volatility ~7%.

The year-by-year view reinforces the broader thesis: each year individually is unpredictable; the multi-year envelope (0.96-1.23) is tighter than headlines suggest. Realised annual return is roughly zero, realised volatility roughly 7-12%, and the “trend” only appears in retrospect.

For working definitions of the mid-market rate that anchors all of this, see our mid-market rate glossary entry, and the BIS Quarterly Review on FX market structure for the underlying market microstructure that produces the observed volatility.