The Model Says Buy

The fair value model — a rolling 252-day OLS of EUR/USD on the US-DE 2-year rate spread — has EUR/USD at 1.1720. Spot is at 1.1517. That's a 203-pip gap, or −2.16 standard deviations below model-implied value.

This is the widest dislocation since March 2nd, when the spread was still at 145bp and spot was grinding toward 1.1700. In two weeks, spot fell 200 pips while the rate spread only compressed 7 basis points. The model moved a little. Price moved a lot.

Fair Value Model · EUR/USD vs 2Y Rate Spread
1.190 1.180 1.165 1.150 1.140 13 FEB 20 FEB 27 FEB 6 MAR 13 MAR 17 MAR FV 1.172 1.152 −203 pips −2.16σ HORMUZ

What the chart shows: spot and fair value tracked each other into late February. Then the Hormuz crisis hit, and they divorced. Spot cratered on risk-off flows. The rate spread — which should be the primary EUR/USD driver — barely moved. The 2Y US-DE spread went from 145bp to 138bp. That's a 7bp compression in a world where oil doubled.

The model's read: EUR/USD should be 1.1720 based on rate differentials alone. The 200-pip discount is pure war premium.

The Rate Story

Rate Architecture · 17 March 2026
UNITED STATES
Fed Funds 3.64%
SOFR 3.70%
2Y Treasury 3.73%
10Y Treasury 4.28%
10Y Real 1.92%
10Y Breakeven 2.36%
EUROPE
ECB Depo 2.00%
ESTR 1.93%
2Y Bund 2.33%
10Y Bund 2.99%
HICP 1.9%
Core HICP 2.3%
2Y SPREAD
138bp
z = −1.27
10Y SPREAD
127bp
z = −0.95
US 10-2
55bp
positively sloped
DE 10-2
66bp
steeper than US

The 2Y spread at 138bp is at the −1.27σ z-score — 64th percentile historically, compressing. It was 145bp at the start of March. The direction is EUR-supportive: the spread is narrowing as the market prices divergent paths (Fed on hold longer, ECB potentially hiking into oil inflation).

But the 20-day change is only −3bp. This is not a rate repricing driving EUR/USD — the rate spread is barely moving. What moved EUR/USD 200 pips in two weeks was pure risk premium. The rate correlation (rolling 60-day EUR/USD returns vs spread changes) is at −0.245 — weakly negative, meaning the rate channel is operating but dominated by other forces.

The German 10-2 curve is steeper than the US curve (66bp vs 55bp). This matters because it reflects the Bund market pricing in the German fiscal bazooka: €126.7 billion in 2026 investment, €500 billion infrastructure fund, defense spending up to 2.8% of GDP. The long end of the Bund curve is repricing for a Germany that borrows and spends. That's structurally EUR-positive.

Who's In the Trade

Net Non-Commercial
+105,144
Specs still net long — but down from +157K in late Feb. Aggressive unwind. z-score = −1.66 on 52-week window.
Contrarian Signal
+0.80
tanh-normalized. Long liquidation has overshot — readings above +0.60 historically precede EUR rebounds within 1-3 weeks.
Est. Avg Entry
1.1680
13-week position-weighted average price. Remaining longs entered around 1.1680 — currently underwater by 163 pips.
Pain Threshold
1.1447
2% loss level for long base. Spot hit 1.1416 on Friday — below pain. Forced liquidation risk is live.
ASSET MGRS
−7,727
net short
DEALERS
−1,538
net short
LEV MONEY
+2,290
net long
VELOCITY
−31,354
wow change

The TFF (Traders in Financial Futures) breakdown adds nuance. Asset managers are short (−7,727). Dealers are short (−1,538). Leveraged money is barely long (+2,290). But the headline non-commercial position — +105,144 — is still net long. What happened: the long base that built through January and February is being dismantled. Specs peaked near +157K in late February and have been cutting at −31K/week pace.

This is a liquidation story, not a crowding story. The z-score at −1.66 says the current long is 1.66 standard deviations below its 52-week average — an unusually aggressive unwind. War-driven flows forced longs to cut, and the contrarian signal at +0.80 says the cutting has gone too far.

Key level: the estimated pain threshold sits at 1.1447 — the 2% loss level for the remaining long base, whose average entry is 1.1680. Spot hit 1.1416 on Friday, breaking below pain. That means forced liquidation is already in play. The question is whether the last longs capitulate (driving EUR lower) or whether the liquidation is nearly exhausted (setting up a bounce). The contrarian model says the latter.

Vol Architecture

REALIZED 5D
9.76%
spiking
REALIZED 22D
6.46%
30th pctile
GARCH(1,1)
7.30%
conditional
VOL-OF-VOL
1.30
elevated

5-day realized vol at 9.76% annualized — that's a 50% premium over the 22-day window (6.46%). When short-dated vol spikes while medium-term vol stays low, it means the market is adjusting to a new information regime but hasn't fully repriced the range yet.

GARCH(1,1) conditional volatility at 7.30% confirms: the model expects vol to stay elevated but not explode. The vol-of-vol at 1.30 is above normal — volatility itself is unstable, which means the daily ranges this week (FOMC Wednesday, ECB Thursday) could be outsized in either direction.

VIX z-score at +0.83 tells us equity vol is modestly elevated but not panicking. This is consistent with the broader signal check from yesterday — fear is fading even as the war intensifies. That's the gap between realized geopolitical risk and priced-in financial risk.

What's Driving EUR — And What Isn't

vs 2Y Spread
−0.245
weak
vs Brent
−0.154
negative
vs VIX
−0.085
decoupled
vs Gold
+0.056
zero
vs DAX
+0.085
zero
vs SPX
+0.178
mild

The correlation map reveals something important: EUR/USD is not being driven by any single cross-asset relationship right now. The rate spread correlation is −0.245 — weak, and this is normally the primary driver. Brent is −0.154 — oil is hurting EUR but not dominating. VIX and Gold are near zero. DAX near zero — German equities have completely decoupled from EUR/USD.

When everything decouples simultaneously, it usually means a single dominant narrative is overriding all the structural relationships. That narrative is the war. EUR is being treated as a European risk asset: oil war → energy import shock → EUR sells. It doesn't matter that rate spreads are compressing (EUR-positive) or that German fiscal spending is steepening the Bund curve (EUR-positive) or that specs are already maximally short. The war is the only input.

This is the kind of environment where the model gets the direction right but the timing wrong. When the narrative breaks — ceasefire, escalation beyond what's priced, or just fatigue — the cross-asset correlations will reactivate and the fair value gap will close. The question is when, not if.

The Macro Under the War

EZ HICP
1.9%
pre-oil. core 2.3%
EZ GDP YoY
1.24%
latest available
EZ UNEMP
6.2%
Jan 2026. near record low
EZ SURPRISE
−0.22
momentum: below trend
IFO BUSINESS
−15.2
Feb 2026 · OECD BCI proxy
US-EZ DIFF
+0.35
EMA. US slightly ahead

HICP at 1.9% is below the ECB's 2% target — pre-oil-shock. This is the last print before Brent went from $70 to $100. The March and April HICP prints will be the ones that matter. If oil pass-through pushes headline HICP above 3%, the ECB has to react. Markets already price 85% odds of a July hike.

German business confidence at −15.2 is ugly. German economic sentiment at 93 (neutral is 100) is the weakest since the 2022 energy crisis. But German factory orders surged 40% annualized on three-month basis — driven by defense and infrastructure spending. The fiscal bazooka is arriving, but it takes quarters to flow through sentiment surveys.

The surprise differential at +0.35 (EMA) favors the US slightly — US data has been less bad than European data relative to expectations. But both are negative in absolute terms (US momentum −0.21, EZ momentum −0.22). Nobody is outperforming expectations. The world is surprising to the downside simultaneously.

Fragmentation Risk

BTP-BUND
71bp
contained · Dec 2025
OAT-BUND
72bp
France = Italy · Dec 2025
BONOS-BUND
43bp
tight · Dec 2025

No fragmentation stress. BTP-Bund at 71bp is well below crisis levels (it hit 250bp in the 2022 repricing). But the notable read here is OAT-Bund at 72bp — France is now trading wider than Italy. That's the Macron deficit premium. It's not a EUR crisis signal, but it's a structural credit story that sits beneath the surface.

The CIP basis proxy at −44bp is consistent with modest dollar funding premium — not stress, but not normal either. The ESTR-SOFR differential at −177bp reflects the rate divergence directly.

This Week: Fed Wednesday, ECB Thursday

FOMC · Wednesday
HOLD
Fed funds at 3.50-3.75%. Unanimous hold expected. Focus on the SEP — does the dot plot still show cuts this year, or does the oil shock push inflation projections up enough to pause indefinitely? Powell's tone on "transitory energy inflation" is the signal.
ECB · Thursday
HOLD → HIKE?
Deposit rate 2.00%. Hold expected. But Lagarde's guidance is the event. Markets price 85% odds of a July hike, with a second hike by year-end at 70% probability. If Lagarde validates the repricing, EUR rallies. If she pushes back, short squeeze may still happen on the "less hawkish than feared" read.

The asymmetry here is interesting. The EUR-negative scenario (hawkish Fed + dovish ECB) requires both central banks to surprise against expectations simultaneously. The EUR-positive scenario (any combination that isn't both) has more paths. Dovish Fed alone = EUR up. Hawkish ECB alone = EUR up. Both dovish = mixed but positioning unwind favors EUR. Both hawkish = EUR up on rate compression.

The only clean EUR-negative outcome this week is a Fed that signals no cuts for the rest of 2026 combined with an ECB that explicitly pushes back on July hike pricing. That's a narrow path.

Synthesis

Signal Summary
Fair Value Model BULLISH · −2.16σ below FV · strong dislocation Positioning BULLISH · Long unwind overdone · Contrarian +0.80 · Pain breached Rate Spread BULLISH · 2Y compressing −3bp/20d · z = −1.27 Surprise Diff NEUTRAL · +0.35 slight US edge · both negative Volatility CAUTION · 5d vol spiking · vol-of-vol elevated · event risk Fragmentation CLEAR · BTP-Bund 71bp · No stress signal Cross-Asset DECOUPLED · All correlations near zero · War narrative dominant Central Banks ASYMMETRIC BULL · More paths to EUR+ than EUR−

Five of eight signals lean EUR-positive. One is neutral, two flash caution on timing (vol and correlations). The model says this is a −2.16σ opportunity to buy EUR/USD — a dislocation that historically reverts. The long base that drove EUR higher in January is being liquidated hard, with the contrarian model saying the unwind has overshot. Rate spreads are compressing. The ECB is repricing hawkish while the Fed stands still. German fiscal spending is a structural EUR tailwind that hasn't been priced yet because everyone is staring at oil.

The catch: all of this was true yesterday too, and spot still fell to 1.1416 — breaking below the pain threshold for remaining longs. The war narrative overrides everything when it's active. The signal is "buy the dislocation" but the timing is "wait for the catalyst." That catalyst is most likely to arrive Thursday at 13:45 CET when Lagarde speaks.

Invalidation Levels

Model invalidation: If the 2Y spread widens back above 150bp (i.e., rate divergence reasserts in the US direction), the fair value estimate drops and the −2.16σ read compresses. Watch for a hawkish FOMC SEP as the trigger.

Positioning invalidation: If spot breaks below 1.1380 (below Friday's 1.1416 low), it means the remaining longs are capitulating, not stabilizing. That's a momentum regime, not a reversion setup — the liquidation has further to run.

Narrative invalidation: If Lagarde explicitly pushes back on July hike pricing ("premature to discuss rate increases"), the hawkish ECB leg collapses and the asymmetry shifts. That's the scenario where being long EUR hurts.

Data Source
Source Kayle DB · PostgreSQL · 82 raw series · 81 derived metrics · 115 CFTC obs · Policy rates, IFO, fiscal figures from public sources Model Rolling 252d OLS: EUR/USD ~ US-DE 2Y spread · GARCH(1,1) skewed-t · DIY surprise index As of 17 Mar 2026 · Prices: 16 Mar close · Bund: 17 Mar · CFTC: 10 Mar report
This material is published for informational purposes only by muFX and does not constitute investment advice, a solicitation, or a recommendation to buy or sell any financial instrument. All views expressed are those of muFX and are subject to change without notice. Past performance is not indicative of future results. Trading foreign exchange carries significant risk of loss.