When the Safe Havens Stop Working — Market Roundup, Week of 29 March 2026

Gold is supposed to go up when the world is on fire. This week, it went down — hard — while stocks collapsed and oil surged to levels not seen since 2008. The traditional playbook for a geopolitical shock isn't playing out, and buried inside our indicator data is the reason why: something more complicated than a simple risk-off rotation is happening, and the credit markets may be starting to price it in.

The Conventional Read

The story analysts have settled on is straightforward. The US-Israeli war on Iran that began February 28 has closed the Strait of Hormuz — a chokepoint for roughly one-fifth of global oil and LNG supply — sending energy prices into a vertical move. Brent crude has surged from around $70 pre-conflict to nearly $120 per barrel, its highest level since 2008. USO, the oil ETF that tracks WTI futures, is up over 55% in 21 days.

Meanwhile, the economic backdrop entering the conflict was already deteriorating. Q4 2025 US GDP came in at a revised 0.7% annualised — down sharply from the 4.4% pace in Q3 — with consumer spending growth revised lower to 2% and exports cratering to -3.3%. Consumer sentiment hit its lowest reading of 2026 as of the March survey, with the Michigan index now 33.9% below its long-run average, a level historically consistent with recession entry. The Conference Board's February Consumer Confidence reading of 91.2 sits well below the 112.8 peak from November 2024, and the ECB has already postponed planned rate cuts while raising its 2026 inflation forecast.

The consensus macro view: stagflation risk is elevated, the Fed is paralysed between growth weakness and energy-driven inflation, and equities should be selling off. On that last point, at least, the market has delivered. The SPY closed Friday at $634.09, down 8% over the past 21 days and now 6.7% below its 50-day moving average. QQQ fared worse: $562.58 on Friday, off 7.7% on a 21-day basis and 7.2% below its 50dma. Both indices are hovering near the 200-day moving average after breaking through the 50-day earlier in the week — a structural breakdown that technical analysts treat as a meaningful regime change.

The VIX closed the week elevated, consistent with the kind of broad fear reading that typically marks periods of peak stress. That part of the conventional read is correct.

What's less obvious — and what our data surfaces — is the divergence hiding beneath these headline numbers.

What the Percentile Data Shows

Flipside tracks four dimensions for each asset using a rolling 365-day percentile framework. Rather than asking "is this RSI number high or low in absolute terms?", the percentile approach asks: how unusual is this reading compared to everything we've seen over the past year? A reading at the 5th percentile means the indicator has only been lower 5% of the time. At the 95th, it's only been higher 5% of the time.

Here's where the key assets sit as of 27 March 2026:

Data as of 27 March 2026. Percentiles based on rolling 365-day window.

A few things stand out immediately. SPY and QQQ are at sub-4th percentile extension — meaning price is more compressed relative to its moving averages than 96%+ of all prior daily readings over the past year. That's not "weak" — that's near maximum oversold. Momentum is similarly crushed, below the 3rd percentile for both. By composite score alone (trend_momentum_score of 4.7 for SPY, 4.1 for QQQ on a 0–100 scale), these are as weak as it gets.

But composite labels are insufficient. The flow dimension tells a different story for each asset. SPY's flow sits at the 33rd percentile — not outright bearish, suggesting the distribution isn't yet at extreme levels despite the price damage. QQQ's flow is worse at the 19th percentile, consistent with more aggressive selling. And GLD's flow is at the 4th percentile — in complete freefall — which is the number that breaks the conventional geopolitical-shock playbook.

Gold should be accumulating in a war-driven risk-off move. Instead, it's experiencing some of the most extreme distribution in the past year.

The Gold Anomaly — And What It Means for Everything Else

Let's dwell on the gold data because it's the most important signal of the week.

GLD closed Friday at $414.70, down 13.1% on a 21-day basis (roc_21: -13.1%), down 10.0% on a 10-day basis (roc_10: -10.0%), and now trading 8.9% below its 50-day moving average. The RSI is 38.4. The flow score sits at 19.4/100 and is falling. Volatility is at the 95th percentile.

This is not a slow, orderly retreat. Gold is being liquidated in a high-volatility environment while a geopolitical crisis is escalating. There are only two plausible explanations historically: forced selling to cover losses elsewhere, or a broader breakdown in the "gold as safe haven" thesis due to structural concerns about the USD or inflation regimes. In the current context — with oil-driven inflation expectations rising and Treasury yields not falling as equities sell off — both may be operating simultaneously.

The GLD data is particularly striking because the 3-month return (roc_63) remains slightly positive at +0.67%, meaning the long-term trend is still intact, but the short-term shock has been violent. This pattern — long-term trend intact, short-term extreme distribution — has historically occurred at inflection points, not trend reversals. But the volatility percentile at the 95th makes the current move more ambiguous than usual.

The Bond Market's Uncomfortable Silence

The second anomaly in the data is what isn't happening in bonds.

In a standard risk-off episode — a 2008, a 2020 — equities fall and bonds rally as money floods into safety. Treasury yields fall. TLT goes up. That's the textbook. This week, it isn't happening.

TLT (20+ year Treasuries) closed at $85.64, with a 21-day return of -5.1% and a 10-day return of -1.0%. The RSI is 38.0. CMF is -0.15 — negative. Flow sits at the 37th percentile. IEF (7-10 year) is similarly soft: -3.1% over 21 days, CMF at -0.16, RSI at 35.

Both are below their 50-day moving averages. Neither is rallying.

The credit market paints an even sharper picture. HYG (high yield corporate bonds) closed at $78.72 with CMF at -0.18 — the most negative reading in this week's dataset for the credit complex — and a 21-day return of -2.2%. LQD (investment grade) shows CMF at -0.13 and a 21-day return of -3.3%. Both have flow scores below 25/100 and trend scores in the low-to-mid teens.

High yield spreading wider while equities sell off is credit stress 101. But the combination — equities down, gold down, bonds down, credit spreads widening — is the configuration that historically suggests a liquidity event rather than a simple risk-off rotation. When investors can't find a safe haven bid, it often means they're not rotating: they're raising cash.

The Morgan Stanley analysis on the Iran conflict noted that defense spending could raise US deficits further, adding pressure to Treasury term premiums — the additional yield investors demand to hold government debt with a challenging fiscal trajectory. A fiscal-pressure bid on yields combined with an energy-inflation scare could explain why the "flight to safety" in Treasuries simply isn't showing up this time.

The One Asset That Follows the Script

Against this backdrop of broken correlations, USO is the one asset behaving exactly as expected — but in a magnitude that makes it its own extreme.

USO closed Friday at $124.20. The 21-day return is +55.7%. The 63-day (quarterly) return is +76.9%. Extension is at the 95th percentile. Flow is at the 91st percentile. Volatility is at the 97th percentile. The trend score is 88.9/100. The flow score is 90.5/100. The RSI is 68.6.

This is a genuine parabolic move. The Strait of Hormuz closure has removed roughly 20 million barrels per day of throughput from global markets — about one-fifth of total supply. As the FPRI analysis noted this week, QatarEnergy's CEO confirmed that approximately 17% of Qatar's LNG export capacity could remain offline for up to five years due to infrastructure damage requiring billions in repairs. Unlike the Russia-Ukraine energy shock of 2022, which was addressable through rerouting and substitution, the Strait of Hormuz closure is a physical chokepoint that cannot be bypassed.

At the 95th percentile extension, USO is in territory where mean reversion risk is high. But the discriminating variable here is whether the physical supply disruption resolves quickly. If it doesn't — and the Al Jazeera analysis and CSIS both suggest it won't — historical parabolic oil moves driven by genuine supply destruction (Gulf War 1990, Hurricane Katrina 2005) sustained themselves longer than the market expected.

The Fundamental Tension

The bull case for markets: Geopolitical crises are typically buying opportunities. Morgan Stanley notes that markets have historically posted double-digit gains in the three to six months after the onset of major conflicts, led by defence and energy. SPY and QQQ are at maximum oversold readings on a 1-year percentile basis. RSIs for both are near 30. A ceasefire or Strait reopening would likely trigger a violent relief rally in equities, a collapse in energy prices, and a gold re-accumulation phase.

The bear case: The data architecture doesn't look like a normal geopolitical-shock oversold. The broken safe-haven correlations — gold down, bonds not rallying, credit distributing — suggest a potential liquidity or solvency stress layer beneath the geopolitical noise. Q4 GDP at 0.7%, consumer confidence at generational lows, and an energy shock layered on top creates the conditions for actual earnings deterioration in Q2 and Q3, not just sentiment damage. The Conference Board noted their GDP revision "does not yet reflect the impact from the war in Iran." With oil at near-2008 levels, the hit to consumer spending, industrial margins, and corporate profitability is not yet in the numbers.

The stagflation scenario — where the Fed can neither cut (inflation is rising) nor raise (growth is already sub-1%) — is the one that turns a temporary geopolitical shock into a structural bear market. In that scenario, the broken correlations we're seeing now are early warnings, not anomalies.

Where We Are Now

Today's profile most closely resembles two historical configurations: the early days of the 2022 Russia-Ukraine energy shock (February–March 2022), and the early weeks of the 2020 COVID crash (February–March 2020), not in terms of cause but in terms of data structure — maximum short-term oversold, active distribution in flow, correlations breaking down.

The 2020 COVID profile eventually resolved in a V-shaped recovery driven by Fed intervention and unprecedented fiscal stimulus. The 2022 Russia-Ukraine profile resolved more slowly: equities bottomed in June 2022 (four months after the initial shock), with the energy impulse continuing to drive CPI well into 2023.

The critical difference today: in 2020, bonds rallied massively, providing a clear signal that the risk-off rotation was orderly. In 2022, bonds also sold off as inflation expectations rose — a more analogous precedent to today. The 2022 SPY reached its final low in October 2022, roughly eight months after the initial shock.

The ambiguity here is genuine. The percentile data shows maximum oversold conditions that have historically preceded sharp recoveries. But the configuration of broken correlations and credit market distribution suggests the potential for a second, more severe leg down if the conflict extends.

What to Watch

1. The TLT/IEF flow percentile. If money starts genuinely flowing into Treasuries — flow percentile rising toward 60–70+ — it signals that the broken safe-haven thesis is resolving and a traditional risk-off rotation is taking hold. That would set up a more standard oversold bounce in equities. If TLT's flow percentile continues declining while equities remain weak, it's the credit-stress/liquidity scenario.

2. Gold's 21-day return turning less negative. GLD's roc_21 at -13.1% is extreme. Watch for stabilisation in the -5% to -8% range over the next week. Any re-accumulation signal in GLD's CMF (currently -0.086, trending from deeply negative) would confirm that forced selling has exhausted itself and the geopolitical bid is reasserting. A continued deterioration below -15% on a 21-day basis would be historically unusual and would warrant elevated concern about broader portfolio deleveraging.

3. The oil price trajectory and Strait of Hormuz status. USO at the 95th extension percentile with flow at 91st is parabolic territory. The single most important observable for the next two weeks is whether there is any credible diplomatic movement toward reopening the Strait, or any evidence of meaningful alternative supply routes being established. A ceasefire announcement would likely collapse USO by 20–30% within days, rerate energy-driven inflation expectations, and potentially allow the Fed to pivot back to a more supportive stance. No ceasefire, and USO's elevated readings become the constraint that keeps everything else under pressure.

Disclaimer

Data as of 27 March 2026. Percentile values calculated on a rolling 365-day window using Flipside's indicator pipeline. Composite scores (trend_momentum_score, flow_accumulation_score, risk_profile_score) are proprietary Flipside calculations based on multiple underlying indicators. Forward returns referenced are derived from historical pattern matching in our database and do not guarantee future outcomes. All macro data sourced from publicly available reports including The Conference Board, University of Michigan, CNN Business, World Economic Forum, Al Jazeera, FPRI, and CSIS, accurate as of the publication date. This article is for informational and educational purposes only. It does not constitute investment advice. Past data patterns do not guarantee future results. Please consult a qualified financial professional before making investment decisions.