HOME / ANALYSIS

Just before the subprime crisis (2007) vs today (September 2026).
Same film, or just the same trailer?

A multi-asset dashboard: gold, oil, gas, the dollar (DXY), the yen, bitcoin, credit derivatives, the S&P 500, the Dow Jones, options and volatility. Built from Q2-Q3 2007 levels, just before the ABCP freeze of 9 August 2007, against live levels from early September 2026. Every figure is sourced. The method is "zero invention".

Data as of 8 September 2026.

⭐ The essentials, explained simply

The plain story first. The figures follow in the sections below.

🏚️ What happened in 2008 (the domino story)

  • People borrowed money to buy homes they could not really afford.
  • Banks bundled all those repayment promises into large packages and sold them, labelled "safe".
  • When people stopped paying, the packages lost their value overnight. Because everyone had bought some, everything fell like dominoes: banks, stock markets, jobs.
  • The most striking part: right before the fall, everything looked calm. The instrument that measures stock-market fear (the VIX) had spent a year between 10 and 15, with a floor of 9.9 in January 2007.

🤔 And today, in September 2026?

  • 🧘 The fear gauge is asleep in the same way. The same displayed level of calm as before 2008, even with a war under way. That's the single most striking parallel.
  • 🏦 But the banks are solid this time. In 2008 they were playing with huge amounts of borrowed money. Today they hold large reserves. The banking dominoes are harder to knock over.
  • 🫣 The danger has changed its hiding place. In 2008 it was hidden in mortgage lending. Today it hides in "private" loans that nobody can see clearly, in gigantic spending on artificial intelligence, and among people paying in instalments "interest-free" (like Tabby in the UAE) while living beyond their means.
  • 🀫 A financial game of pick-up-sticks is on the move. Investors borrow in Japan (where it is almost free) to bet elsewhere. The yen is strengthening again, and it's already pulling at the sticks: two small tremors have already happened (2024, 2025) without bringing the pile down.
  • 🥚 All the eggs in one basket. Ten large companies make up 41% of the US stock market. If artificial intelligence disappoints, the whole index feels it.
  • 💪 And the jobs market? Unemployment is low. But it's mostly the state (through healthcare) doing the hiring: strip that out and private employment is falling. Not a healthy engine.
In one sentence: no, it's not the same film as 2008, the banks are holding up better, but it is the same dangerous recipe: deceptive calm plus debt hidden in dark corners. There's no certainty of a storm; there's plenty of dry wood. Watch, don't panic, and above all don't believe anyone who says "this time it's different, everything's fine".

1. The 30-second verdict

The geometry of 2026 resembles 2007. The fuel doesn't.

🔴 What resembles 2007 A crushed VIX (~14.5, a level partly compressed by zero-day options flows) despite a major geopolitical shock (the US-Iran war, oil above $100 at times) · record equity concentration (top ten ≈ 41% of the S&P, versus 22% for financials in 2007) · leverage that has moved out of banks and into shadow banking (the basis trade, $0.8-2.4tn, private credit) · recent regulatory easing (the eSLR change of December 2025 echoes the SEC's 2004 rule) · a yield curve that has dis-inverted after a record inversion · a divided Fed late in the cycle · a yen that strengthened by 1.7% in a single session, with a 76% probability priced in for a BoJ rate rise in September (a carry-trade unwind under way right now).
🟢 What is different No mass subprime mortgage credit (residential prices +2% year-on-year, sales weak but no purge) · over-capitalised banks (an SLR of ~6.5% versus 30:1 leverage at Bear Stearns and Lehman) · positive growth (Q2 GDP +1.5%) · the cycle has already absorbed five shocks (SVB/Credit Suisse in 2023, the yen carry trade in 2024 and 2025, Tricolor/First Brands in 2025, the Iran war in 2026) without a systemic break · bitcoin, which did not exist in 2007, now absorbs part of the market's beta risk (down 28% from its all-time high without spreading contagion). But one nuance stands out on closer inspection: the headline "4.1% employment" figure is a surface reading. Healthcare and social assistance (+680,500 jobs from March 2025 to March 2026, a sector funded by Medicare and Medicaid) created almost DOUBLE the total net job gain (+208,800). Strip out the publicly-funded sectors and the US private economy is net NEGATIVE on jobs. 2025 has just been revised down from 585,000 to 181,000 net job creations. Much of the 2026 "full employment" narrative is an artefact of public spending.
🟠 The real issue "2026's subprime", if it exists, lives exactly where banking statistics don't look: private credit (~$1.7tn), AI capital expenditure (~$700-725bn a year across four groups, exceeding their own cash flows), the Treasury basis trade, office commercial real estate (a 12.3% CMBS delinquency rate, an absolute record, worse than after Lehman). And at the household level, "interest-free" buy-now-pay-later debt: Tabby and Tamara in the UAE, for instance, where paying in four instalments at 0% means living beyond one's means with no trace on a credit file, only added to the Etihad Credit Bureau since July 2026. In 2007 as in 2026, risk grows precisely in the regulators' blind spot. That's the precedent repeating, almost to the letter.

2. Multi-asset dashboard: 2007 (just before) vs 2026 (now)

The "pre-crisis" benchmark is June and July 2007 (before the ABCP freeze of 9 August 2007 / BNP Paribas). The 2008 column shows the peak or trough of the acute phase for reference. 2026 levels are live, early September, sources listed in section 9.

AssetMid-2007 (pre-crisis)2008 (acute phase)Sept 2026 (live)Multiple 2026/2007Reading
🥇 GOLD (spot $/oz)$655.75 (fix, 15 Jun 07)$1,023.50 fix / ~$1,032 (17 Mar 08)$4,436 (8 Sep) · futures $4,480×6.8Off the charts. A different monetary regime (central banks bought 244t in Q1-26; JPMorgan sees ~755t for 2026) but a parabolic path: a closing peak of $5,318.40 on 29 Jan 26 (intraday $5,586) → down 15.8% from that peak, up 23.1% over 12 months, up 3.6% year-to-date. Gold has already given back a third of its rally, much as it did between March and August 2008.
🛢️ WTI CRUDE OIL$65.09 (1 Jun 07) · $78.20 (31 Jul 07)$147.27 (11 Jul 08)$92.94 (8 Sep) · Brent $97.59×1.4A hot market: up 61.9% year-to-date (the Strait of Hormuz shock: US inventories below the five-year floor, EIA STEO, 11 Aug 26); a closing peak of $112.95 (7 Apr 26) → down 17.7%, a low of $55.27 (16 Dec 25). As in July 2007 into early 2008, energy is rising BEFORE the crisis hits, and the EIA does not see it easing back to ~$69 until 2027.
🔥 HENRY HUB NATURAL GAS$6.40-7.89 (Jun 07, FRED)peak $13.31 (2 Jul 08)$2.95 (8 Sep, down 20% YTD)×0.42The opposite picture in the US. Gas there in 2026 trades at half its 2007 price (shale and LNG). But there is a total split with Europe: TTF gas at €71.95/MWh, up 155% YTD, at its 52-week peak this week (€73.63 on 2 Sep): the 2026 gas stress is European (a Hormuz-LNG shock), not American.
💵 DXY (DOLLAR INDEX)~81-8276.70 (2007 close) · low of 70.70 (16 Mar 08)98.83 (8 Sep)×1.21Neutral, with a positive tilt. 2025 was the dollar's worst year against the G10 on record (down 9.4%), followed by a 2026 rebound (a 96-102 range all year). This is not a dollar crisis, but a gradual loss of status, and gold at $4,400 is its direct counterpart, and the DXY's low on 27 Jan coincided almost to the day with gold's peak (29 Jan).
💴 YEN (USD/JPY)122.10 (1 Jun 07) · peak 124.09 (22 Jun 07) · carry trade ~$1,000-1,500bn107.51 (day before Lehman) → ~87 (1 Sep 09)153.44 (8 Sep, down 1.7% on the day) · peak 163.86 (28 Jul) · carry trade ~$500bn×1.25The key parallel, live. In 2007 the yen carry trade unwound in August and swept everything with it. In 2026 it is unwinding right now: the yen gained 2% on 3 Sep, then fell a further 1.7% on 8 Sep, following a record joint US-Japan intervention in July (near 160); markets are now pricing 76% odds of a BoJ rate rise in September (versus 24% at end July, Tokyo Tanshi survey). Roughly $500bn of carry trade remains open (half the 2007 stock) after TWO mini-crashes already (August 2024: VIX at 65 intraday; August 2025: VIX ~38). This is the only "2007-style" asset whose unwind is happening in front of us: a peak in late July followed by a 6.3% fall in six weeks, accelerating this week.
₿ BITCOINn.a. (did not exist)n.a.~$79,000n.a.A new player: it did not exist in 2007. In 2026: ~$79,000 (Binance $79,062 · CoinMarketCap $79,311 · Coinbase $79,906), down 28% from its January 2025 all-time high (~$109,000). Its role in 2026: an institutionalised risk valve (spot ETFs). Its correction without contagion (unlike 2022) suggests some of the market's beta risk is being absorbed by this asset class. Read it as a risk-appetite gauge, not as a hidden-damage detector: bitcoin is liquid, continuously priced and collateral for almost nothing, so a fall reveals no gap between marked value and real value.
⚠️ CREDIT (CDS)CDX IG ~35 bp · iTraxx ~23-25 bp (a record-tight zone for 2006-07; exact print unconfirmed)iTraxx above 110 bp · CDX ~200 bpIG OAS 81 bp · HY 265 bp · JPMorgan CDS 37.7 / Goldman Sachs CDS 53.4 bpn.a.A striking parallel. The 2026 high-yield spread (265 bp, FRED, 3 Sep) sits at the exact level of 2007's record tight (~240 bp): the same richness in credit pricing. Bank CDS show no visible stress, but the industry has just launched CDS on private credit and a CDX Financials index (April 2026): synthetic instruments are moving back onto the hidden risk.
📈 S&P 5001,503.35 (29 Jun 07) → peak 1,565.15 (9 Oct 07)1,251.70 (day before Lehman) · 676.53 (9 Mar 09, down 57%)7,718.60 (4 Sep) · closing all-time high 7,798.99 (13 Aug 26)×5.1Euphoria? No: a wall of worry. Up 12.75% year-to-date; a trough of 6,343.72 (30 Mar 26, the oil shock) followed by a 21.7% recovery in five months and a fresh record on 13 Aug. 86% of Q2 earnings beat estimates. But the top-ten concentration (41%) exceeds the 2007 financials peak (22.4%): the index IS the AI trade.
🏛️ DOW JONESpeak 14,164.53 close (9 Oct 07)11,421.99 (day before Lehman) · 6,547.05 (9 Mar 09, down 54%)~53,414 (all-time high above 54,000 on 4 Aug 26)×3.9A mirror image. The same trajectory as the S&P; the Dow has always historically preceded market breaks with new highs (1929, 2007). No technical top signal in 2026, quite the opposite: an all-time high just five weeks before this report.
🎯 OPTIONS & VOLATILITYVIX 13.94 (15 Jun 07) · low 9.89 (24 Jan 07) · broad volatility selling · flat skewVIX 80 to 89 (Oct-Nov 08)VIX 14.53n.a.The most quoted parallel, and the trickiest to read raw. June 2007 (13.94) and today (14.53) show the same complacency, with a war active, oil up $4 in a single session, and the yen carry trade under strain. But today's VIX is no longer quite the same instrument: it is computed from 23-37 day options, outside the 0DTE bucket, whose record volume (62.4% of SPX turnover in August 2026) dampens intraday realized volatility through dealer gamma hedging and anchors the whole short curve lower. Today's 14.5 is partly manufactured, so the raw comparison overstates the family resemblance.
Multiples such as "×6.8" compare the 2026 level to the mid-2007 level, in nominal terms, not adjusted for inflation (cumulative CPI is roughly +48% from 2007 to 2026: in real terms gold is up ×4.6, the S&P is up ×3.4, and oil is only up ~×0.9; real oil in 2026 remains BELOW its 2007 level).

3. Warning signals: 2006-2007 vs 2026

The classic crisis checklist. In the "reading" column: SIMILAR means it looks like pre-2008, DIFFERENT means it is not the same configuration.

Signal2006 to 2007Sept 2026Reading
US yield curve2s10s inverted for ~285 days (Jun 06 → Mar 07, a trough of −19 bp); 3m10y inverted for over a year; the 10-year yield peaked at 5.26% on 12 Jun 07; average lag from inversion to recession ≈ 22 months2s10s at +41 bp; 3m10y at +87 bp (4 Sep 26), the dis-inversion completed at the end of 2025; the 10-year yield is around 4.78%A similar sequence, and it is the dis-inversion that has preceded every recession. But this cycle has already produced two false positives.
Fed funds rate5.25% (Jun 06 → Sep 07), then cuts (4.75% on 18 Sep 07)3.50-3.75%; a divided Fed under Chair Warsh (9-3); JPMorgan expects a hike, Goldman Sachs expects two cuts; CPI at 3.4%A divided late cycle, but a tighter inflation constraint: 525 bp of ammunition in 2007, roughly 150 bp today. A shock with a Fed that cannot respond looks more like 1973-79 than 2008.
Bank leverageBear Stearns 36:1, Lehman ~30:1, under the SEC's 2004 ruleGoldman Sachs/Morgan Stanley SLR ~6.5% (≈15×), but the eSLR was eased in December 2025, buybacks resumed, and SLR floors hit a record low in Q1 2026A regulatory echo: leverage has moved out of banks and into hedge funds: the Treasury basis trade sits at $830bn (Fed) to $2.4tn (The Economist), with hedge-fund Treasury positions of ~$2tn (OFR).
HousingCase-Shiller down 9.1% year-on-year (Dec 07, a record); 1.3 million subprime ARM resets in 2008 aloneResidential prices up 2.1% year-on-year, median price at a record $431,000, sales at 4.05 million (weak); office CRE CMBS delinquency at 12.3%, an absolute record (worse than 2009); ~$1.8tn of CRE loan wallsResidential is different; CRE is worse, but slower. An erosion absorbed through extensions, not an identified trigger.
Consumer credit / spreadsCredit-card charge-offs ~3.6-4.7%; HY OAS at a record tight of ~240 bp (late May to June 07) → above 2,000 bp by Dec 08; Baa minus 10-year ≈ 150 bpCredit-card delinquency (30+ days) at 8.8% (below the 9.05% peak of Q2 2024), stable for two years; stress concentrated among younger and lower-income borrowers; 2026 spreads: IG OAS 81 bp · HY OAS 265 bp (FRED, 3 Sep 26), the exact level of 2007's record tightAn inverse pattern: already deteriorated but stabilised, rather than healthy-but-deteriorating.
ConcentrationTech made up ~35% of the S&P at the March 2000 peak (top ten ~27%); financials 22.4% at their 2007 peakTop ten make up ~41%; Nvidia alone 8.6%; AI capex among the big four ~$700-725bn (up 60-98% year-on-year), exceeding cash flow from Q3 26More extreme than 2000 and 2007 combined, with AI in the thematic role, this time funded by debt. But a 2000-style ending (an equity purge with sound banks) is not a 2008-style one (a credit crash): two different films, and the current readings point to the first.
Recession probabilityNY Fed model ~27% (Jul 07) → ~40%; recession began Dec 07Yield-curve model ≈ 28.9%; JPMorgan 40%, Moody's 42%; Q2 GDP +1.5%The same range, roughly 30-40% in both cases, except that 2026 has not yet tipped into recession.
Employment4.4% (spring 07) → 5.0% (Dec 07), a breaking trajectory4.1%, stable, +162,000 in August, but the composition tells a different story: healthcare and social assistance added +680,500 over 12 months, almost DOUBLE the total net job gain (+208,800); excluding publicly-funded sectors (Medicare, Medicaid), the private sector is net NEGATIVE; 2025 was revised from 585,000 down to 181,000; the 12-month average is ~31,000/month; the Sahm rule triggered in Jul 2024 WITHOUT a recession followingA plateau against a slope: the same headline level, an opposite trajectory; the 2026 plateau is being propped up by public healthcare spending, a fiscal pillar rather than a private engine.
Shocks already absorbedAlmost none before 9 Aug 2007 (two Bear Stearns hedge funds, BNP Paribas)SVB/First Republic and Credit Suisse (Mar 2023) · the yen carry trade twice (Aug 2024, VIX 65; Aug 2025, VIX ~38) · Tricolor/First Brands (2025) · the US-Iran war (2026)Already stress-tested: five shocks absorbed without contagion: real resilience, but every shock wears down the buffers.
The named "subprime"(identified too late)Private credit (JPMorgan's Dimon: "hell to pay") · AI (Michael Burry: "the final months of the bubble") · the basis trade · CRE · the Tricolor auto lenderThe same topology. The risk being named sits off banks' balance sheets and is poorly measured. Just as in 2007.

4. Asset by asset: what prices said then, what they say now

🥇 Gold, from a sideshow to the centre of gravity (×6.8)

2007: $655.75 an ounce (London fix, 15 Jun 07), a secondary market nobody talked about. It took the crisis to push it past $1,000 ($1,023.50 on 17 Mar 08). 2026: $4,436 (spot, 8 Sep, 06:22; futures $4,480-4,484), but a parabolic path: a closing peak of $5,318.40 on 29 Jan 26 (intraday $5,586.20; a low of $3,636.90 on 11 Sep 25) → down 15.8% from that peak, up 23.1% over 12 months, up 3.6% year-to-date. Central-bank buying: 244 tonnes in Q1-26 (up 17% quarter-on-quarter; Poland 64t, Uzbekistan 33t; JPMorgan sees ~755 tonnes for 2026; a World Gold Council survey found 89% of central banks expect their reserves to rise). The nominal ×6.8 multiple (×4.6 in real terms) signals a monetary regime shift. Fine, but the 2026 chart already resembles the 2008 chart: a spike, a peak, then a third given back while the rest of the market ignores the signal (between the March 2008 peak and Lehman, gold corrected ~20% before rallying again).

🛢️ Oil, a war premium ahead of the storm

2007: WTI between $65 and $72 that summer, with everyone looking elsewhere. Then $147 by July 2008: the energy bubble accompanied and then accelerated the crisis (a supply shock plus speculation plus BRIC demand). 2026: WTI $92.94 · Brent $97.59 (live, 8 Sep, Yahoo Finance: WTI up 62% year-to-date), a closing peak of $112.95 (7 Apr 26, intraday $119.48), a low of $55.27 (16 Dec 25); spikes above $100 during the US-Iran war (Wall Street Journal; Reuters, 1 Sep: up $4 in a single session); US commercial inventories below the five-year floor (EIA STEO, 11 Aug 26: it sees Brent averaging ~$87 for 2026, easing back to only $69 in 2027; the pre-conflict consensus had pointed to $62, Reuters, Nov 25). This configuration looks closer to 1973-79 (a geopolitical shock plus a Fed constrained by 3.4% inflation) than to the 2007-08 bubble. Its role as a "stress trigger", though, is identical.

🔥 Gas, the US anti-parallel, and the European shock

Henry Hub gas is at $2.95/MMBtu (8 Sep, down 20% YTD; EIA STEO projects a Q3-26 average of $2.87, with the CME curve running 2.93 to 3.86 by January 27). In 2007 the price was around $6.5-7.5 post-Katrina. US gas remains structurally abundant thanks to shale and LNG: there's no US gas bubble here. But European TTF gas is at €71.95/MWh, up 155% year-to-date, at its 52-week high (€73.63 on 2 Sep 26; a low of €26.60 on 10 Dec 25), a Hormuz-LNG shock. The US-Europe divergence (down 20% versus up 155%) is a sign that 2026's stress is geopolitical and regional, not a global demand surge.

💵 The dollar (DXY), neither crisis nor confidence

2007: around 81-82, total calm. 2026: 98.83 (live, 8 Sep, down 0.34% on the day; a full-year range of 96-102: a low of 96.22 on 27 Jan, a high of 101.61 on 24 Jun), after the greenback's worst G10 year in recent memory (2025: down 9.4%, HSBC). The dollar isn't in crisis. It's undergoing a slow erosion of status, and gold at $4,400 is its direct counterpart: the DXY's low on 27 Jan 26 coincided almost to the day with gold's peak (29 Jan). In 2008 the dollar WAS the safe haven (a flight to quality into T-bills); in 2026 the safe-haven role is shared out among gold, the franc, the yen. And it shows.

💴 The yen, the trigger pulling for a third time?

2007: USD/JPY at 122.10 on 1 Jun, a peak of 124.09 on 22 Jun 07 (FRED), the carry trade at its maximum, estimated at ~$1,000bn (CFR/Pi Economics) to $1,500bn of Japanese liabilities (Fed IFDP 899). The unwind of August 2007 was the spark for the ABCP freeze. The real "moment" was the violent unwind that took USD/JPY from 124 (Jun 07) down to ~87 (Jan 09). 2026: USD/JPY at 153.44 (live, 8 Sep, down 1.72% on the day) after peaking at 163.86 (28 Jul 26); the yen gained 2% on 3 Sep (CNBC); a record joint US-Japan intervention in July (around 160); markets are now pricing 76% odds of a BoJ rate rise in September (Tokyo Tanshi survey, up from 24% at the end of July); 10-year JGB yields ~1.95%, an 18-year high. Roughly $500bn of carry-trade positions remain open, half the 2007 stock, and they've already produced TWO global mini-crashes (August 2024: VIX at 65 intraday, TOPIX down 12%, BIS Quarterly Review No. 90; August 2025: VIX ~38). This is the only "2007 asset" whose unwind is happening in front of us right now: a late-July peak followed by a 6.3% fall in six weeks, accelerating this week.

₿ Bitcoin, the canary that didn't exist

Non-existent in 2007 (the whitepaper is dated 31 Oct 2008, in the middle of the crisis, a complete irony). 2026: ~$79,000 (Binance $79,062 · CoinMarketCap $79,311 · Coinbase $79,906), down 28% from its January 2025 all-time high (~$109,000). Its function in 2026: an institutionalised risk valve (spot ETFs). Its correction without contagion (unlike 2022) suggests part of the market's beta risk is now absorbed by this asset class. One temptation must be resisted here: calling bitcoin "the ABX index of this cycle". The ABX mattered in 2007 not because it fell, but because it was the only quoted price on a stock of illiquid assets nobody else was marking to market, held in leveraged balance sheets. Bitcoin has none of those properties: it is liquid, priced continuously and pledged as collateral almost nowhere, so a drop reveals nothing hidden. The functional equivalent in 2026 sits where the same structure sits, in private credit: listed business development companies traded at roughly 0.74 times their net asset value in March 2026, the widest discount since October 2020, and for a single loan held by several managers the valuation gap exceeds five points. That is the marked-versus-market spread of this cycle.

⚠️ Credit derivatives, 2008's heart, 2026's periphery

2007: a market of roughly $42.6tn of CDS notional outstanding at end-June 2007, peaking near $58tn by end-2007 (BIS), uncleared and bilateral. The systemic amplifier: AIG had sold hundreds of billions of dollars of protection with no capital behind it; Lehman's failure required roughly $400bn of CDS to be settled. The CDX IG index sat at ~35 bp from 2006 to mid-June 2007, and iTraxx Europe was at its historic record tight, a zone of roughly 23-25 bp (the exact print is unconfirmed). Calm seas, in other words. And the signal already existed: the ABX.HE BBB- index (subprime mortgage securities) had fallen from 97.47 to 67.27 between issuance and March 2007, with BBB- tranches down 20% as early as February 2007, a Geiger counter nobody was watching (Gary Gorton, The Panic of 2007). Bear Stearns's CDS rose 50% on 2 Aug 2007, then to ~700 bp by 14 Mar 2008. 2026: the CDS market has been reformed (central clearing, compression, notional outstanding roughly ten times smaller, though 2025 volume hit a record $41.8tn, ISDA, 29 Jul 26). AIG-style counterparty risk has been neutralised. Bank CDS in September 2026: JPMorgan 37.7 bp · Deutsche Bank 43.4 · Goldman Sachs 53.4 bp (Investing.com, 4 Sep), zero visible stress. But credit risk itself has migrated to private credit (~$1.7tn, off-index, largely untransparent), and the industry has just built the instruments to synthesise it: CDS on private credit and a CDX Financials index launched in April 2026 (Wall Street Journal / Financial Times / Reuters). In 2006, synthetic CDOs backed derivatives with hidden mortgage risk. In 2026, CDS are backed by hidden private-credit risk. IG spreads (81 bp OAS) and HY spreads (265 bp, FRED, 3 Sep 26, 2007's record-tight level) tell the same calm-seas story as June 2007.

📈 S&P 500, the same shape, a different engine (×5.1)

2007: a peak of 1,565.15 on 9 October (end of Q2: 1,503.35 on 29 Jun 07). Underneath, though, financials (22.4% of the index) were carrying record leverage. The index was literally built on the bomb. 2026: 7,718.60 (4 Sep), a closing all-time high of 7,798.99 on 13 Aug 26, a trough of 6,343.72 on 30 Mar 26 (a 21.7% recovery after the oil shock), up 12.75% year-to-date, 86% of Q2 earnings beating estimates. The top ten make up 41% of the index, with AI capex of ~$700-725bn a year exceeding operating cash flow from Q3 26 (Epoch AI). Here too, the index is built on a single thesis. The difference: the "Magnificent Seven"-style tech names generate massive real cash flow; 2007's financials generated accounting profits built on leverage. The similarity: in both cases, diversification disappeared exactly when implied volatility hit the floor.

🏛️ Dow Jones, ×3.9, the same story told heavier

A peak of 14,164.53 on 9 October 2007 → 6,547.05 on 9 March 2009 (down 54%). Today, ~53,414 (an all-time high above 54,000 on 4 Aug 26, up 10.7% year-to-date). The Dow has preceded every major break with new highs in the six to twelve months before (1929: May; 2007: October; 2026: August, five weeks before this report). Not a sell signal in itself. More a reminder that crises always start from record levels, never from lows.

🎯 Options and volatility, complacency has changed tools

2007: a VIX between 10 and 15 (a floor of 9.89 on 24 Jan 07), a flat skew, broad volatility selling, capital-guaranteed structured products backed by that sold volatility. When the shock came, implied volatility exploded (VIX 89 in October 2008) and the sellers fed the spiral. 2026: the VIX is at 14.5 to 15.3 (Cboe), with a war active and oil moving $4 in a single session. But matching levels are not the whole story: the VIX is computed from 23-37 day SPX options, outside the 0DTE bucket, and those zero-days (a record 62.4% of SPX volume in August 2026, Cboe, plus covered calls, buffer funds and retail-packaged volatility-selling products) dampen intraday realized volatility through dealer gamma hedging, anchoring the whole short curve lower. Today's 14.5 is partly manufactured, and the raw comparison with 2007's 13.94 overstates the family resemblance. Volatility selling has become the market itself: the 2026 seller is no longer a trading desk but the public, every day, automatically. That is also the fragility: on 5 February 2018, after a year spent mostly between 9 and 12, the VIX moved from 17.2 to 37.32 in one session, its largest one-day rise on record, and a listed product built on short volatility lost more than 80% that same evening. The mechanism that compresses the number is the one that releases it. Bank capital, this time, is solid.

5. A closer look at credit and derivatives: where the hidden risk sits

⚠️ 2007: credit derivatives were the gunpowder

  • ~$62tn of CDS notional outstanding at end-2007, bilateral, uncleared
  • AIG Financial Products: $527bn of protection notional sold (a toxic core of $78bn in multi-sector CDO CDS), unprovisioned, with $182bn of eventual public support (ProPublica / NBER)
  • Lehman: ~$400bn of gross CDS to be settled → an auction on 10 Oct 08, an 8.625% recovery rate, only $5.2bn net actually paid out (DTCC / Reuters); press fears at the time had pointed to $365bn
  • CDX IG at ~30 bp in 2006 (full compression) → above 200 bp after Lehman
  • The ABX subprime index was already in free fall by January 2007. The signal was READABLE, and ignored
  • SIVs and conduits: off-balance-sheet leverage backed by ABCP, frozen on 9 August 2007

🔀 2026: the gunpowder has relocated

  • CDS are now cleared and compressed, so AIG-style counterparty risk is neutralised; notional outstanding is roughly ten times smaller, but 2025 volume hit a record $41.8tn (ISDA, 29 Jul 26)
  • Bank CDS in September 2026 show no stress: JPMorgan 37.7 · Deutsche Bank 43.4 · Goldman Sachs 53.4 bp (Investing.com, 4 Sep). But April 2026 saw the launch of CDS on private credit and a CDX Financials index: synthetic instruments are moving back onto the hidden risk
  • Private credit, ~$1.7tn: off-index, marked to model. Dimon, on it: "losses are underestimated… there will be hell to pay"
  • The Treasury basis trade: $830bn (Fed) to $2.4tn (The Economist) of leveraged hedge-fund positions
  • The eSLR was eased in December 2025, and SLR floors hit a record low in Q1 2026. Bank capital went back into buybacks
  • Office commercial real estate: 12.3% CMBS delinquency, beyond the 2009 peak, managed by extending loans rather than resolved
  • Zero-days-to-expiry options: a record 62.4% of SPX volume in August 2026 (Cboe), daily, automatic retail volatility selling
  • "Interest-free" buy-now-pay-later debt, the invisible household lever: in the UAE, Tabby and Tamara let consumers pay in four to twelve instalments at 0%, meaning someone can live durably beyond their means. And until July 2026, those obligations appeared nowhere in credit reports at all (the Etihad Credit Bureau has only just started including them). The UAE BNPL market is worth roughly $5bn in 2026; Tabby is backed by $700m of JPMorgan debt financing and obtained its central bank Stored Value Facility licence in April 2026. Regulation is running behind debt that has already been taken on
The structural lesson: in 2007 as in 2026, systemic risk grows in the areas regulators don't measure. In 2007 it was banks' off-balance-sheet exposure. In 2026 it's unlisted private credit, hedge-fund leverage, volatility sold by the public, and, at the very bottom, consumer BNPL debt disguised as "0% interest". The exact parallel with 2007: the American household stacked up subprime loans that nobody was aggregating; the Gulf household stacks up "interest-free" instalments that the credit bureau itself couldn't see. The common denominator isn't an asset. It's blindness.

6. Synthesis: six parallels, six differences, three scenarios

🔴 Six parallels with 2007

  • Implied volatility at the floor (VIX 14.5) despite clear and active risks
  • Record index concentration on a single thesis (41% top ten versus 22% financials)
  • Massive leverage outside regulators' sight: shadow banking, private credit, the basis trade
  • Recent regulatory easing (the 2025 eSLR change echoes the SEC's 2004 rule)
  • A dis-inverted curve after a record inversion, plus a divided Fed late in the cycle
  • Energy rising ahead of the storm. And a yen carry trade under strain

🟢 Six major differences

  • No toxic mortgage credit; residential property is stable (up 2.1%)
  • Over-capitalised banks (SLR ~6.5% versus 30:1); centrally cleared CDS
  • Employment (4.1%) and growth (+1.5%) positive ON PAPER. But net job gains are carried by publicly-funded healthcare and social assistance (excluding which the private sector is net negative). A fiscal cushion, not a private engine
  • Five shocks already absorbed without contagion (2023-2026)
  • The Fed is CONSTRAINED: 3.50-3.75% against 3.4% inflation, limited ammunition. A 1970s-style scenario, not a 2008-style one
  • New risk absorbers: bitcoin (down 28% without contagion), ETFs, deep Treasury markets
Scenario A, "2000 all over again" (moderate probability): the right precedent is 2000, not 2008: a concentrated equity bubble bursts while the banks are solid. At the March 2000 peak, tech weighed about 35% of the index and the top ten about 27%; the 2001 recession was short, but the NASDAQ lost 78% in two years and regained its high only fifteen years later. The 2026 version: concentration is more extreme still (ten names at ~41%), the AI-driven part of the index corrects by 30-40%, the banks hold, and the recession is classic rather than financial-apocalyptic. A liquidity trigger (a full yen carry-trade unwind, a private-credit default, a major geopolitical event in Iran) and the spiral through zero-days-to-expiry options and the basis trade then act as the spark and the accelerator, not the root cause.
Scenario B, "a 1973-79 reboot" (moderate-to-high probability): the shock is energy and inflation, not credit. Oil at $100 to $150, CPI heading back towards 5%, a Fed under Chair Warsh forced to hike into a slowdown (JPMorgan's scenario is already on the table) → soft stagflation, equities moving sideways in nominal terms, gold remaining the receptacle for the regime (up 23% over 12 months, its January parabola corrected by 16%; a stagflationary twist would relaunch it).
Scenario C, "an extended soft landing" (non-zero probability): AI genuinely lifts productivity, the capex pays for itself, growth absorbs the energy shock. Concentration dilutes over time. Maybe the five shocks already absorbed were proof of resilience, not of exhaustion.
What this analysis is not: investment advice. It's a comparative reading of market configurations, with the uncertainties shown openly (volatile live data, mixed sources, unknowns flagged as such). No figure has been invented. Where we don't know, we say so.

7. Method and limits

  • Method: four independent research passes run in parallel (pre-crisis 2007 / live 2026 spot levels / CDS and options / macro signals), plus direct cross-checks of spot prices (gold, S&P, Dow, WTI, gas, DXY, VIX, bitcoin, USD/JPY). Every figure is tied to a public source.
  • The "just before 2008" benchmark: Q2-Q3 2007 (June-July), before 9 August 2007 (the ABCP freeze, BNP Paribas), the effective "start" of the crisis. Lehman Brothers' collapse (15 September 2008) is its acute phase, not its beginning.
  • Limits: live data shifts from one day to the next. Some 2026 metrics (the current Sahm rule reading, exact prime-broker leverage, the June 2026 Summary of Economic Projections) were not published in the sources we could access, and are treated as unknown. Nominal multiples are not inflation-adjusted (cumulative inflation of roughly +48% from 2007 to 2026).
  • Selection bias, stated openly: 2007 was chosen as the benchmark because it is the modern archetype of a financial crisis, but the same geometry has existed without any drama: in 2017 the VIX spent the year mostly between 9 and 12, and the accident that followed (February 2018) was a 10% correction, not a systemic crisis. A calm market is a condition that favours a crisis, not a signal that predicts one. Symmetrically, mid-2007 was calm only in equities: New Century had gone bankrupt in April, the ABX index had already collapsed in February, the Bear Stearns funds had just failed in June. 2007's calm sat in stock prices, not in credit; 2026's does too, with offices and credit cards already impaired.
  • 2007 consistency checked: mid-2007 levels for gold, WTI, Henry Hub gas, DXY, USD/JPY, the S&P 500 and the VIX were confirmed against FRED series (LBMA, EIA, the Federal Reserve Board of Governors, Cboe) in addition to press sources.

8. Main sources

2026 spot levels (final reading, morning of 8 Sep, Yahoo Finance v8): gold spot $4,436 (Stooq, 8 Sep, 06:22) · futures GC=F $4,480-4,484, a closing peak of $5,318.40 on 29 Jan 26, a low of $3,636.90 on 11 Sep 25 · WTI CL=F $92.94 (up 61.9% YTD, a peak of $112.95 on 7 Apr) · Brent BZ=F $97.59 · natural gas NG=F Henry Hub $2.95 (down 20% YTD) · TTF=F €71.95/MWh as of 4 Sep (up 155% YTD, a 52-week peak of €73.63 on 2 Sep) · DX-Y.NYB 98.83 (2026 range: 96.22-101.61) · JPY=X 153.44 (down 1.72% on the day, a peak of 163.86 on 28 Jul) · S&P 7,718.60 (4 Sep), a closing all-time high of 7,798.99 (13 Aug), a trough of 6,343.72 (30 Mar), all cross-checked against Kitco ($4,422), JM Bullion ($4,420, 7 Sep), APMEX ($4,457), S&P Dow Jones Indices (DJIA 53,414.25), Cboe (VIX 14.53 as of 4 Sep), Binance/CoinMarketCap/Coinbase (bitcoin ~$79,000), the EIA's Short-Term Energy Outlook (Henry Hub Q3-26 at $2.87; Brent averaging $87 for 2026, easing to $69 in 2027; US inventories below the five-year floor), gold.org's Q1/Q2-26 data (central-bank buying of 244 tonnes in Q1), CNBC (3 Sep 26: the yen's 2% gain, the joint July intervention, a 76% probability of a September BoJ hike), and HSBC (the dollar down 9.4% against the G10 in 2025)
2026 employment composition (independently verified): the US Bureau of Labor Statistics' "health care and social assistance employment increased by 2.9%, or 680,500, from March 2025 to March 2026" · Revelio Labs (healthcare added 410,700 jobs since Jan 25, roughly double the total net job gain of 208,800) · Bancreek (aged-care and disability services added 281,400 in 2025, more than the revised total nonfarm figure of 181,000, implying the rest of the economy was net negative) · the BLS employment situation report (a 2025 revision from 585,000 down to 181,000; a 12-month average of ~31,000/month; August 2026 at +162,000) · the Yale Budget Lab (on healthcare's role in job growth)
2026 macro data: the Federal Open Market Committee, 29 Jul 2026 (federalreserve.gov, 3.50-3.75%, a 9-3 vote) · Reuters, 12 Aug 26 (CPI at 3.4%) · FRED series T10Y3M (+0.87 as of 4 Sep 26) · YCharts (2s10s at +41 bp; recession probability at 28.9% as of 31 Aug) · a Federal Reserve note, 22 Jun 26 (the basis trade at $830bn) · the Office of Financial Research, 19 Aug 26 (hedge-fund Treasury positions at $2tn) · The Economist, 1 Jun 26 ($2.4tn) · Risk.net (record-low SLR floors) · the Federal Register (the eSLR rule change, Dec 2025) · the National Association of Realtors (sales of 4.05 million) · Trepp (office CRE at 12.3%) · the New York Fed's Household Debt and Credit report (credit-card delinquency at 8.8%) · RBC (41% concentration) · CNBC, 6 Feb 26 and 4 Sep 26 (AI capex; 4.1% employment) · the Bureau of Economic Analysis (Q2 GDP +1.5%) · BIS Quarterly Review No. 90 (August 2024) · the New York Times' DealBook, 6 Apr 26 (Jamie Dimon's letter) · CNBC, 8 May 26 (Michael Burry) · GlobeSt, 17 Apr 26 (Tricolor/First Brands) · the Wall Street Journal / Reuters, 1 Sep 26 (the US-Iran war, oil prices)
2007-2008 history: S&P and Dow Jones peaks on 9 Oct 2007 and troughs on 9 Mar 2009 (S&P Dow Jones Indices) · WTI at $147.27 on 11 Jul 2008 (EIA) · Federal Reserve FEDS Notes 2008-63 (1.3 million ARM resets in 2008) · the Financial Crisis Inquiry Commission (leverage of 36:1 at Bear Stearns, ~30:1 at Lehman, under the SEC's 2004 rule) · ISDA (CDS notional of ~$62tn in 2007) · the timeline of the ABCP freeze (9 Aug 2007) and Northern Rock (14 Sep 2007)
2007 series (FRED/Cboe): LBMA gold, $655.75 (fix, 15 Jun 07) → $1,023.50 (17 Mar 08) · FRED DCOILWTICO, WTI $65.09 (1 Jun 07) → $78.20 (31 Jul 07) → $147.27 (11 Jul 08) · FRED DHHNGSP, Henry Hub $6.40-7.89 (Jun 07) → $13.31 (2 Jul 08) · FRED DTWEXBGS, 76.70 (2007 close) → a low of 70.70 (16 Mar 08) · FRED DEXJPUS, a peak of 124.09 (22 Jun 07) → 107.51 (day before Lehman) · FRED VIXCLS, a low of 9.89 (24 Jan 07), 13.94 (15 Jun 07) · Markit's ABX.HE BBB- index, 97.47 → 67.27 · the iTraxx Main index at a zone of ~23-25 bp (2006-07, exact print unconfirmed) · BIS CDS notional at its end-2007 peak: $58tn (original publication) / $61.2tn (2018 revision) · DTCC/Reuters, October 2008 (Lehman: $5.2bn net paid out on ~$400bn gross, an 8.625% recovery rate) · ProPublica, 2010, and NBER Working Paper 21108 (AIG Financial Products at $527bn, ~$182bn of eventual support)
2000 and 2018 references: Reuters (the tech sector above 35% of the S&P 500 at the March 2000 peak, the top ten ~27%) · the NASDAQ Composite peaking at 5,048.62 on 10 Mar 2000, down 78% to its October 2002 trough, its high regained only in 2015 · CNBC and the BIS Quarterly Review (5 February 2018: the VIX from 17.2 to 37.32 at the close, its largest one-day rise on record; an inverse listed volatility product down more than 80% the same evening and wound up shortly after) · Cboe (VIX methodology: 23-37 day SPX options)
2026 credit: FRED series BAMLC0A0CM / BAMLH0A0HYM2 (IG OAS 81 bp · HY 265 bp as of 3 Sep 26) · ISDA, 29 Jul 26 (2025 CDS volume: $41.8tn, a record) · Investing.com, 4 Sep 26 (five-year CDS: JPMorgan 37.7 · Deutsche Bank 43.4 · Goldman Sachs 53.4 bp) · the Wall Street Journal / Financial Times / Reuters, April 2026 (the launch of CDS on private credit and a CDX Financials index) · Cboe (zero-days-to-expiry options at 62.4% of SPX volume in August 2026) · Gulf BNPL: Arabian Business, on BNPL debt being added to Etihad Credit Bureau reports from July 2026 · ResearchAndMarkets (the UAE BNPL market at ~$5bn in 2026, Tabby's Stored Value Facility licence in April 2026, its $700m JPMorgan debt facility)
This page is an informational analysis. It's not investment advice.

Analysis by I am in.Digital · September 2026
Status: final. The 2007 (FRED/Cboe), 2026 spot, CDS/options and macro research streams are consolidated. Unknowns are marked as such.