degen.gauges

Mon, Sep 14, 2026
latest
SHUT
Legs basing
2/6 basing
VIX settling
17.1
Credit calm
normal
STRC < 90 · $98.9 — de-risk STRC < 80 · cut hard VIX > 22 · at 17.1 CTA breach · +4.2% away VVIX > 100 · 94.9

The strangest risk-off catalyst of the cycle arrived this weekend: the AI labs themselves pumped the brakes. Anthropic's CEO published an essay urging the industry to deliberately pace capability advancement so safety work can keep up; OpenAI's CEO endorsed the idea, called a 2026 IPO "ill-advised" (pushing one of history's most anticipated listings to 2027 at the earliest), and floated a coordinated inter-lab pact to slow frontier development. The market's response was not a panic — it was a sorting. The AI-capex hardware complex was repriced in one session (semiconductor ETFs down 5-6% Monday, memory and optics names worse, data-center infrastructure equities now more than 20% off their highs) while the software layer above it ripped: security names most violently (CrowdStrike +14%, Zscaler +16%, Palo Alto +13% on the straightforward logic that a more dangerous AI world expands security budgets), and the enterprise-software rerating cohort with them. Even the Mag7 split down the middle — the platform names caught the software bid at fresh highs while the chip-and-capex names wore the tax. In real time, the tape is dividing "the AI trade" into three different trades: hardware capex (sold), platforms and software (bid), and the private labs themselves (now self-throttling).

The gauges say stress, but a narrow kind. The one flashing panel is rate volatility — the MOVE index at its 93rd percentile — because markets now price roughly nine-in-ten odds that the Fed HIKES at Wednesday's meeting, with meaningful odds of another by year-end. Credit could hardly be calmer: high-yield spreads sit near their tightest decile, the crypto-credit canary this dashboard tracks is fully healed, and the vol curve remains in contango at a 17-handle VIX. By this framework's rules, calm credit plus stressed positioning equals correction mechanics, not crisis mechanics. But the positioning math deserves respect: S&P breadth halved again in two weeks (50% to 39% of members above their 50-day), fear-and-greed sits at 31 with its breadth and strength components at extreme fear, tail-hedge demand is at the 90th percentile, and the index trades just 4% above the level where trend-following systems flip to mechanical selling — while volatility-targeting strategies remain near max-long. The fast money already de-grossed; the slow machines have not. That gap is the remaining air pocket, and a binary Fed decision lands in the middle of it this week.

Two more threads worth tracking. Oil jumped 10% in five sessions as the war premium reflated, with gold at $4,330 and tanker rates squeezing — geopolitics is quietly funding one side of every macro book again. And the seasonal pattern strategists keep citing — midterm-year Septembers are historically the worst stretch of the four-year cycle, with recoveries typically beginning in October — is so far tracking to script. The posture that follows from all of it: this is the setup phase, not the deployment window. Let the mechanical sellers show whether they get triggered, let Wednesday's Fed decision clear, buy the flush rather than the forecast — and treat the software strength as validation to be added on a reset, not chased on a +14% day.

Sources: Anthropic pacing essay and industry response (CNBC, 9/12: cnbc.com/2026/09/12/anthropics-amodei-proposes-plan-to-slow-the-pace-of-advancing-ai-capabilities.html) · OpenAI IPO delay (Fortune, 9/14: fortune.com/2026/09/14/openai-ipo-wait-grows-longer-as-ai-safety-fears-mount-cfo · Axios, 9/12: axios.com/2026/09/12/openai-public-ipo-delay-sam-altman) · Cybersecurity sector bid (Benzinga, 9/14: benzinga.com/trading-ideas/movers/26/09/61763825 · Seeking Alpha, 9/14: seekingalpha.com/news/4642523) · Fed hike odds per CME FedWatch pricing, 9/14.

  • VIX 15 → 17
  • SPX breadth 50% → 39%
  • F&G 50 → 31
  • STRC 96.9 → 98.9
  • legs basing 4 → 2

Macro regime

risk-on
1/6
Credit (HY OAS)2.65%252d pctile 4%, 1m -0.06pp, tightening
Rate vol (MOVE)84252d pctile 93%
Equity vol (VIX TS)VIX 17.1 / 3M 19.3 → contango
Fin. conditions (NFCI)-0.56looser than average
Real rate (10y TIPS)2.60%1m +0.18pp
Breadth (RSP/SPY)50d -1.3%narrowing

Mechanical label can mislead while internals rot — read the signals, not the headline.

Fear & Greed

contrarian
feargreed
31
fear
MomentumStrengthBreadthPut/CallVolatilitySafe-havenJunk-bond
does revenue show up first?

AI ROI coverage

14%of capex covered
exogenous 9% · ~35% circular
Lab ARR$54B+157%
vs capex$400B/yr+60%
Exogenous9%
Circular (NVDA→OpenAI→Azure)~35%

ARR vs capex is Clock A's numerator. Exogenous-vs-circular is the honesty check. asof 2026-06-22.

AI-infra demand

$0.20cheapest frontier $/Mtok
median $6.00
Cheapest frontier$0.20/Mtok
Median frontier$6.00/Mtok
Frontier-class models128
Total models445

Price (the Jevons denominator) only. Falling = intelligence commoditizing; volume must outrun it.

Consumer (the demand base)

+1.7%spend-over-income gap
savings 3.0%
Real PCE YoY+2.1%
Real DPI YoY+0.5%
Gap (spend - income)+1.7%
Savings rate3.0%
Revolving credit YoY+3.6%
CC delinquency2.85%-0.19pp/yr
Debt service / DPI11.2%+0.06pp/yr
Initial claims206k-24k/qtr

Spend>income + low savings + rising debt-service = the consumer-funded leg is stretched.

Distribution (K-shape)

+2.4%productivity-pay wedge → capital
labor share -3.4% YoY
Productivity YoY+2.2%
Real pay YoY-0.1%
Wedge → capital+2.4%
Labor share (2017=100)93.4-3.4% YoY
Corp profits YoY+28.2%

Gains to capital income-cap the demand base. The K-shape slows Clock A (ROI).

Labor / jobs

-0.07Sahm rule (≥0.5 = recession)
unemployment 4.1% · firm
Unemployment4.1%-0.2pp/yr
Payrolls MoM+162k
Quits rate1.9%
Job openings7,271k
Sahm value-0.07
IT-services emp YoY-1.3%

jobs = the consumer income engine (Clock A); Sahm rule = the recession trigger. Tech line is IT-services employment (CES6054150001) — a substitution *hint*, not proof.

the leading edge / the fuse

Crypto / AI-infra credit

$98.91STRC (par 100)
-1.1% vs par · normal
STRC discount-1.1%
Strategy prefs3/4 below par-0.6% 5d
MSTR vs BTC (21d)+42.0%
BTC$78,168
MSTR$136.94

The leading credit edge. STRC<90 falling = de-risk; <80 = cut hard. Dress rehearsal for AI-infra credit.

Credit quality ladder

10.0ppCCC-IG dispersion
leaking (bottom edge)
IG OAS0.80%+0.01pp/mo
BB OAS1.50%
HY OAS2.65%
CCC OAS10.76%+0.52pp/mo
Private-credit / BDC-3.4% off-hi-1.2% ex-SPY
Lev loans (BKLN)+0.0% off-hi
Regional banks (KRE)-4.9% off-hi-2.7% ex-SPY

corporate quality ladder (IG→CCC, FRED) + the levered/shadow-bank edge (de-beta'd). CCC + private credit cracking while IG/banks calm = early/confined; IG widening or banks breaking = systemic.

Funding plumbing

-3bpSOFR - IORB
buffer drained
SOFR3.62%
IORB3.65%
SOFR - IORB-3bp
RRP buffer$1B+1B/mo
Bank reserves$2.99T+47B/mo

A plumbing leak is a different failure mode than spreads; SOFR spiking >IORB = the 2019 repo channel.

Private credit / AI-infra debt

-22.9%worst basket off-hi
cracking
Private credit (BDCs)-6.9% off-hi-3.8% 5d
PC worst nameOWL -18%
AI-infra debt-22.9% off-hi-9.7% 5d
Infra worst nameVRT -34%

de-beta'd equity proxy for the shadow-bank / AI-infra-debt edge (CDS/CLO/NAV are paywalled). Infra basket (ORCL/VRT/DLR) is tech-multiple beta, NOT debt stress. Confirms credit_stress/funding; never a standalone trigger.

Neocloud watch

-39.6%basket off-hi
9/9 cracking >15% · cracking
BRUN-62.5% off-hi-17.4% 5d
APLD-47.2% off-hi-6.8% 5d
WULF-46.6% off-hi-6.3% 5d
CIFR-46.4% off-hi-11.9% 5d
CORZ-41.8% off-hi-5.2% 5d
CRWV-29.6% off-hi-7.1% 5d
IREN-29.1% off-hi-3.4% 5d
HUT-26.8% off-hi-2.6% 5d
NBIS-26.0% off-hi-6.3% 5d

levered GPU-cloud operators (CRWV/IREN/…) — the sharpest, most faith-dependent Clock-B edge; cracks first. De-beta'd vs SPY. Bifurcation = name-specific.

how violent the unwind, if it comes

Momentum / crowding

2/6basing
Momentum factor
MTUM/SPY
-0.5%
Mag7 mega-cap
MAGS/SPY
+2.2%
Semis
SMH/SPY
-3.3%
High-beta/low-vol
SPHB/SPLV
-2.4%
Breadth (eq/cap)
RSP/SPY
-0.6%
Growth/value
VUG/VTV
+0.8%
off-hi · 5d basingVIX 17.1 · VVIX 94.9

off-hi = unwind so far · run63 = fuel left · 5d ≥ 0 = basing. Dip-buy needs legs basing.

CTA systematic flows

+4.2%to short
SPX spot 7,620
spot
short+4.2%
medium+8.6%
long+15.3%

Breach = systematic supply ON. Levels asof 2026-06-09.

SPX breadth

n=502
% above 50dma39%
% above 200dma59%

The load-bearing breadth measure (n≈500).

Retail froth (payload size)

+46.6%margin debt YoY
$742B
Margin debt$742B+46.6% YoY
High-beta SPHB/SPLV-9.8% off-hi-2.4% 5d
2x-ETF casino off-hi-22.0%
2x-ETF casino 5d-7.6%

The payload size, not the fuse — froth amplifies the move; credit + ROI trigger the break.

Mag7 concentration

5/7above 50dma
color only — not breadth
AAPL333.08+0.2%
MSFT505.41+2.0%
GOOGL349.39+3.2%
AMZN253.54-1.3%
NVDA210.96-3.4%
META665.60+2.7%
TSLA358.97-1.8%

n=7 is not breadth — the breadth measure is the SPX panel.

Buffett indicator

256%market cap / GDP
valuation backdrop
Total mkt cap / GDP256%

Valuation backdrop — magnitude, not a trigger.

Cross-asset tape

SKEW 152tail-hedge demand
pctile 90%
DXY99.47
Gold$4,330
BTC$78,168
Copper$6.39

Macro cross-asset backdrop.

Memory super-cycle

+40-50%3Q26 QoQ forecast
vs consensus +15-20%
3Q26 forecast+40-50%
Consensus+15-20%
Cycle-top marker2028 ASP fall — China NAND catch-up + 15-20% supply growth; CSP LTAs at 50% of capacity (could go 70%), squeezing consumer
Latest print2026-06-24

Contract-price read vs the super-bull call. The crux gauge for the memory leg.

Memory tape (Korea / EWY)

EWY 176.2Korea-beta canary · memory tilt
-6.6% 1d
EWY176.22
1d-6.6%
5d-6.7%
Off 63d high-19.6%

broad Korea/memory tilt (EWY) — a Korea-beta & Asia-risk canary, NOT the memory duopoly; direct makers live in the maker panel.

Makers (supply oligopoly)

-29.4%basket off-hi ex-SPY
raw -31.5% · n=7
Basket off-hi-31.5%
Excess vs SPY-29.4%
Avg 5d-7.6%
Worst name285A.T -53%

semicap/memory supply oligopoly (Samsung/Hynix/TSMC/ASML/Infineon/MU) — EQUITY-BETA health (de-beta'd vs SPY), NOT supply tightness; can't see CoWoS lead times / HBM allocation (no free feed).