Gekko & Fox reads the market every Sunday: a state for the S&P 500, eleven sectors and the layers of the AI stack, computed by a published method. How it works

The weekly market state

AI hardware breaks down while staples and health care take the lead

The market's structural state has slipped into Risk-Off at −27, and AI hardware is where it broke. Defensives took the bid, and health care holds the only Risk-On read on the board.

S&P 500

against T-bills

Tactical (next 2–6 w)

−11Balanced−3

23rd percentile of 30 weeks

90

Tactical · S&P 500, 30 weeks
+100+24−24−100
Mar 29Oct 18

Structural (next 3–9 m)

−27Risk-Off−13

26-week low

100

Structural · S&P 500, 30 weeks
+100+24−24−100
Mar 29Oct 18
Read the call

Three things to know

  1. 01The structural state fell 13 points to −27, Risk-Off and its lowest in 26 weeks, with structural breadth at −49 doing most of the damage.
  2. 02AI hardware led the board lower: networking tactical fell 40 points to −37, the largest drop on the board, and semis, memory, networking and data centers all sit at their lowest structural readings in 30 weeks.
  3. 03Defensives took the bid: staples tactical rose 34 points to +18, its highest in 30 weeks, and health care structural at +37 is the only Risk-On read on the board.

The S&P 500

Structural state at −27, its lowest in 26 weeks

−21 −49
−32 −26
−17 −27
−9 −11
−6—
+39—

Positioning is contrarian: a positive score means leveraged funds are more short, or less long, of S&P 500 futures than usual, which the method reads as supportive. A negative score is crowded longs.

The two horizons read the same market through different sub-metrics, so one week gives two sets of components. Tactical trend is price against the 50-day average, structural against the 200-day; tactical relative strength is the 1- and 3-month return, structural the 6-month; breadth is the share of S&P 500 members above their 50-day average (tactical) or 200-day (structural). Momentum quality and positioning are tactical only. Weights: tactical trend 25, relative strength 30, breadth 15, momentum 10, liquidity 15, positioning 10; structural trend 25, relative strength 30, breadth 20, liquidity 25. Every sub-metric is scored against its own trailing distribution, clipped at ±2.5 standard deviations and scaled to ±100; a component is the mean of its sub-metrics.

The crowd

AAII Investor Sentiment Survey · Wed, Oct 14

Bulls 38.7% · Neutral 19.7% · Bears 41.6% · spread −2.9· long-run average +6.0

AAII Investor Sentiment Survey, week to Wed, Oct 14 (published Thu, Oct 15): 38.7% bullish, 19.7% neutral, 41.6% bearish — a bull–bear spread of −2.9 points, against AAII's long-run average of +6.0 (37.5% bullish, 31.5% bearish); +1.3 the week before. AAII Investor Sentiment Survey — displayed, not scored.

The tape

Tactical −11 Balanced with conviction 90 at p23: breadth −21, trend −17, momentum −6, rs −9 and liquidity −32 all point down, and positioning at +39 is the lone support, reflecting funds net short 21.1% against a 17.2% mean. Structural −27 Risk-Off with conviction 100 at p13, lowest in 26 weeks, led by breadth at −49.

The story

Investors are weighing expected Q3 blended earnings growth near 29.5% [3] against 3.7% CPI [1] and a 5.09% 10-year [14], and took comfort from Fed signals that an October hike is not urgent [2].

The gap

The earnings story is broad; the tape is narrow, with structural breadth at −49 saying fewer stocks are participating than a 29.5% growth quarter implies.

What would change the view
Risk-On Market S&P 500 Structural state recovers to −20 or better, back into Balanced. structural ≥ −20 · Fri −27 7 pts away Risk-Off Market S&P 500 Tactical state falls to −21 or lower, joining the structural read in Risk-Off. tactical ≤ −21 · Fri −11 10 pts away

What changed · week to Fri, Oct 16

AI hardware sold off hardest while defensives took the bid

AI hardware sold on financing, not demand

A Goldman Sachs basket of Asian AI hardware stocks fell 11% on the week, its steepest decline since mid-July, as investors repriced the supply chain against high borrowing costs [8]. The US layers followed: networking tactical fell 40 points to −37, semis 32 to −34, compute 30 to −25 and memory 22 to −36. This came in the same week that memory contract prices kept rising [6][7], so the selling is about funding and valuation rather than end demand.8 6 7
On the week AISemiconductors & Equipment−32AIMemory & Storage−22AINetworking & Optics−40AICompute (accelerators)−30AIData Centers & Neoclouds−13

The spenders kept spending

Oracle guided net capital spending of up to $70 billion for fiscal 2027, from about $48 billion, with gross capex of $90-95 billion a year [4]; Meta narrowed its 2026 range to $130-145 billion [10] and TSMC held its $60-64 billion budget on demand it calls extremely robust [12]. Microsoft's trim to about $175 billion is an accounting change, not a cut [11]. Hyperscaler capex near $800 billion is increasingly debt- and equity-funded, with combined liabilities above $4 trillion [9], which is the part the tape is pricing.4 10 11 12 9
On the week AIAI & Hyperscalers−17AIData Centers & Neoclouds−13AIAI Software−15AISemiconductors & Equipment−32

Hot CPI, lower long yields

September CPI rose 0.6% on the month and 3.7% on the year, with core at 2.6% [1], yet the 10-year yield fell 15 bp to 5.09% and the 2-year 6 bp to 4.74% [14] as Fed officials signalled no urgency to hike in October [2]. Rate-sensitive sectors lifted from low bases: utilities tactical rose 29 points, financials 23 and real estate 18.1 14 2
On the week MarketS&P 500−3SectorFinancials+23SectorUtilities+29SectorReal Estate+18

Rotation into staples and health care

Staples rose 34 points tactically to +18 and 36 points structurally to +19, the largest structural gain on the board, while health care structural rose 21 points to +37. September retail sales rose 0.8% on the month [17], and discretionary tactical gained 12 points to −6, but its structural state stays Risk-Off at −46.17
On the week SectorConsumer Staples+34SectorHealth Care+16SectorConsumer Discretionary+12

The sectors · all 11

Only health care holds a Risk-On read

Every GICS sector through its Select Sector ETF, read on the same scale as the S&P 500. A row opens to the week’s note and the AI layers that draw on the sector; a row with neither stays as it is.

Financials XLF −17+2380 −4+1570

Note

Structural +37 is the only Risk-On read on the board, at p98 of 30 weeks on rs +71 and breadth +69. Tactical +20 after a 16-point gain puts it at the edge of Risk-On on that horizon too.

Note

No note on this sector in this issue.

In the AI stack

AI 01 AI & Hyperscalers −2

Note

Tactical rose 34 points to +18, its highest in 30 weeks, and structural rose 36 to +19, the largest structural gain on the board. Breadth of +40 and +51 says the move is broad, not a few names.

Note

No note on this sector in this issue.

In the AI stack

AI 07 Power −28

Note

Structural −65 Strong Risk-Off at p10, with breadth at −93 and trend at −79. A 9-point tactical gain still leaves it at −30 Risk-Off.

In the AI stack

AI 06 Data Centers & Neoclouds −39 AI 07 Power −28
Materials XLB −6+2555 −17+3280

Note

No note on this sector in this issue.

In the AI stack

AI 07 Power −28

Note

No note on this sector in this issue.

In the AI stack

AI 06 Data Centers & Neoclouds −39

Note

No note on this sector in this issue.

In the AI stack

AI 01 AI & Hyperscalers −2 AI 08 AI Software −6

Scale −100 to +100 in five bands; a state inside ±24 is Balanced and makes no call.
Marks: Tactical, Structural, Balanced; Conviction is the share of the weight on the state’s side, 0 to 100, under each state.
The short line behind a mark is the previous issue’s position.
Source: State Street Select Sector ETFs, daily closes via Massive; breadth over each sector’s S&P 500 constituents.

The AI stack

Spenders lift capex while four hardware layers hit 30-week lows

A cross-cut of the market, not a sector: eight layers of the AI economy in dependency order, each an equal-weight basket of US-listed names drawn from across the index and beyond it. The order is the chain, so it is never sorted.

Layer 04 of 08· Hardware

Memory & Storage

Tactical

−36Risk-Off−22

9-week low

90

Structural

−55Risk-Off−12

Lowest on record (30 weeks)

100

Basket · 7 names

Draws from

Sector Technology 7 names · −18
The gap

The widest gap on the stack: prices are rising while the structural state sits at its lowest in 30 weeks. The tape is pricing either a cycle peak or a financing-led capex slowdown, and the board does not say which.

The tape

Tactical −36 Risk-Off after a 22-point fall, at p7 and lowest in 9 weeks: rs −52, breadth −66, trend −39, momentum −27. Structural −55 Risk-Off with conviction 100 at p0, lowest in 30 weeks, with breadth at −100.

The story

The pricing cycle is strong: HBM3E contracts at $1.50-1.60 per gigabit, HBM4 near $2.00 [6], conventional DRAM forecast up 10-15% and NAND 15-20% in Q4, and DDR4 up 24% over three months [7].

What would change the view
Risk-On AI Memory & Storage Memory tactical momentum turns non-negative, from −27 to 0 or better. tactical momentum quality ≥ 0 · Fri −27.1 27 pts away Risk-Off AI Memory & Storage Memory structural state falls to −60 or lower, into Strong Risk-Off. structural ≤ −60 · Fri −55 5 pts away
−66 −100
−32 −26
−39 −61
−52 −42
−27—
+39—

What to watch

Semis sit two points from a Strong Risk-Off read

Triggers closest to firing

Risk-Off AI Semiconductors & Equipment Semis structural state falls to −60 or lower, into Strong Risk-Off. structural ≤ −60 · Fri −58 2 pts away
Risk-On AI Data Centers & Neoclouds Data centers structural state recovers to −59 or better, out of Strong Risk-Off. structural ≥ −59 · Fri −61 2 pts away
Risk-Off AI AI & Hyperscalers AI & Hyperscalers structural state falls to −21 or lower, into Risk-Off. structural ≤ −21 · Fri −16 5 pts away
Risk-On AI Compute (accelerators) Compute tactical state recovers to −20 or better, back into Balanced. tactical ≥ −20 · Fri −25 5 pts away
Risk-Off AI Memory & Storage Memory structural state falls to −60 or lower, into Strong Risk-Off. structural ≤ −60 · Fri −55 5 pts away
Risk-Off AI Data Centers & Neoclouds Data centers tactical state falls to −45 or lower. tactical ≤ −45 · Fri −39 6 pts away
Risk-On AI AI & Hyperscalers AI & Hyperscalers tactical momentum turns non-negative, from −7 to 0 or better. tactical momentum quality ≥ 0 · Fri −6.8 7 pts away
Risk-On Market S&P 500 Structural state recovers to −20 or better, back into Balanced. structural ≥ −20 · Fri −27 7 pts away
Risk-Off AI Power Power structural state falls to −50 or lower. structural ≤ −50 · Fri −43 7 pts away
Risk-On AI Power Power tactical state recovers to −20 or better, back into Balanced. tactical ≥ −20 · Fri −28 8 pts away
Risk-Off Market S&P 500 Tactical state falls to −21 or lower, joining the structural read in Risk-Off. tactical ≤ −21 · Fri −11 10 pts away
Risk-Off AI Networking & Optics Networking tactical state falls by 10 points or more again by 25 Oct. weekly change in tactical ≤ −10 · on the week −40 10 pts away
Risk-Off AI Compute (accelerators) Compute structural state falls to −21 or lower, into Risk-Off. structural ≤ −21 · Fri −10 11 pts away
Risk-Off AI AI Software AI Software structural state falls to 0 or lower, losing its positive read. structural ≤ 0 · Fri +12 12 pts away
Risk-On AI Semiconductors & Equipment Semis tactical state recovers to −20 or better, back into Balanced. tactical ≥ −20 · Fri −34 14 pts away
Risk-On AI Networking & Optics Networking tactical state recovers to −20 or better, back into Balanced. tactical ≥ −20 · Fri −37 17 pts away
Risk-On AI AI Software AI Software tactical trend turns non-negative, from −18 to 0 or better. tactical trend ≥ 0 · Fri −18 18 pts away
Risk-On AI Memory & Storage Memory tactical momentum turns non-negative, from −27 to 0 or better. tactical momentum quality ≥ 0 · Fri −27.1 27 pts away

The week ahead

Open Mon–Fri, 9:30 AM–4:00 PM · no holidays

  1. Throughout US Treasury par yield curve. The 10-year fell 15 bp to 5.09% in the week to 16 Oct; whether it holds decides if the rate-sensitive lifts in utilities and real estate extend.
  2. Throughout Q3 earnings season continues. Expected blended growth near 29.5% is the story behind a tape with structural breadth at −49.
  3. Thu, Oct 22 US weekly initial jobless claims. A labour read ahead of the FOMC, with a market still weighing a hike after 3.7% CPI. After the close, Thu, Oct 22
  4. Fri, Oct 23 CFTC Commitments of Traders. Funds are net short 21.1% against a 17.2% mean; a cover would cut the +39 positioning support across every area. After the close, Fri, Oct 23
  5. Wed, Oct 28 FOMC decision at the end of the 27-28 October meeting. Warsh gave no signal; a hike would hit liquidity, already −32 tactical and −26 structural across the board.

Risk map

Every area at the close of Fri, Oct 16

Sunday's states as published, both horizons; the change is against the previous issue's board; conviction is the share of the horizon's weight on the state's side.
TacticalStructural
AreaStateBandOn the weekConvictionStateBandOn the weekConviction
S&P 500 −11Balanced−390−27Risk-Off−13100
Technology −18Balanced−3190−7Balanced−3845
Financials −17Balanced+2380−4Balanced+1570
Health Care +20Balanced+1685+37Risk-On+2175
Consumer Discretionary −6Balanced+1255−46Risk-Off+1100
Consumer Staples +18Balanced+3485+19Balanced+3675
Energy +18Balanced+1385+9Balanced+1355
Industrials −30Risk-Off+980−65Strong Risk-Off+10100
Materials −6Balanced+2555−17Balanced+3280
Utilities −7Balanced+2965−54Risk-Off+20100
Real Estate −28Risk-Off+1880−36Risk-Off+7100
Communication Services −20Balanced+290−40Risk-Off+6100
AI & Hyperscalers AI−2Balanced−1740−16Balanced−1780
Compute (accelerators) AI−25Risk-Off−3090−10Balanced−2845
Semiconductors & Equipment AI−34Risk-Off−3290−58Risk-Off−20100
Memory & Storage AI−36Risk-Off−2290−55Risk-Off−12100
Networking & Optics AI−37Risk-Off−4090−70Strong Risk-Off−28100
Data Centers & Neoclouds AI−39Risk-Off−1390−61Strong Risk-Off−6100
Power AI−28Risk-Off−1790−43Risk-Off−14100
AI Software AI−6Balanced−1565+12Balanced−2250

Liquidity and rates

  • Credit appetite −1.45% · −0.98% · +0.36% high yield minus investment grade · the week to Fri, Oct 16, the month from Wed, Sep 16, the quarter from Fri, Jul 17
  • Dollar −0.34% · +1.83% · +2.08%
  • Long bonds +2.65% · −1.03% · −5.29%
  • Treasury curve not available — No daily was published for the 2026-10-16 close.
  • Positioning CFTC Traders in Financial Futures, S&P 500 consolidated, report dated Tue, Oct 13 (published Fri, Oct 16): leveraged funds net short 21.1% of open interest, from net short 21.3% the week before. It stands in every state.
  • The crowd AAII Investor Sentiment Survey, week to Wed, Oct 14 (published Thu, Oct 15): 38.7% bullish, 19.7% neutral, 41.6% bearish — a bull–bear spread of −2.9 points, against AAII's long-run average of +6.0 (37.5% bullish, 31.5% bearish); +1.3 the week before. AAII · displayed, not scored
  1. 1 September CPI hotter than expected, fuels inflation concerns bls.gov · Wed, Oct 14https://www.bls.gov/news.release/cpi.nr0.htm
  2. 2 Fed Chair Warsh silent on October rate hike; market eyes hold economictimes.indiatimes.com · Fri, Oct 16https://economictimes.indiatimes.com/markets/us-stocks/wall-street-guide/us-market-fed-chair-warsh-offers-no-signal-on-october-interest-rate-decision/articleshow/134984000.cms?from=mdr
  3. 3 Q3 earnings season kicks off with major banks reporting clarin.com · Sun, Oct 11https://www.clarin.com/us-news/the-stock-market-this-week-october-1216-big-bank-earnings-september-cpi-and-retail-sales-to-test-record-rally_0_h4cVcTwKuu.html
  4. 4 Oracle projects significant increase in FY2027 capital spending seekingalpha.com · Thu, Oct 15https://seekingalpha.com/article/4954480-oracles-ai-selloff-is-creating-a-rare-setup
  5. 5 Meta launches new "Muse Wonder" AI model with subscription revenue focus mediapost.com · Tue, Oct 13https://www.mediapost.com/publications/article/418689/meta-launches-ai-model-called-muse-wonder.html
  6. 6 High-Bandwidth Memory (HBM) prices continue upward trend capitalandcompute.net · Fri, Oct 16https://capitalandcompute.net/memory-prices/hbm/
  7. 7 DRAM and NAND prices expected to rise significantly in Q4 2026 insightorys.com · Tue, Oct 13https://insightorys.com/2026/10/13/ram-prices-october-2026-buy-now-or-wait/
  8. 8 Asian AI hardware stocks experience sharp selloff amid financing concerns briefs.co · Sun, Oct 18https://www.briefs.co/news/tech-selloff-puts-asia-s-ai-supply-chain-on-notice-as-borrow/
  9. 9 Hyperscaler capex approaches $800 billion, raises funding and debt concerns seekingalpha.com · Tue, Oct 13https://seekingalpha.com/article/4953737-several-potential-threats-that-could-pop-the-ai-bubble
  10. 10 Meta narrows 2026 capex guidance amid AI spending focus seekingalpha.com · Fri, Oct 16https://seekingalpha.com/article/4954820-meta-the-capex-problem-is-starting-to-look-like-a-strategy
  11. 11 Microsoft adjusts 2026 capex forecast due to accounting change finance.biggo.com · Sat, Oct 17https://finance.biggo.com/news/931470d5-0cd8-417a-9650-34b82fe9b585
  12. 12 TSMC reiterates robust AI demand and capital budget for 2026 finance.biggo.com · Sat, Oct 17https://finance.biggo.com/news/931470d5-0cd8-417a-9650-34b82fe9b585
  13. 13 IMF and World Bank highlight AI's role in global economic growth nationthailand.com · Thu, Oct 15https://www.nationthailand.com/business/economy/40072336
  14. 14 US Treasury yield curve, daily par yields home.treasury.gov · Fri, Oct 16https://home.treasury.gov/resource-center/data-chart-center/interest-rates/TextView?type=daily_treasury_yield_curve
  15. 15 Federal Reserve: Minutes of the Board's discount rate meetings on August 17, September 9, and September 16, 2026 federalreserve.gov · Tue, Oct 13https://www.federalreserve.gov/newsevents/pressreleases/monetary20261013a.htm
  16. 16 Census Bureau: Manufacturing and Trade Inventories and Sales census.gov · Thu, Oct 15https://www.census.gov/mtis/index.html
  17. 17 Census Bureau: Advance Monthly Sales for Retail and Food Services census.gov · Thu, Oct 15https://www.census.gov/retail/index.html
  18. 18 Census Bureau: Business Formation Statistics census.gov · Wed, Oct 14https://www.census.gov/econ/bfs/

Methodology v1 · universe v2 · published Tue, Oct 20, 5:48 PM UTC by Dario Dario AB · percentiles read over the 30 prior Sundays the issue could see