Macro & Market Structure

Ten Themes — Map Room

Market data 2026-10-16
Yield curve: US Treasury · Factors: 1y daily
Descriptive view — no trade conclusions
← Desk · Grant map · Backtest · Winner →
ratesstable
8.0 bp 1m
dollarstrengthening
1.9 % 1m
volelevated
82.0 pctile 1y
creditrisk-off
-0.7 % 1m
copperflat
-0.4 % 1m

1 · Macro Scorecard

Level, trailing changes, and position within the one-year range for each market-based macro indicator. The marker on the range bar shows where today sits between the 1-year low and high.

VIX
Equity implied volatility
19.99
1W+34.7%1M+29.5%3M+17.2%1Y+22.1%
13.47 — 31.05 · 82nd pctile
Dollar (UUP)
US dollar index, via the UUP ETF that tracks DXY futures
28.92
1W-0.3%1M+1.9%3M+1.9%1Y+3.9%
26.47 — 29.13 · 96th pctile
Copper (CPER)
Copper futures, via the CPER ETF — industrial demand proxy
39.50
1W-2.0%1M-0.4%3M+2.8%1Y+28.5%
30.53 — 41.43 · 79th pctile
WTI Crude (USO)
Front-month crude oil, via the USO ETF
161.76
1W+9.2%1M+4.2%3M+28.9%1Y+134.5%
66.17 — 161.86 · 100th pctile
Gold (GLD)
Gold, via the GLD ETF
382.01
1W-0.7%1M-4.1%3M+3.9%1Y-1.4%
362.32 — 495.90 · 25th pctile
Credit Appetite (HYG/LQD)
High-yield relative to investment grade — rising means risk appetite improving
0.74
1W-1.4%1M-0.7%3M-0.0%1Y+3.1%
0.72 — 0.76 · 75th pctile
Bitcoin (IBIT)
Crypto risk appetite, via the IBIT spot bitcoin ETF
46.15
1W-0.9%1M+6.6%3M+25.1%1Y-26.9%
33.29 — 65.28 · 62nd pctile
TLT 20y+ Treasury
Long-duration Treasury price
80.05
1W+2.6%1M-2.1%3M-4.6%1Y-11.7%
77.11 — 92.06 · 7th pctile

Treasury curve

Current par yields against one month and one year ago. 2s10s at 0.35%, 10-year 8 bp over the past month and sitting at the 94th percentile of its 1-year range.

TenorNow1M agoΔ bp1Y ago
3M 4.26% 4.14% +12% 4.02%
2Y 4.74% 4.74% +0% 3.48%
5Y 4.89% 4.86% +3% 3.60%
10Y 5.09% 5.01% +8% 4.03%
30Y 5.43% 5.35% +8% 4.62%

Curve shape

2 · Macro Factor Sensitivity

Univariate regression of each theme's equal-weighted daily returns on each factor, over the trailing year. Cells show beta with R² beneath. Rates and curve betas are expressed as percent move per 10 basis points; the rest are per 1% move in the factor. Cells are dimmed where |t| < 2, meaning the relationship is not statistically distinguishable from noise.

Theme10y yield
bp
2s10s slope
bp
US dollar
%
Crude oil
%
Gold
%
Copper
%
Credit appetite
%
VIX
%
S&P 500
%
Top macro
excl. mkt
Ex-beta
residual
AI Compute & Semiconductors-1.46
0.04
-1.04
0.01
-1.28
0.03
-0.20
0.04
0.52
0.10
0.87
0.26
1.17
0.01
-0.23
0.33
2.77
0.54
 Copper
0.260
 Copper
0.014
Power & Grid for AI Datacenters-1.09
0.04
-0.59
0.00
-1.11
0.04
-0.16
0.04
0.37
0.08
0.57
0.17
0.52
0.00
-0.15
0.21
1.69
0.31
 Copper
0.174
 Copper
0.012
Defense Tech & Autonomy-0.68
0.02
-0.37
0.00
-0.79
0.03
-0.02
0.00
0.26
0.06
0.30
0.07
0.21
0.00
-0.07
0.06
1.02
0.16
 Copper
0.067
 Crude oil
0.016
Energy Storage & Critical Minerals-1.07
0.02
-0.52
0.00
-1.71
0.04
-0.19
0.03
0.69
0.14
0.88
0.22
1.80
0.01
-0.24
0.31
2.68
0.42
 Copper
0.220
 Gold
0.040
Robotics & Physical AI-1.02
0.03
-0.45
0.00
-1.19
0.04
-0.26
0.11
0.40
0.09
0.75
0.30
1.82
0.03
-0.20
0.41
2.32
0.59
 Copper
0.299
 Copper
0.024
Space & Satellite Economy-1.58
0.03
0.35
0.00
-2.31
0.07
-0.17
0.02
0.62
0.10
0.75
0.14
1.49
0.01
-0.22
0.20
2.52
0.32
 Copper
0.135
 US dollar
0.019
Fintech & Payment Rails-1.19
0.04
0.02
0.00
-1.11
0.04
-0.16
0.04
0.35
0.07
0.52
0.15
0.95
0.01
-0.17
0.28
1.94
0.42
 Copper
0.147
 2s10s slope
0.006
Biotech & AI Drug Discovery-1.05
0.05
-0.22
0.00
-1.06
0.05
-0.15
0.05
0.31
0.08
0.30
0.07
-0.18
0.00
-0.12
0.20
1.24
0.24
 Gold
0.079
 Gold
0.015
Enterprise Software & AI Agents-0.41
0.01
0.30
0.00
-0.64
0.01
-0.08
0.01
0.13
0.01
0.23
0.03
0.70
0.00
-0.10
0.09
1.28
0.19
 Copper
0.030
 Copper
0.008
Cybersecurity-0.15
0.00
0.23
0.00
-0.15
0.00
0.01
0.00
0.02
0.00
0.17
0.01
1.38
0.01
-0.08
0.04
1.30
0.13
 Copper
0.011
 Crude oil
0.021
How to read this, and the honest caveat. VIX and the S&P are the market factor in two costumes — VIX "wins" any naive horse race simply because equities fall when volatility rises, which is why both are excluded from the "Top macro" column and reported separately below. Copper is the strongest genuine macro driver across most themes, consistent with these being industrial-buildout trades. But note the magnitudes: the best true-macro R² across all ten themes is 0.30, against a market-factor R² reaching 0.59. Strip market beta out and the strongest residual macro relationship falls to 0.040. On this year of data these themes are overwhelmingly beta stories, not macro stories.

Market factor versus macro — variance explained

Side by side: how much of each theme's daily variance the market explains, versus its best macro factor, versus its best macro factor after market beta is removed.

ThemeS&P betaMarket R²Best macro R²Residual macro R²
Robotics & Physical AI 2.32
0.59
0.299
0.024
AI Compute & Semiconductors 2.77
0.54
0.260
0.014
Energy Storage & Critical Minerals 2.68
0.42
0.220
0.040
Fintech & Payment Rails 1.94
0.42
0.147
0.006
Space & Satellite Economy 2.52
0.32
0.135
0.019
Power & Grid for AI Datacenters 1.69
0.31
0.174
0.012
Biotech & AI Drug Discovery 1.24
0.24
0.079
0.015
Enterprise Software & AI Agents 1.28
0.19
0.030
0.008
Defense Tech & Autonomy 1.02
0.16
0.067
0.016
Cybersecurity 1.30
0.13
0.011
0.021

3 · Rolling Correlation

Sixty-day rolling correlation of each theme against copper and against the 10-year yield. Regime shifts appear here as lines crossing zero or converging — a theme whose correlation is unstable is one where the macro relationship is not durable.

vs Copper

vs 10y yield change

4 · Regime Lens

Trailing-year days bucketed by rate direction and dollar direction over the preceding month, showing each theme's annualised average return within each bucket. The current regime is rates flat / USD↑, highlighted below.

Themerates↑ / USD↑
50d
rates↑ / USD flat
39d
rates flat / USD flat
37d
rates flat / USD↓
37d
rates flat / USD↑
35d
AI Compute & Semiconductors+65%+137%+194%+138%-190%
Power & Grid for AI Datacenters+15%-61%+70%+32%-42%
Defense Tech & Autonomy-38%+68%-33%+13%-58%
Energy Storage & Critical Minerals-169%-16%+182%+152%-232%
Robotics & Physical AI+13%+56%-6%+3%-92%
Space & Satellite Economy+23%+178%+207%+141%-381%
Fintech & Payment Rails+25%-93%+39%+18%-71%
Biotech & AI Drug Discovery+70%+68%-31%+42%-90%
Enterprise Software & AI Agents+44%-32%+107%+80%-71%
Cybersecurity+48%-27%+137%+138%-21%
Small-sample warning. This is one year of history sliced into buckets. The current regime bucket contains only 35 days, and annualising a handful of daily observations produces large, unstable numbers. Read the sign and the relative ordering, not the magnitudes. Buckets with fewer than 15 days are omitted entirely.

5 · Sector Reality Check

What GICS sector each theme is actually made of. The right-hand column is concentration — the share of the basket sitting in its single largest sector.

Theme
Basic Materials
Communication Services
Consumer Cyclical
Financial Services
Healthcare
Industrials
Technology
Utilities
Top sector share
AI Compute & Semiconductors7100%
Biotech & AI Drug Discovery8100%
Cybersecurity7100%
Fintech & Payment Rails7188%
Enterprise Software & AI Agents1788%
Defense Tech & Autonomy5271%
Space & Satellite Economy14167%
Power & Grid for AI Datacenters3457%
Robotics & Physical AI14257%
Energy Storage & Critical Minerals2112129%
Three themes are single-sector. AI Compute & Semiconductors, Cybersecurity, and Biotech & AI Drug Discovery are each 100% one GICS sector — they are sector exposures with thematic labels, and will behave like the sector regardless of the narrative. At the other end, Energy Storage & Critical Minerals spans five sectors with only 29% in its largest, which is why it behaves least like a single coherent bet.

Relative strength versus benchmarks

Theme return minus benchmark return. The sector column asks whether the theme beat its GICS sector; the thematic column asks whether it beat the closest thematic ETF. Positive on both is where thematic selection added something beyond simply owning the sector.

ThemeSector1M3M1YThematic1M3M1Y
Enterprise Software & AI Agents XLK +2.6%+15.2%-19.4% IGV +1.6%+7.4%+17.6%
Defense Tech & Autonomy XLI -4.1%+9.8%-28.0% ITA -2.5%+12.8%-18.8%
Fintech & Payment Rails XLF +4.6%+9.1%-14.1% FINX +8.2%+13.2%+19.0%
Cybersecurity XLK -1.0%+8.8%+29.4% HACK -0.6%+6.8%+23.2%
Biotech & AI Drug Discovery XLV -2.2%+5.0%-1.2% ARKG +12.0%-6.0%-22.0%
Robotics & Physical AI XLI -2.7%+4.8%+2.1% BOTZ +2.6%+2.6%+22.4%
Power & Grid for AI Datacenters XLU +3.5%+1.3%+7.4% URA +21.7%+3.6%+37.3%
AI Compute & Semiconductors XLK +0.4%-2.4%+70.7% SMH +1.7%+5.4%+41.9%
Space & Satellite Economy XLI -13.2%-11.1%-3.8% ITA -11.7%-8.1%+5.4%
Energy Storage & Critical Minerals XLB -21.1%-30.2%-53.9% LIT -11.0%-24.1%-50.8%

6 · Market Maps — Value Chain by Theme

All 72 public tickers and 74 private companies placed in the layer where their centre of gravity sits, ordered upstream to downstream. Hover any chip for the placement rationale.

TICKER public TICKER spans 3+ layers Private Co private private only no public pure play no private rival incumbent-held

AI Compute & Semiconductors

7 public · 8 private · dominant GICS Technology (100%)
1Foundry & Manufacturingno private rival
TSM
1 public · 0 private
2Memory & HBMno private rival
MU
1 public · 0 private
3Accelerators & Custom Silicon
AMDARMAVGONVDACerebras SystemsEtchedGroqSambaNova SystemsTenstorrent
4 public · 5 private
4Interconnect & Networkingno private rival
MRVL
1 public · 0 private
6Models & Labsprivate only
AnthropicOpenAIxAI
0 public · 3 private
Private-only at primary layer (OpenAI, Anthropic, xAI all pre-IPO); largest AI capex allocators are entirely inaccessible to public-market investors as pure plays.

Power & Grid for AI Datacenters

7 public · 8 private · dominant GICS Utilities (57%)
0Fuel & Generation Assets
CEGTLNVSTFermi America
3 public · 1 private
1Generation Technology
GEVAntares NuclearBlue Energy GlobalTerraPowerValar Atomics
1 public · 4 private
Four private SMR/next-gen nuclear startups versus only GEV as public incumbent; SMR commercialization wave has no investable pure-play public equity yet.
2Transmission & Grid Equipmentno private rival
ETN
1 public · 0 private
3EPC & Constructionno private rival
PWR
1 public · 0 private
4Datacenter Operatorsprivate only
CoreWeaveCrusoe EnergyVantage Data Centers
0 public · 3 private
Zero public primaries; CoreWeave (recently IPO'd) and Vantage are the only names, making hyperscale datacenter operation effectively a private-only layer for new investment.
5Retail Power & Tradingno private rival
NRG
1 public · 0 private

Defense Tech & Autonomy

7 public · 8 private · dominant GICS Industrials (71%)
0Primes & Platforms
LDOSLMTRTXAnduril Industries
3 public · 1 private
1Autonomy Software & AI
PLTRApplied IntuitionHelsingShield AI
1 public · 3 private
2Sensors, EW & Commsno private rival
LHX
1 public · 0 private
3Munitions & Effectorsprivate only
CastelionEpirus
0 public · 2 private
Zero public primaries despite growing loitering munitions demand; Epirus and Castelion are private-only, leaving the counter-drone and hypersonics wave untracked publicly.
4Uncrewed Systems
AVAVKTOSSaronic TechnologiesSkydio
2 public · 2 private

Energy Storage & Critical Minerals

7 public · 6 private · dominant GICS Technology (29%)
0Mining & Extraction
ALBMPMariana Minerals
2 public · 1 private
1Refining & Materialsprivate only
Sila Nanotechnologie
0 public · 1 private
2Cell Manufacturing
EOSEFSLRTSLACMBlu EnergyForm Energy
3 public · 2 private
3Systems Integration & BESSno private rival
ENPHFLNC
2 public · 0 private
4Grid & Utility Deploymentprivate only
Base Power
0 public · 1 private
Zero public primaries; Base Power is the only private name, while FLNC covers it as secondary — a gap given utility-scale BESS deployment is accelerating rapidly.
5Recycling & Circularityprivate only
Redwood Materials
0 public · 1 private
Zero public primaries with only Redwood Materials as a private player; battery recycling mandates are growing but have no public pure-play investment vehicle.

Robotics & Physical AI

7 public · 8 private · dominant GICS Industrials (57%)
1Industrial Arms & Automationno private rival
FANUYROKSYMTER
4 public · 0 private
2Humanoids & General-Purposeprivate only
1X TechnologiesAgility RoboticsApptronikFigure AINeura RoboticsUnitree Robotics
0 public · 6 private
Crowded with six private challengers (Figure, Apptronik, 1X, Agility, Neura, Unitree) and zero public primaries; largest near-term robotics growth vector is entirely private.
3Mobile & Delivery Roboticsno private rival
SERVZBRA
2 public · 0 private
4Surgical & Medical Roboticsno private rival
ISRG
1 public · 0 private
5Perception & Robot Foundation Modelsprivate only
Physical IntelligencSkild AI
0 public · 2 private
Zero public primaries; Physical Intelligence and Skild AI are private-only, meaning the robot-brain software layer has no public investable proxy.

Space & Satellite Economy

6 public · 8 private · dominant GICS Industrials (67%)
0Launch
BARKLBSpaceXStoke Space
2 public · 2 private
1Satellite Manufacturingprivate only
Apex SpaceK2 Space
0 public · 2 private
Zero public primaries; K2 Space and Apex Space are pure-play private bus manufacturers, while SpaceX manufactures Starlink satellites internally at scale.
2Components & Subsystemsno private rival
HEI
1 public · 0 private
3Constellation Operatorsno private rival
ASTSIRDM
2 public · 0 private
5In-space Services
LUNRAxiom SpaceImpulse SpaceTrue AnomalyVarda Space Industri
1 public · 4 private

Fintech & Payment Rails

8 public · 6 private · dominant GICS Financial Services (88%)
0Card Networks & Acquiring
MAPYPLVXYZStripe
4 public · 1 private
1Issuing & Banking-as-a-Serviceprivate only
Revolut
0 public · 1 private
Zero public primaries; Revolut is the dominant private player while Unit and Marqeta (outside dataset) are the only names, suggesting BaaS is under-represented publicly.
2Stablecoins & Crypto Rails
CRCLBridge (Stripe subsi
1 public · 1 private
3Exchanges & Brokerage
COINHOODFigure AI
2 public · 1 private
4Consumer Credit & BNPL
AFRMKlarna
1 public · 1 private
5Infrastructure & Complianceprivate only
Ramp
0 public · 1 private
Zero public primaries despite the AML/KYC compliance software market growing rapidly; Ramp covers it as primary but is private, and no public pure-play exists.

Biotech & AI Drug Discovery

8 public · 7 private · dominant GICS Healthcare (100%)
0Tools & Sequencingno private rival
ILMN
1 public · 0 private
1Discovery Platforms
RXRXSDGRChai DiscoveryEvolutionaryScaleGenesis TherapeuticsIambic TherapeuticsIsomorphic LabsXaira Therapeuticsinsitro
2 public · 7 private
Crowded with seven private challengers vs. only two public names (RXRX, SDGR); Discovery Platforms is the most contested private biotech layer and may see consolidation.
4Commercial Pharmano private rival
ABBVLLYNVOVRTX
4 public · 0 private
5Diagnostics & Real-World Datano private rival
TEM
1 public · 0 private

Enterprise Software & AI Agents

8 public · 7 private · dominant GICS Technology (88%)
0Cloud Infrastructureno private rival
GOOGLMSFT
2 public · 0 private
1Data Platforms
MDBSNOWDatabricksNeon (Databricks sub
2 public · 2 private
2Observability & DevOpsno private rival
DDOG
1 public · 0 private
3Agent Frameworks & Orchestrationprivate only
Cognition (Devin)
0 public · 1 private
Zero public primaries; Cognition (Devin) is the only private primary, while MSFT/GOOGL and NOW/CRM own it as secondary — the emergent agentic layer has no public pure-play.
4Vertical & Horizontal Applications
AICRMNOWWriter
3 public · 1 private
5Model Providersprivate only
AnthropicCohereOpenAI
0 public · 3 private
Zero public primaries; OpenAI, Anthropic and Cohere are all private, making the most strategically critical AI layer entirely inaccessible as a direct public-market investment.

Cybersecurity

7 public · 8 private · dominant GICS Technology (100%)
0Identity & Access
OKTAInforcer
1 public · 1 private
1Endpoint & XDRno private rival
CRWDS
2 public · 0 private
2Network & Edge/SASEno private rival
FTNTNETPANWZS
4 public · 0 private
Crowded with four public primaries (PANW, FTNT, NET, ZS) and zero private challengers at primary layer; incumbents appear entrenched with no notable new entrant.
3Cloud & Data Securityprivate only
Abnormal AISocketZenity
0 public · 3 private
Zero public primaries; Socket, Abnormal AI and Zenity are private challengers competing in CNAPP/data-security without a dedicated public pure-play.
4SecOps, SIEM & Automationprivate only
Eye SecurityTorqZafran Security
0 public · 3 private
Zero public primaries; Torq, Zafran and Eye Security are private-only, while CRWD, S and PANW cover it only as secondary layers.
5Offensive & Autonomous Securityprivate only
XBOW
0 public · 1 private
Zero public and only one private (XBOW); AI-driven offensive security / red-teaming is nascent with almost no investable exposure.

7 · Competitive Pressure

The public-versus-private relationships in the dataset, counted. These same relationships drive the moat factor in the ratings model — shown here as structure rather than a score.

Public names facing the most private challengers

TickerThemeRivalsNamed challengersMoat
NVDAAI Compute & Semiconductor
8		 OpenAI, Anthropic, xAI, Cerebras Systems, Groq, SambaNova Systems, Tenstor 3.00
VSTPower & Grid for AI Datace
8		 Fermi America, Valar Atomics, Blue Energy Global, Antares Nuclear, Crusoe 2.00
RKLBSpace & Satellite Economy
8		 SpaceX, Stoke Space, K2 Space, True Anomaly, Impulse Space, Apex Space, Va 2.00
ROKRobotics & Physical AI
8		 Figure AI, Physical Intelligence, Skild AI, Apptronik, 1X Technologies, Ag 2.00
CRWDCybersecurity
8		 XBOW, Socket, Torq, Abnormal AI, Zafran Security, Zenity, Eye Security, In 2.00
AMDAI Compute & Semiconductor
7		 OpenAI, Anthropic, xAI, Cerebras Systems, SambaNova Systems, Tenstorrent, 2.00
CEGPower & Grid for AI Datace
7		 Fermi America, Valar Atomics, Blue Energy Global, Antares Nuclear, Crusoe 2.00
GEVPower & Grid for AI Datace
7		 Fermi America, Valar Atomics, Blue Energy Global, Crusoe Energy, CoreWeave 2.00
LMTDefense Tech & Autonomy
7		 Anduril Industries, Shield AI, Saronic Technologies, Helsing, Epirus, Cast 3.00
RTXDefense Tech & Autonomy
7		 Anduril Industries, Shield AI, Saronic Technologies, Helsing, Epirus, Cast 3.00
LHXDefense Tech & Autonomy
7		 Anduril Industries, Shield AI, Saronic Technologies, Helsing, Epirus, Cast 2.00
SYMRobotics & Physical AI
7		 Figure AI, Physical Intelligence, Skild AI, Apptronik, 1X Technologies, Ag 2.00
PANWCybersecurity
7		 XBOW, Socket, Torq, Abnormal AI, Zafran Security, Zenity, Eye Security 2.00
RXRXBiotech & AI Drug Discover
7		 Isomorphic Labs, Chai Discovery, Xaira Therapeutics, EvolutionaryScale, in 2.00

Private companies pressuring the most public names

CompanyValuationTargetsTickers
Anduril Industries
Defense Tech & Autonomy
 $61B 7 LMTRTXLDOSKTOSAVAVLHXPLTR IPO watch
OpenAI
Enterprise Software & AI Agent
 $730B 6 MSFTGOOGLCRMNOWSNOWAI IPO watch
Anthropic
Enterprise Software & AI Agent
 $965B 5 MSFTGOOGLCRMNOWAI IPO watch
Shield AI
Defense Tech & Autonomy
 $13B 5 LMTRTXAVAVKTOSLHX IPO watch
SpaceX
Space & Satellite Economy
 $800B 5 RKLBASTSLUNRIRDMBA IPO watch
Databricks
Enterprise Software & AI Agent
 $190B 5 SNOWDDOGMDBMSFTGOOGL IPO watch
Cohere
Enterprise Software & AI Agent
 $7B 5 MSFTGOOGLCRMAISNOW IPO watch
Fermi America
Power & Grid for AI Datacenter
 — 4 VSTNRGCEGGEV
Valar Atomics
Power & Grid for AI Datacenter
 $6B 4 VSTCEGGEVNRG
Blue Energy Global
Power & Grid for AI Datacenter
 — 4 VSTCEGGEVNRG
Antares Nuclear
Power & Grid for AI Datacenter
 — 4 VSTCEGNRGETN
Crusoe Energy
Power & Grid for AI Datacenter
 $10B 4 VSTCEGGEVPWR IPO watch
CoreWeave
Power & Grid for AI Datacenter
 — 4 VSTCEGGEVPWR
Vantage Data Centers
Power & Grid for AI Datacenter
 — 4 VSTGEVPWRETN

8 · Breadth & Dispersion

Share of each theme's constituents trading above their 50- and 200-day averages, with return dispersion. Low dispersion means the theme moves as one bet; high dispersion means constituent selection matters more than the theme call.

Theme>50d>200d1M3M1YDispersionAvg betaBest 1YWorst 1Y
Fintech & Payment Rails
50%
75%		 +4.0%+8.2%-9.8% 23 1.90 VCOIN
Enterprise Software & AI Agents
50%
75%		 +0.2%+19.6%+9.7% 30 1.25 DDOGAI
Biotech & AI Drug Discovery
38%
75%		 +2.5%+16.0%+22.7% 65 1.20 ILMNRXRX
Power & Grid for AI Datacenters
29%
29%		 +4.7%-4.8%-1.7% 34 1.66 GEVNRG
Defense Tech & Autonomy
29%
14%		 -4.8%+4.0%-18.1% 32 1.01 RTXAVAV
Robotics & Physical AI
29%
29%		 -3.4%-1.0%+12.0% 62 1.96 TERSERV
Cybersecurity
29%
100%		 -3.4%+13.2%+58.5% 47 1.30 OKTAZS
AI Compute & Semiconductors
0%
86%		 -2.0%+2.0%+99.8% 110 2.74 MUAVGO
Energy Storage & Critical Minerals
0%
0%		 -20.7%-28.4%-39.1% 30 2.65 ALBEOSE
Space & Satellite Economy
0%
17%		 -13.9%-16.9%+6.1% 60 2.55 IRDMASTS

Method & Limitations

Rates come from the US Treasury daily par yield curve CSV, not from Yahoo's ^TNX, which returned only 16 usable daily closes over a year. Everything else — volatility, credit, sector and thematic ETFs — is CBOE's delayed daily history. The Nasdaq 100, dollar, crude, gold, copper and bitcoin are shown through the ETFs that track them (QQQ, UUP, USO, GLD, CPER, IBIT), so levels are ETF prices and changes track the underlying net of fees. Credit appetite is the HYG/LQD price ratio, a proxy for the high-yield spread rather than the spread itself.

Factor betas are univariate ordinary least squares on daily returns over the trailing year. Univariate rather than multivariate on purpose: with nine correlated factors and roughly 250 observations, multivariate coefficients are unstable and hard to interpret. The trade-off is that univariate betas do not separate overlapping factors — a copper beta partly reflects the market beta riding underneath it, which is exactly why the residual column exists.

Residual exposure regresses theme returns on the S&P first, then regresses the residuals on each macro factor. That isolates macro sensitivity that is not simply high beta.

These are descriptive, not predictive. A beta measured over one year describes what happened; it does not forecast. Correlations over a single year are unstable, which is what the rolling chart is for. R² is shown on every cell so weak relationships read as weak instead of being quietly presented as signal, and cells where |t| < 2 are dimmed.

Regime analysis is a small sample. One year sliced into direction buckets leaves tens of days per bucket. Signs and ordering are informative; magnitudes are not.

GICS classification is Yahoo's, which occasionally differs from S&P's official assignment. Value-chain layers are an editorial taxonomy — a company's centre of gravity is a judgment call, and the spans 3+ layers flag marks vertical integrators whose single-layer placement understates their reach.

Hard macro is absent. CPI, payrolls, and ISM are not reachable from this environment. Everything here is market-priced and daily, which suits correlating against daily equity returns but omits the official economic series entirely.

Nothing here is investment advice. This page states relationships and leaves the conclusions to the reader by design.