Capital Flows
Global Macro Livestream
Thursday, July 02, 2026
Capital Flows Daily Livestream
Streaming M-F 11:30 MST (1.5 hours before cash equity close)
Built for
Active risk takers who put real capital to work and want their book aligned with the biggest drivers in global markets.
Traders and allocators hunting asymmetric, home-run trades instead of nickel-and-diming basis points.
People who want their actions in sync with the macro regime so months of disciplined decisions don't get erased in a single week.
Especially if you...
Need to navigate recession risk, inflation shocks, macro liquidity, and credit cycle turns without guessing the narrative of the day.
Want a clear read on rates, FX, and US equities so you know when to press risk and when to get out of the way.
Make capital allocation decisions for yourself or others as a PM, CIO, prop trader, wealth manager, family office, or serious individual trader.
If you're responsible for risk, not just opinions, this is built for you.
Macro Livestream Format
Map the macro regime so you're on the right side of it, and isolate the few large, asymmetric bets that can become home-run trades.
Free Macro Playbook
Macro Flows From First Principles
What Actually Drives Markets
How The Structural Regime Is Developing
Filtering False Narratives And Institutional Bias
Reading The Daily Data In Regime Context
Global Capital Flows
Macro Regime
Positioning & Systematic Flows
Macro Liquidity
Credit Cycle
Cross-Border Flows
Growth · Inflation · Liquidity
Rates & FX
Equities
Market Microstructure
Option Flow
Momentum & Mean Reversion
A Complete Read On The Macro Backdrop
Clarity on the Macro Regime
AND
Conviction in Asymmetric Bets
Session Output
During the Livestream
Live Q&A
After the Livestream(20-Minute Read)
Proprietary Macro Flows & Positioning Report
Recording of the Private Session
Synthesized Transcript of the Private Session
THE SESSION AHEAD
What you'll walk away with
By the end of this session, you will have a working map of the US-China AI arms race — every moving part, how they wire into one machine, and the dated watchlist that tells you if that machine starts running in reverse.
We take the machine apart: five moving parts, each examined through three lenses — technology, geopolitics, markets — each starting from one intuitive sentence before we add any depth.
Then we rebuild it: the parts wire into one reflexive loop, we run the clock on the 30 days that just assembled it — and we finish with the six signals to watch from here.
Timing is not an accident: the strongest US model came back online yesterday after an 18-day government pulldown, and the executive order that started it hits its first deadline today.
TODAY'S QUESTION
Could China crash the entire AI trade?
The fault line is not who wins AI — it is the PRICE of intelligence falling faster than the most crowded trade in history can adjust.
The claim making the rounds: “Kill the US labs with free great models. Then use cheap electricity to export compute. The chip is the blocker — but Huawei catches up soon.”viral post · July 2026
Ranked by threat credibility: free models — strongest (it only needs to compress token prices, not kill anyone) · compute export — partial (Global South first) · chip catch-up — weakest near-term (memory is the real gate).
What makes it a market question: the trade pricing in the old assumptions is roughly half the S&P 500 — and since 2025 it runs on borrowed money. Margin repricing, not capability parity, is the crash mechanism.
01 · THE PRICE WAR 02 · THE OPEN-WEIGHTS GATE 03 · THE SILICON CHECKPOINT 04 · THE ENERGY FLOOR 05 · THE CREDIT-FUNDED TRADE falling cost of intelligence ungatable distribution chip / memory choke power buildout debt-financed concentration
US-CHINA AI ARMS RACE -- THE WHOLE PICTURE
The machine we’re about to take apart
THE MACHINE
Lenses
TECH
GEOPOLITICS
MARKETS
Machine State
MOVING PARTS5
LENSES3
ACTIVE LENS
PACKET AT
Concentration
50%
Share of S&P 500
in 28 AI stocks
Five moving parts, one machine -- break one part and the whole thing can run in reverse.
ILLUSTRATIVE MECHANICS -- NOT A FORECAST
Framework: Capital Flows -- July 2, 2026
THE PRICE WAR
What does intelligence cost now?
The quality gap is one model generation. The price gap is ten to a hundred fold.
Tech lens: the best Chinese open-weight model scores 51 on the composite intelligence index vs 60 for the US frontier — roughly where the US frontier was one generation ago. The gap keeps closing faster than forecast.
Tech lens: output tokens: US frontier $25-50 per million; comparable-tier Chinese open-weights $0.28-4.40. Same workloads, a different order of magnitude.
Geo lens: this is a business-model asymmetry, not charity — the leading Chinese labs are funded by hedge funds, patient capital and the state. They do not need API margins. US labs need those margins to fund the next model.
Market lens: enterprise buyers noticed. One major crypto exchange cut its AI bill ~50% by defaulting to Chinese models; an agent startup moved 100% of traffic; a US CEO now calls US token pricing “a wealth tax” on live TV.public statements · June 27 - July 1, 2026
The tension: if the premium band is only nine index points wide, how much revenue can it defend — and what happens to the spend built on defending it?
US FRONTIER 60 CHINA OPEN-WEIGHT 51 MARGIN ZONE -- FUNDS FRONTIER R&D THE PRICE FLOOD TOKEN PRICES COLLAPSING
$0.28 $1 $2 $4.40 $10 $25 $50 $ / M TOKENS OUT LOG SCALE PRICE FLOOR
MOVING PART 1 -- TECHNOLOGY
The falling price of intelligence
THE PRICE WAR
Token Price -- $ / M Out
US frontier$25-50
China open$0.28-4.40
China floor
Intelligence Index
US frontier60
China open-weight51
Quality gap~9 PTS
China price vs US
China now sells ~85% of the intelligence at 1-10% of the price.
ILLUSTRATIVE MECHANICS -- NOT A FORECAST
Model index & pricing: public leaderboards and list prices, July 2026
THE PRICE WAR
Nine points of quality, a hundred-fold price gap
Read it: the top Chinese open-weight model sits nine index points below the US frontier at roughly one-tenth the price. The premium is real -- but it is a band, not a moat.
THE PRICE WAR
The buyers already voted
Read it: eighteen months ago Chinese models were ~1% of routed tokens; by early 2026 they crossed half. Not a shrinking pie -- everyone's volumes grew, China's ~4x faster. Downloads flipped the same way, independently. (Routing platforms skew price-sensitive; US labs' first-party premium traffic not shown.)
THE OPEN-WEIGHTS GATE
Why the gate closed on an empty river
You can gate a closed model. You cannot gate weights that are already published everywhere.
Geo lens: June 12 — Washington ordered the strongest US model suspended worldwide over a jailbreak report. First time a deployed frontier model has ever been pulled. Anthropic complied within hours.
Geo lens: June 13 — within ~24 hours, Beijing's champion lab open-sourced a near-frontier rival under MIT license, no regional restrictions: “frontier intelligence should not belong to only a few, nor should it be withdrawn by a few rules at any time.”
Tech lens: the asymmetry is structural. A closed model lives on its maker's servers — gateable. Open weights replicate onto any datacenter on earth the day they ship. Each US gating action raises the value of the ungatable alternative.
Market lens: 18 days later Washington reversed. But every CTO learned the lesson: hold a Chinese open-weight fallback. On one major gateway, the leading Chinese model went <1% to 17% of tokens in a single month.platform data · May-June 2026
The tension: the ban was meant to protect the moat — the market's verdict on who it actually helped prints on the next chart.
RESERVOIR -- CLOSED MODEL MIT LICENSE ENTERPRISE DELTA
JUN 12 -- SHUTDOWN JUN 13 -- OPEN-SOURCE RELEASE JUN 30 -- REVERSAL SELF-REPLICATION FORK
MOVING PART 2 -- GEOPOLITICS
The gate closed on an empty river
THE OPEN-WEIGHTS GATE
DOWNSTREAM THROUGHPUT -- OPEN-WEIGHT PACKETS
You cannot export-control something that is already everywhere.
Export Control
Days gated
Gate
Downstream River
Rival release+1 DAY
Enterprise fallbacks
ILLUSTRATIVE MECHANICS -- NOT A FORECAST
Timeline: US Commerce actions & model releases, June-July 2026
THE OPEN-WEIGHTS GATE
Who is the open-weight champion?
The state-backed lab that gives its models away is now a HK$1 trillion company.
Who they are: Zhipu AI — a Tsinghua University spinoff (2019, founded by Prof. Tang Jie's knowledge-engineering lab), maker of the GLM model family. One of China's “six AI tigers” — and the first pure-play LLM maker ever to IPO (Hong Kong, January 2026).
The state wiring: government funds from Beijing, Shenzhen, Hangzhou, Shanghai and more sit on the cap table (over RMB 10bn raised); on the US Entity List since Jan 2025 over alleged military-AI ties. No formal “national champion” title — the shareholder register does the talking.
The business model: MIT-licensed weights are the funnel, not the product — ~85% of revenue is on-premise deployments for state firms, banks and governments. Latest year: ~$105M revenue (+132%) against a ~$650M net loss. It burns ~$10 per $1 of revenue — and the state keeps writing checks. Loss-leader pricing IS the strategy.
Fit in the machine: its open-source answer landed within days of the US pulldown; it exports sovereign AI (Malaysia, Singapore, UAE, Saudi Arabia, Kenya — Huawei-chip “AI-in-a-Box”), targeting ~20 Belt-and-Road states. Founder to Musk, on a Fable-class Chinese model: “It won't take that long.”
Why it matters: HK$1 trillion of market cap on ~$105M of revenue means the market is pricing a strategic weapon, not a company. The chart on the next slide is what that repricing looked like — and July 8, when ~69% of IPO shares unlock, is its first real stress test.
THE OPEN-WEIGHTS GATE
The market priced the backfire in one week
Read it: Zhipu AI's Hong Kong-listed stock -- +1,700% since the January IPO, past HK$1 trillion in market value in June. The 18-day US pulldown was the best marketing the open-weight champion ever got. Watch July 8: the first lock-up expiry.
THE SILICON CHECKPOINT
The counterargument: silicon
The chip choke is real — and the true gate is memory, not logic.
Tech lens: domestic Chinese fabrication is stuck at 7nm; even aggressive scenarios put Huawei's aggregate AI compute at ~4% of Nvidia's. The genuinely binding constraint is high-bandwidth memory — the stacked RAM every AI chip needs.
Tech lens: the disputed variable: China's memory champion claims mass HBM production this year (a $4.2bn IPO funds it); independent analysts say the timeline is slipping. This single number decides how long the choke holds.
Geo lens: the twist — Beijing is now blocking its own firms from buying US chips, deliberately starving them to force the domestic flywheel. A $295bn national computing grid is being drafted around 80% domestic silicon.
Market lens: China's answer meanwhile is brute force: its flagship cluster beats the US rack on raw compute by spending 3x the money and 4x the power — viable precisely because power is the one thing China has in surplus.
The tension: sanctions made Chinese compute expensive, not impossible — and expensive is a problem money and megawatts can attack.
THE FRONTIER
LITHOGRAPHY (7nm) GATE 1 -- WIDE ADVANCED PACKAGING GATE 2 -- NARROWER MEMORY (HBM) GATE 3 -- NARROWEST, RUNS HOTTEST DOMESTIC BYPASS ? CAPACITY DISPUTED
Throughput
Agg compute (index)
100
US
~4
China
Beijing policy -- state
Checkpoint reads
Domestic node7nm
Huawei vs Nvidia compute~4%
Memory gate
The bottleneck is not the chip -- it is the memory stacked on top of it.
MOVING PART 3 -- TECHNOLOGY x GEOPOLITICS
The bottleneck behind the bottleneck
THE SILICON CHECKPOINT
ILLUSTRATIVE MECHANICS -- NOT A FORECAST
Compute & node estimates: public research, 2026 -- bypass width disputed
THE SILICON CHECKPOINT
Brute force vs efficiency
Read it: China's flagship cluster out-computes the US rack by brute force -- at 3x the price and 4x the power. Efficiency is still America's edge; abundance is China's.
THE SILICON CHECKPOINT
Who wins if the flywheel is forced?
Beijing's $295bn computing grid must be built from a handful of listed names — and the market already found them.
The accelerator — Cambricon (688256): China's purest Nvidia substitute — US-blacklisted since 2022, so a 100% domestic story. Revenue +4,300% in one half-year; targeting 500k chips in 2026 vs 116k last year; ByteDance reportedly pre-ordered ~200k. Briefly overtook Moutai as China's most valuable stock.
The chokepoint — SMIC (0981 HK): every domestic accelerator routes through its fabs (it manufactures Huawei's and Cambricon's silicon). 7nm capacity reportedly doubling this year, 5nm pilot runs for Huawei and Alibaba, four new fabs under construction — one dedicated to Huawei AI chips.
The shovels — Naura (002371): China's #1 chip-equipment maker — every domestic fab capex dollar passes through its etch and deposition tools. Revenue +31% in 2025 with ~30% growth forecast through 2026. Same logic: AMEC (688012).
The rest of the stack: Hygon (688041) — domestic server CPUs + AI co-processors, revenue +57%; Moore Threads (688795) — the GPU startup that closed +425% on its December debut, the year's biggest Shanghai tech IPO.
The one to watch — CXMT (IPO pending): China's memory champion — the disputed memory gate from two slides ago IS this company. Filing for 2026's largest mainland IPO, a $3bn Tencent supply deal signed this week, memory-stacking fab targeted for year-end.
The honest counterweight: Chinese chip executives themselves admit a 5-10 year silicon lag; advanced memory still binds; Cambricon's top-5 customers are 94% of its revenue and it trades at thousands of times earnings. These are policy trades at policy multiples — the next chart shows the market pays them anyway.
THE SILICON CHECKPOINT
The market already found the winners
Read it: the domestic-substitution complex re-rated hard in a year -- the accelerator +263%, the equipment maker +156%, the CPU maker +136%, the foundry +85%. A forced flywheel met a willing market. (Shanghai / Shenzhen / Hong Kong listings.)
THE ENERGY FLOOR
More muscle vs better brains
China overbuilt power. America overbuilt demand.
Geo lens: in one year China added ~429 GW of generation capacity; the US added ~51 GW. In 2025 the gap widened to roughly 8x. China's grid carries reserve margins of 80-100% vs ~15% in typical US regions.
Market lens: the US side binds: bank research projects a 44 GW datacenter power shortfall through 2028 — up to a fifth of planned AI demand with nowhere to plug in. Time-to-power is the US buildout's scarcest input.
Tech lens (the honest counterweight): US datacenters are far more cost-efficient per unit of compute, Chinese industrial power is not meaningfully cheaper, and Beijing is already cracking down on datacenter overcapacity.
Geo lens: so the “cheap electricity” claim is really about buildout speed and slack — China treats AI datacenters as a sponge for surplus power and exports the full stack (chips + models + sovereign cloud) Belt-and-Road style.
The tension: intelligence is becoming an energy business — and the two contenders built opposite constraints into their grids.
Moving Part 4 — Geopolitics
More muscle vs better brains
The Energy Floor
Efficiency Edge: US · cost per compute
Capacity added / yr
429 ≈8× 51
Reserve margin 80–100%
Reserve margin ~15%
China
United States
China · new capacity
GW /YR
US · new capacity
GW /YR
Shortfall by 2028
China overbuilt power. America overbuilt demand.
Illustrative mechanics — not a forecast
Capacity additions & shortfall: public grid data and bank research, 2024–2026
THE ENERGY FLOOR
An eight-fold buildout gap
Read it: China added eight times more power in a year than the US, while US datacenters stare at a 44 GW shortfall. China's constraint is chips; America's is megawatts.
THE ENERGY FLOOR
Who gets paid for the power edge?
In America, scarcity made utilities an AI trade. In China, abundance means the wire gets paid — not the power plant.
The verified winner — grid equipment: State Grid spent a record RMB 650bn in 2025 and plans RMB 4 trillion (~$574bn) for 2026-30, +40%. The day that plan dropped (Jan 16), grid-equipment names went limit-up +10%. Sieyuan Electric (002028); NARI Tech (600406) — State Grid's own tech arm, >60% of ultra-high-voltage control systems.
The pipeline — nuclear: 10 new reactors approved for the fourth straight year (>RMB 200bn); 30 under construction ≈ half the world's total; the world's first commercial small modular reactor due this half. CGN Power (1816 HK) and China Nuclear (601985) won the latest batch.
The trap — the generators: hydro -11%, wind -30% over the past year. The 80%-renewable mandate for computing hubs makes them suppliers — but 80-100% reserve margins mean no pricing power. The abundance that gives China its edge is exactly why its power plants don't re-rate.
The policy wiring: a four-ministry AI-energy action plan (May 8) and a “power dictates computing” siting doctrine — put the datacenter where the electrons are. Datacenter demand growing ~19%/yr toward ~289 TWh by 2030, more than double today.
The inversion to remember: the US prices power scarcity — its utilities re-rated as AI trades. China prices power abundance — cheap electrons subsidize the compute layer, and the investable flow is the grid capex wiring western power to eastern data. Sensitivity ranking: grid equipment > nuclear > generators.
THE ENERGY FLOOR
The wire gets paid, not the plant
Read it: grid equipment (+113%) is China's real AI-power trade; the generators sit flat to down (wind -30%) because abundance kills pricing power. The exact inverse of the US, where scarcity re-rated the utilities.
THE CREDIT-FUNDED TRADE
Who is actually holding this risk?
Half the index, funded on borrowed money, priced for a premium that is nine points wide.
Market lens — concentration: 28 AI stocks = ~50% of S&P 500 market cap; the biggest seven alone are ~a third. US equities are ~72% of developed-market indices — an AI repricing is a global wealth event.large-bank market research · 2026
Market lens — financing: hyperscaler capex consumed ~40% of operating cash flow in 2023. Projection for 2026-27: ~94%. The marginal AI dollar is now borrowed, not earned.
Market lens — credit: AI-linked debt is now ~$1.2 trillion — 14% of the US investment-grade index, bigger than banks. $152bn more issued in the first 4.5 months of 2026 than all of 2025. Single-A tech already trades at BBB-profile spreads.
Geo lens: even the BIS — the central bank of central banks — flagged the AI boom as a financial-stability issue in its annual report four days ago. This stopped being just an equity story.
The tension: an equity repricing now tightens the credit channel that funds the capex that justifies the equity. That circle is the machine's most dangerous wire.
MOVING PART 5 -- MARKETS
The engine that borrows its fuel
THE CREDIT-FUNDED TRADE
Equity Flywheelthe AI trade
Credit Pump
Cash Flow
Borrowed
Spread Valve
Capex Pistonsdatacenters · chips · power
← Reflexive Return Belt
6%
94%
The Financing Flip · 2023 → 2026-27
Capex / cash flow
AI share of US IG index14% > BANKS
Issued in 4.5 months$152BN
Credit Spreads · Aperture
Marginal Dollar · Mix
% of capex funded by credit
The trade now borrows the fuel it burns to justify itself.
ILLUSTRATIVE MECHANICS -- NOT A FORECAST
Financing mix & index shares: bank and asset-manager research, 2025-2026
THE CREDIT-FUNDED TRADE
The shock absorber got thinner
Read it: the AI buildout stopped being self-funded -- at 94% of operating cash flow and the largest slice of the investment-grade bond market, the marginal AI dollar is borrowed.
THE WHOLE MACHINE
January 27, 2025 was the dress rehearsal
The market does not reprice on Chinese revenue. It reprices on beliefs about the cost curve.
What happened: one Chinese model release changed beliefs about how cheaply frontier AI could be trained — and erased $589bn from a single stock in one day (-17%), the largest one-day value loss in market history at the time.
Why it recovered: spending accelerated — cheaper intelligence meant more demand for compute, not less. Nine months later the same stock closed above $5 trillion. Demand absorbed the shock.
Why this time is structurally different: that absorption ran on cash-funded capex (~40% of cash flow). The 2026 loop runs at ~94%, debt-funded — the shock absorber is thinner and wired to credit spreads.
The loop, named: belief about cost → AI equity (half the index) → credit spreads ($1.2T) → capex plans → revenue assumptions → back to belief. It ran upward for three years. Nothing in its wiring prevents reverse.
The question that decides everything: does the demand offset still work when the fuel is borrowed? That is what the next visual runs, both directions.
BELIEF ABOUT COST INTACT SHOCKED JAN 27 2025 -$589BN IN ONE DAY AI EQUITY (HALF THE INDEX) BID REPRICING CREDIT SPREADS ($1.2T) OPEN WIDENING THEN: CASH-FUNDED 40% NOW: DEBT-FUNDED 94% CAPEX PLANS EXPANDING CUT REVENUE ASSUMPTIONS RISING SLASHED
FLOW DIRECTION
REBUILD -- ALL FIVE PARTS WIRED
The same loop runs in both directions
THE WHOLE MACHINE
The loop that carried the trade up is the same loop that runs it down.
EXPANSION
SHOCK
UNWIND
RESET
ILLUSTRATIVE MECHANICS -- NOT A FORECAST
Loop mechanics: Capital Flows framework -- precedent: January 27, 2025
THE WHOLE MACHINE
The proof the loop exists
Read it: this is Nvidia's share price. January 27, 2025 -- one belief change erased $589bn of its market value in a day (-17%). It recovered to a $5 trillion peak because cash-funded demand absorbed the shock. Today it trades below peak with China revenue assumed at zero.
THE CLOCK
Thirty days that built the machine
The same story, told in time -- each release wired another part in
Jun 2 — Executive order signed: frontier models must clear a government benchmark-and-review process. The architecture everything else flows through.
Jun 9 — The strongest US model yet ships. Three days of normal.
Jun 12 — Washington orders it suspended worldwide over a jailbreak report — the first-ever pulldown of a deployed frontier model.
Jun 13 — Within ~24 hours, Beijing's champion open-sources a near-frontier rival under MIT license: “frontier intelligence should not belong to only a few.”
Jun 15-22 — The market votes: the rival's maker jumps 48% in a day and tops HK$1 trillion — up 1,700% since its January IPO.
Jun 26 — Partial retreat: restricted trusted-access programs; the next US flagship ships to ~20 government-approved customers only.
Jun 30 — Full reversal: the license requirement is withdrawn — in exchange for standing government oversight. Same day, a US lab cuts prices.
Jul 1 — The model returns worldwide, capped at 50% of normal usage through July 7.
Jul 2 — TODAY: the executive order's first 30-day deliverable comes due. The clock is still running. One month in: the pulldown precedent, the open-weights asymmetry, and the release calendar as a geopolitical variable — every CTO on earth now keeps a Chinese fallback.
THE RISK MACHINE · FIVE COMPONENTS · ONE LOOP PRICE WAR OPEN WEIGHTS CHIPS ENERGY CREDIT JUN 2 AI EXECUTIVE ORDER SIGNED JUN 9 US FRONTIER MODEL LAUNCHES JUN 12 WORLDWIDE PULLDOWN ORDERED JUN 13 CHINA OPEN-SOURCES RIVAL (+1 DAY) JUN 22 RIVAL'S MAKER TOPS HK$1T JUN 26 GOV-APPROVED ACCESS ONLY JUN 30 WASHINGTON REVERSES JUL 1 MODEL RETURNS, CAPPED JUL 2 FIRST ORDER DEADLINE (TODAY)
JUNE 2 -- JULY 2, 2026
Thirty days that built the machine
THE CLOCK
Machine Completion
The Window
DAYS30
GATED DAYS18
RIVAL RESPONSE+1 DAY
Carriage Position
TIMELINE SCRUB · JUN 2 → JUL 2
Each headline didn't just make news -- it assembled the machine.
ILLUSTRATIVE MECHANICS -- NOT A FORECAST
Timeline: public policy actions & model releases, June-July 2026
THE CLOCK
The tape traded every beat of it
Read it: through June's gating saga the most crowded stock in the world lost ~14% while the index barely moved -- the market is already trading the machine we just rebuilt.
THE WATCHLIST
Six dials for what happens next
If the machine starts running backward, one of these six dials moves first.
Jul 7 — usage caps lift on the restored US frontier model. Does premium demand snap back — or did 18 days teach it to live elsewhere?
Jul 8 — the open-weight champion's first lock-up expiry (~25.7M shares) plus a planned Shanghai listing. Does the crown jewel of the China trade absorb its own supply?
~Aug 1 — the classified benchmarking process comes due. Does every future US release become a negotiated deployment?
Ongoing — China's domestic memory-chip volumes: the single most disputed number in the thesis. If domestic memory scales, the silicon checkpoint loosens.
Ongoing — AI credit spreads: single-A tech already prices like BBB. Widening here tightens the fuel line on the whole machine.
The confidence tell — if Beijing ever unblocks US chip purchases, it is admitting the domestic flywheel is weaker than advertised. Bullish the choke, bearish the sprint.
Backdrop: today's US payrolls printed 57k vs 113k expected — a softening tape has less capacity to absorb an AI belief shock. Keep the two stories on one screen.
Usage Caps Lift
Jul 7
01
Days
:
Hrs
:
Min
Rival Lock-Up Expiry
Jul 8 · 25.7M Shares
02
25.7M
Shares
Jul 8
Benchmark Process Due
~Aug 1
03
~AUG 1
China Memory-Chip Volumes
Disputed
04
LOW HIGH
?
AI Credit Spreads
Single-A Priced Like BBB
05
SINGLE-ABBB
Beijing Unblocks US Chips?
The Confidence Tell
06
Blocked
Open
The Watchlist
What Tells You The Machine Is Reversing
Six dials for what happens next
SIGNALS 6 · NEXT: JUL 7
Six dials tell you if the machine starts running backward.
ILLUSTRATIVE MECHANICS -- NOT A FORECAST
Dated catalysts: public filings & policy deadlines, July-August 2026
Macro Livestream Format
Map the macro regime so you're on the right side of it, and isolate the few large, asymmetric bets that can become home-run trades.
Free Macro Playbook
Macro Flows From First Principles
What Actually Drives Markets
How The Structural Regime Is Developing
Filtering False Narratives And Institutional Bias
Reading The Daily Data In Regime Context
Global Capital Flows
Macro Regime
Positioning & Systematic Flows
Macro Liquidity
Credit Cycle
Cross-Border Flows
Growth · Inflation · Liquidity
Rates & FX
Equities
Market Microstructure
Option Flow
Momentum & Mean Reversion
A Complete Read On The Macro Backdrop
Clarity on the Macro Regime
AND
Conviction in Asymmetric Bets
Session Output
During the Livestream
Live Q&A
After the Livestream(20-Minute Read)
Proprietary Macro Flows & Positioning Report
Recording of the Private Session
Synthesized Transcript of the Private Session