• ...do everything you can to manage and protect yourself...
• ...always keep an eye on when something “stops working”...
• ...the market has already done the Fed’s tightening for it...
• ...AIThe shock should be divided into “how value is distributed” and “whether the overall pie is shrinking”...
Column subscription link:https://www.meridianresearch.app/home?product=cross-market-strategies
Notice: By reading, subscribing to, browsing, forwarding, saving, or using the content of this Planet in any way, you are deemed to have fully read, understood, and agreed to this disclaimer. If you disagree, please stop using it and leave immediately. All content on this Planet is for personal research, information compilation, and learning and discussion only. It does not constitute any investment advice, buy or sell recommendation, invitation to make an offer, promise of returns, asset management service, or investment advisory service, nor does reading, subscribing, interacting, or using it create any legal relationship of client, agency, trust, or advisory services. The content does not take into account any recipient’s financial circumstances, investment objectives, risk tolerance, or suitability requirements. Recipients should verify the information themselves, make their own independent assessments and decisions, and bear the investment risks and consequences themselves. Information in the text may come from public sources, third-party data, market information, personal judgment, or model analysis, and may be delayed, incomplete, erroneous, biased, or no longer valid; the author does not guarantee its authenticity, accuracy, completeness, timeliness, or suitability. Past performance, historical examples, backtest results, or the accuracy of past views do not indicate future performance and do not constitute a guarantee of returns. Without permission, you may not copy, take screenshots, make screen recordings, forward, adapt, compile, or use the content for stock recommendations, fundraising, managing money on behalf of others, commercial promotion, or other such purposes. Any consequences arising from such actions shall be borne by the persons responsible!
_____________________________________________________
Zero: The key is whether you can keep yourself in the game
SOXXAfter the breakdown, the technology tail-risk short position opened using SOXS was at one point up approximately 35% during the week, and was ultimately closed for a small profit above breakeven. Some friends gradually took profits around SO overbought levels, which is perfectly fine and shows that you have thoroughly learned the introductory handbook column. I’m very happy for you.
To be precise, this was a hedge against extreme risk. The purpose of an offensive trade is to use the best trading instrument to express a directional view, while the purpose of a hedge is to add the words “risk-controlled” to the distribution’s right tail (a black swan), so that you can protect yourself when an extreme event occurs.
The story of Situational Awareness and Citadel has spread widely. The exact circumstances are impossible to know, but the substantial drawdown resulting from its highly leveraged, concentrated exposure is an absolute fact.
Leopold is someone who has still managed to survive in the arena, and it is also an absolute fact that his YTD remains impressive even after the drawdown.Another absolute fact is that individuals who made many of the same decisions as he did lost their chance to keep surviving in the market. Behind the brutal figures for forced liquidations and blown-up accounts are individual families. When your own money is gone, it is gone.
It is a good thing if a black swan does not occur. It is foreseeable that without this right-tail “rescue” action, once SOXS reached T1, more people would inevitably have been visited by a brutal reality.
When certain things happen, it is difficult for individuals to survive them. Markets move in cycles, and once a cycle passes, everything becomes calm and quiet again—but you only have one life and one pot of money.As individuals in the market, without pricing power, the only thing we can do is everything in our power to manage and protect ourselves under market conditions at every moment.
Unrealized gains are unrealized gains; don’t become attached to winning or losing. Correct actions lead to correct results. There is a long road ahead; it’s not just about this one trade.
...
Stay in the game.
_____________________________________________________
Zero point five: A response to a good question
Some questions are good, and some are bad. The distinction has nothing to do with whether someone is a beginner or mature. Many people who have just started out can also ask questions worth answering. The difference is: 1) whether you have genuinely thought it through yourself before asking, 2) whether you understand that asking me is the best way to get an answer, and 3) whether the perspective from which you ask the question is the right one...
Image: Response to a question about the Hong Kong Dragon FMRB panel (https://www.meridianresearch.app/home?product=asia-strategies）
Always focus on when it “stops working,” because that is the answer to “where you will die.”
There is no Holy Grail.
_____________________________________________________
I. Macro Trading Positioning
Figure: Macro Drivers Positioning Map
Macro-theme logic, trading and unusual market moves (as of 260802)
On 7 29, FOMC voted 9 to 3 to keep the federal funds target range at 3.50%–3.75%, while three dissenters favored a 25-basis-point hike; afterward, BofA projected three more hikes this year, while Barclays and Goldman Sachs leaned toward leaving rates unchanged for the year. The same set of data has yielded policy paths that differ by 75 basis points, and the divergence has gone beyond the significance of any single meeting to center on how the committee weighs inflation, growth, and financial conditions. Markets have to demand compensation not only for upside inflation surprises in the future, but also for the opacity of the central bank’s rules; the latter feeds into bond yields, exchange rates, and equity discount rates, and this is the source of the reaction-function uncertainty premium we are seeing now.
Red indicates unanimous bearish positioning, green unanimous bullish positioning, and gray disagreement; a gold-rimmed dot indicates movement toward the upper right (as of 260802)
The annualized 1.5% growth rate in U.S. GDP GDP in the second quarter masks still-resilient private demand. According to BEA data, private domestic final purchases grew 3.9%, and consumption grew 3.2%; real PCE in 6 still rose 0.4% month over month, while headline PCE and core PCE grew 3.7% and 3.3% year over year, respectively. Nominal spending in the card data sample grew 4.1% year over year, and remained at 3.3% after excluding gasoline.
Behind resilient consumption lies a stark distributional divide.Real disposable income is expected to grow by only about 1% in 26, while higher energy prices will erode household cash flow in the second half of the year, hitting lower-income groups first.
About sixty percent of households live in owner-occupied homes and also hold stocks more broadly; gains in home and share prices over the past several years have improved their net worth;but households that rent, lack financial assets, and depend on wage income for cash flow face higher bills for rent, food, and energy.
The stability of aggregate consumption cannot hide the pressure on marginal consumers—that is, the households most likely to cut their next purchase as bills rise. There is also a growing disconnect in perceptions: politically, the cost of living and affordability—that is, whether household income can cover housing, food, energy, and credit bills—are seen as the most urgent issues; most institutions, however, are currently focused on long-term breakevens, meaning the average inflation compensation implied by the difference between nominal Treasury yields and Treasury Inflation-Protected Securities yields, and then say that everything appears normal (the former measures current bills and income, while the latter measures bond investors’ pricing of average inflation and risk compensation over many years). The White House has replaced the expiring 10% Section 122 global tariffs with new tariffs. Their current impact on the effective tariff rate is limited, but a second investigation could push rates back to their pre-Supreme Court ruling levels; if implemented, Canadian tariffs could reduce its GDP by about 0.2–0.3 percentage points and strengthen the case for the BoC to stay on hold for the rest of the year. Tariffs are already affecting U.S. core goods inflation, Canadian growth, and North American interest-rate differentials at the same time.
Aggregate activity is slowing, demand has not collapsed, and inflation is still far from a level that would let the Fed pivot easily. The macro backdrop looks more like slowing growth + resilient demand + inflation not declining fast enough; recession, stagflation, and a Goldilocks outcome have in fact all yet to be confirmed.
Red indicates unanimous bearish positioning, green unanimous bullish positioning, and gray disagreement; a gold-rimmed dot indicates movement toward the upper right (as of 260802)
Curve steepening and yields in 10 and 30 are both in Leading, while U.S. duration is in Lagging, indicating that long-duration assets, which are most sensitive to interest rates, are still underperforming relatively. The longer the duration, the greater the price decline caused by the same rise in yields. Long-end yields must also compensate investors for the inflation, fiscal supply, and interest-rate-path uncertainty involved in locking up funds for the long term; this additional compensation is the term premium. The long end is absorbing inflation, Treasury supply, policy credibility, and corporate financing needs all at once; it is not merely a passive reflection of the next FOMC.
Warsh’s approach in practice amounts to choosing a steady-state in which the market, rather than the Fed, makes the fine adjustments (View note）
The market has already tightened policy on the Fed’s behalf.Specifically, rising long-end yields increase the cost of funds used for mortgages, corporate bonds, and equity valuations; this is what it means for financial conditions to tighten. It also gives the Fed a reason to keep waiting. But if waiting is understood as tolerance of inflation or fiscal risks, investors will demand a higher term premium, the long end will rise further, and financial conditions will tighten further, creating a self-reinforcing cycle. The 5-year 5-year forward yield is currently about 25 basis points above fair value, equivalent to only about 0.5 standard deviations: it is elevated, but there is still room for overshooting.
Beyond that, watching the Fed today is no longer a game of parsing the chair’s every word; the significance of the vote count is beginning to rise. Each committee member’s reaction function, the chair’s ability to assemble a majority, and how the vote count shifts with the data will all start to enter prices. A chair being overruled by the committee remains a tail risk, but its probability is no longer zero. If inflation data remain unfavorable and the number of votes for a hike continues to grow, the front end of the curve will also face the risk of a sudden shift in the distribution of votes between meetings.
Reducing forward guidance—that is, reducing the central bank’s conditional indications about the future path of interest rates—makes this conflict harder to trade. Saying less can indeed reduce the moral hazard of markets mistaking forecasts for promises and taking excessive risks on that basis, but it cannot eliminate the “hall of mirrors”: the central bank reads economic expectations from market prices, while the market sets prices based on its own guesses about the central bank’s response; each side treats the other’s judgment as information, and ultimately may simply amplify the same misreading in a feedback loop. If the public does not understand the reaction function, apparently fewer false claims of precision will only translate into greater uncertainty. The sharper contradiction is that one side has just said inflation cannot be judged by sight alone, while the other has described the current policy stance as “watchful thinking.”If there is no clear explanation of what will be monitored while waiting, what data will trigger action, and how interest rates feed through to demand and prices, Observation and reflection are still just another way of saying “wait.”; The answers at the press conference about the dual mandate and transmission channels also felt more like wordy replies with an incomplete chain of logic. Although reducing the number of annual meetings has not become a formal proposal, it could compress what were originally spread-out adjustments into 50 or 75 basis points, raising the risk of front-end jumps and higher term premiums. At the same time, a broader definition of “price stability” than 2% PCE, but one without a public metric, would make markets feel the central bank has a secret inflation target.
This is the problem Warsh urgently needs to solve next.
Chart: Two-year Treasury yield rises above the federal funds rate again
U.S. 2-year Treasury yield and federal funds rate (1990–2026, as of 2026 year 7 month 31 day)
More anomalously, since the Fed cut rates by about 150 basis points cumulatively in 2024 year 9 month, the 30-year yield has instead risen from about 4.02% to about 5.20%. This can be read as damage to policy credibility, or explained by rising real rates, supply, and term premiums, since long-term inflation breakevens remain no higher than when Warsh took office.
So far, since 2024 year 9 month, the 30-year nominal yield has risen by about 125 basis points, of which about 108–113 basis points came from real yields, while long-term inflation compensation contributed only about 12–17 basis points. This indicates the market is not primarily pricing in an out-of-control long-term inflation outlook, but is raising the return required to hold long-term real cash flows.At a deeper level, this is not entirely due to stronger growth and upward revisions to the neutral rate; massive Treasury supply, the fiscal path, policy credibility and interest-rate volatility, together with the entire tech sector’s wild spending, have all pushed up the real term premium.Therefore, the long end is currently trading primarily on rising real discount rates; supply pressure is an important source of this increase, while inflation is merely a background risk sustaining the high premium.
US30Y (as of 260802)
There is a saying: bond traders can stop panicking only when the Fed itself starts to panic, but before action, the only thing observable is pricing itself.At pressure levels, watch the 10-year nominal yield approaching 5% and the 10-year real yield approaching 2.75%; the risk lies more in the speed of arrival.
If long bonds stop making new lows after high inflation, a poor auction, or hawkish remarks, that means bad news can no longer push prices lower, and only then is a reversal credible. Until then, crowded shorts merely provide fuel for future covering; they are not in themselves a buy signal.
Chart: Real financial conditions (market pricing) overlaid with precious metals, tech, and alternative asset composites
Financial Conditions&Precious Composite&Tech Composite&Crypto Composite (as of 260802)
...
Chart: The weekly newsletter from last weekend (Read article）
US10Y (as of 260802)
breakout.
_____________________________________________________
Second, Tech Trading
Chart: Tech leader South Korea’s KOSPI, breaking down and moving lower as expected
KOSPI (daily chart, as of 260802),Move the short take-profit from 7284.43 in last week’s weekly newsletter to 6755.77
Still in the short zone; turning bullish on tech requires corroboration.
KOSPI (daily chart, as of 260802); more micro-level content beyond swing trading is available in the tactical positioning section (https://www.meridianresearch.app/home?product=vega-trading）
SK Hynix (daily chart, as of 260726),Shorts’ take-profit level has moved from 2006k in last week’s weekly note to 1816k
Resistance below the T1 target; still in bearish territory. A shift back to bullish tech requires confirmation across assets.
Review it yourself
The scale of leveraged ETF assets in Korea fell from about 530 hundred million U.S. dollars to 150 hundred million -160 U.S. dollars. The forward P/E at the recent low was about 4.4 times, at a level more than three standard deviations below the norm. The headline forecast of 317% earnings growth in 2026 has a clear base effect, while the expected growth rate of about 38% in 2027 indicates that memory earnings have not been completely written off.
Extreme valuations and forced deleveraging have created the conditions for a local bottom, but a new uptrend requires confirmation from sustained flows and price action (watch for price action that provides a clear basis for entering and exiting, rather than blindly trying to catch the bottom!).
Beyond market forces, the Korean government has quite limited policy tools it can deploy quickly. The proposed 10 trillion won stabilization fund amounts to only about 0.22% of KOSPI’s market capitalization; the National Pension Service of Korea NPS must adhere to its long-term strategic asset allocation, making it difficult to temporarily ramp up domestic equities on a large scale to support the market. Reimposing a blanket short-selling ban would also undermine market openness and price discovery, harming Korea’s goal of securing MSCI developed-market status, so the likelihood is low. Fiscal support is relatively more likely, but would still be no substitute for earnings. The Bank of Korea may still be headed toward rate hikes, but the pace is likely to be more gradual and data-dependent; consecutive hikes at successive meetings are not the base case. Korea’s medium-term direction is still determined by memory earnings, while the short-term path continues to depend on leverage, foreign investors, and policy expectations.
SOXX (daily chart, as of 260726). 6.23 failed to resume its long position, was stopped out, and has continued downward since; the purple rectangle forms the range, and the bull-bear price levels should not be moved.
Still in bearish territory, having not reclaimed the key bull-bear price level.
STAR Market 50 (daily chart, as of 260802)
The isolated breakout failed; likewise, it remains in bearish territory.
...
Overall, KOSPI, SK Hynix, and SOXX are all currently in bearish territory, For tech bulls, resuming long positions requires at least waiting for price to reclaim the level above the bull-bear line Blindly trying to catch the bottom is not a rational choice, because you have no clear basis for entering or exiting. You may indeed catch the bottom, but you may also buy halfway down. The gains and losses here are all a gamble: those who win and survive will be celebrated, while those who lose are simply gone, and no one will care about them.
..
Chart: Hyperscaler capital expenditure, cloud revenue, and margins
..
A word on cloud: trading in cloud service providers is actually diverging today.AI industry demand is still expanding. Capital expenditure by major cloud providers is expected to nearly double year over year in 2026 to about 7300 hundred million U.S. dollars; it could exceed 1 trillion U.S. dollars annually in 2027 and 2028, while cloud revenue growth remains around 35%-45%.
The real change is happening on the financing side. Investment-grade bonds issued directly by hyperscalers rose from 1080 hundred million U.S. dollars in 2025 to 1940 hundred million U.S. dollars year to date in 2026; the full-year figure could reach 2500 hundred million U.S. dollars, and may rise further to 4000 hundred million U.S. dollars in 2027. The credit spread on 30-year cloud-provider bonds widened by about 18 basis points at one point, then retraced about half of that move. A credit spread is the additional yield a corporate bond pays over a government bond of the same maturity; a widening spread means investors are demanding more compensation for credit and liquidity risk.The market has not questioned these companies’ ability to meet their obligations, but it has indeed started to demand a higher long-term cost of funding.
Capital expenditure by AI turns into revenue for semiconductor, memory, networking, and power-infrastructure suppliers, while also consuming cloud providers’ operating cash flow, reducing free cash flow after capital expenditure, and increasing long-term bond issuance. The greater the bond supply, the more easily spreads and term premiums can rise; the resulting increase in discount rates also reduces the present value of profits far in the future. Thus, AI demand and supply-chain revenue can be strong even as risks to cloud providers’ equity and long-term credit rise at the same time.
If you want to express concern about the returns on AI capital expenditure, long-term cloud-provider credit spreads are a closer fit to the issue itself than directly shorting the overall AI equity beta. Equity beta bundles together revenue growth, margins, buybacks, index flows, and market risk appetite; even if returns on capital expenditure deteriorate, other factors may still support share prices. Long-term credit, by contrast, is more directly exposed to debt supply, free cash flow, and duration risk. That is the logic behind shorting 30-year investment-grade cloud-provider bonds: the more they issue, the weaker their cash flow, and the higher the credit compensation investors demand, the more likely their bonds are to fall in price relative to government bonds. The trade still requires separate consideration of spreads that have already widened, cash-bond liquidity, and duration gains or losses caused by changes in government bond yields, but its chain of assumptions is at least shorter than “AI has poor returns on investment, so the Nasdaq must fall.”
The AI shock also needs to be divided into “how value is distributed” and “whether the overall pie is shrinking.”China’s open-source models and lower unit inference costs first change the distribution of value among models, cloud, chips, storage, networking, applications, and users. They will increase single-stock volatility, reduce correlations, and create a changing lineup of winners, but lower costs may also expand adoption, which may continue to run on U.S. cloud platforms.Only if companies cannot turn AI of investment into revenue and cash flow, leading cloud providers to cut capital expenditure, would that constitute a shock to the overall pie.The former shock has already emerged, while the latter remains an important tail risk and has yet to be confirmed by companies’ capital expenditure behavior. If the latter occurs, stocks and long-term yields could fall together, rather than the move remaining an internal rotation within tech.
Beyond that, U.S. earnings show no evidence of a fundamental collapse (at least not so far). Approximately 80% of S&P companies have reported earnings above expectations; reported earnings grew 36% year over year, including a one-off contribution from the revaluation of Google’s investment. Excluding that, growth was still about 24%. Earnings guidance and analyst revisions have generally remained healthy. 36% includes one-off items, so extrapolating it as is would overstate the trend; 24% excluding those items is likewise sufficient to rule out the explanation that “the tech decline is entirely due to an earnings recession.”
Chart: This week’s Prime Book deleveraging was the second largest in the past ten years
Beyond the narrative. Global tech long selling reached an all-time extreme in the reported sample, and year-to-date flows into semiconductors turned negative. The momentum basket had risen about 57% at one point in the first half, then fell about 30% in 7 month. (Prime Brokerage is a business through which investment banks provide hedge funds with financing, securities lending, clearing, and position management; its aggregated data offers a closer view of what risks leveraged capital is actually adding or reducing. The data recorded a bout of de-grossing extreme on a ten-year scale: leveraged accounts cut both long and short gross exposure at the same time, shrinking their balance sheets to a lower-risk level. Investors do not have to first change their long-term views; rising volatility, financing constraints, or redemption pressure alone can force positions down, so prices can fall ahead of fundamentals.) Some of the unusual selling pressure came from the forced liquidation of Situational Awareness, whose assets fell from about 450 hundred million dollars to about 100 hundred million dollars; Citadel took over some of the distressed assets.
Still, it is a classic Wall Street story: when a strategy puts the entire portfolio into one theme and keeps adding leverage as it rises, the so-called genius eventually becomes a target for everyone. The fund at one point used leverage of about 400%,The tech rebound this week may therefore also be just a “dead-cat bounce”—a sharp rebound after forced selling temporarily stops, which in itself does not prove that new long-term buying has taken hold.This decline at least partly involved forced trading, and overhead supply from buyers who entered near previous highs may also constrain the rebound: when prices return to around their cost basis, these underwater positions may take the opportunity to sell, limiting the rebound’s height.The share of the overall price move attributable to forced trading still cannot be identified precisely. A genuine “sellout” needs to be confirmed by volume, the number of advancing stocks, and sustained buying by medium-term capital; one fund’s story is nowhere near enough.
Systematic models show that major trend positions in the Nasdaq have already been substantially reduced, while the S&P, Russell, and European equity indices still retain considerable long exposure.
Chart: Within tech
Daily chart, as of 260802
On our monitoring dashboard, the global tech complex is in Weakening, while U.S. tech relative to Asian tech has entered Improving, but has not yet returned to genuine leadership. The most crowded positions have already retreated ahead of fundamentals, but whether the AI cycle has ended and whether the tech trend has recovered remain unanswered. The key question is:Keep a close eye on the cross-validation long-short lines across the planets.
...
Finally, as promised, the Tech Monitor for all our friends(This section will not be updated indefinitely)：
Chart: Mainland China tech trading, Basket as an Asset
Daily chart, as of 260802
The color indicates the overall long-short direction of Basket as an Asset. The unrealized profit in the chart shows the unrealized profit in the basket’s current position direction. B at the bottom is short for Bars, meaning how many trading days the current long-short direction has lasted; the % in the upper right indicates the number of constituents held long in the Basket.
Note: Having more than 50% of Basket constituents held long does not necessarily mean Basket as an Asset itself turns long; having more than 50% of constituents held short does not necessarily mean Basket as an Asset itself turns short.
Chart: Global tech trading, core micro-proxy
Daily chart, as of 260802
The color indicates the micro-proxy’s own long-short direction. The unrealized profit in the chart shows the unrealized profit in the current position direction. D in the chart is short for Days, meaning how many trading days the current long-short direction has lasted.
Chart: Three-year effectiveness of Long&Short
Daily chart, as of 260802
Further elaboration: 1)View note，2）View note，3）View note
...
Two further points. This section addresses the discussion in last week's and the week before's newsletters about tokens and the so-called power thesis. Beyond total Capex, the most important factor on the demand side is unit economics: for every dollar AI a company spends, how much labor and software expense can it save, and how much additional revenue can it generate? A token is the basic unit used to bill and compute when a model processes text, images, or other information. The median enterprise employee currently generates less than 11 dollars in token spending per month; an economic task may cost only AI 2–5 dollars, yet save a company about 55 dollars. If this relationship holds, lower model prices do not necessarily reduce overall demand for computing power. The Jevons paradox describes precisely this situation: the lower the cost per use, the more previously uneconomic tasks begin using AI; the frequency and scope of use grow faster than the unit price falls, so total consumption rises instead.
The sources of corporate adoption budgets are also quite specific. When an AI agent can independently handle spreadsheets, charts, emails, and workflows, companies cut existing software subscriptions and shift that budget to models and computing power; AI spending does not need to come entirely from new IT budgets.
According to widely cited estimates, the number of tokens may grow by about 24 times over the next five years; an extremely bullish scenario even assumes that computing demand doubles every six months. In any case, price deflation and growth in unit volumes are two sides of the adoption process.
As for power, we addressed it separately last week because power is indeed closer than financing to being a constraint on actual progress. The available power capacity of hyperscalers could rise from about 30 GW in 2025 to about 120 GW in 2028. US grid capacity under construction or contracted is about 30 GW, less than half the 68 GW expected to be needed by data centers. Grid connections in some areas require a wait of five to seven years, and skilled electricians, welders, and pipefitters are also in short supply. On the supply side, companies are bringing the shale-era fracking playbook to data centers: using on-site generation, fuel cells, energy storage, gas turbines, and retrofitted existing high-capacity sites to bypass slow grid approvals, trading modular construction and regulatory adaptation for speed of delivery. This is more likely to defer revenue than to end demand.
The real trigger for the dangers of overinvestment may instead be a leap in efficiency. If a new model or chip enables the same power and data-center capacity to support more than ten times the effective computing power, the economics of existing projects will be reassessed;The volatility triggered by Chinese open-source models reflects the market's concern that an efficiency shock could arrive before demand expands.This concern has the same roots as concerns about the internet at the time. Back then, people were not wrong about demand for the internet, but by 2001, the amount of fiber-optic cable laid exceeded what humanity would ultimately need.
...
_____________________________________________________
Three. Global Equities
Figure: Positioning Map
Global Equities: Unusual Activity in Structural Trades (as of 260802)
Global Equities: Unusual Activity in Economic Trades (as of 260802)
The two main trade themes reverse (Technology vs. Resources) (as of 260802)
Red indicates unanimous bearish positioning, green unanimous bullish positioning, and gray disagreement; a gold-rimmed dot indicates movement toward the upper right (as of 260802)
Peripheral European markets have moved into Leading; this was discussed in detail in an earlier newsletter. Latin America's improvement is consistent with the appeal of FX carry. Overall, EM leadership remains unconfirmed.
Figure: US Cross-Border Fund Flows
Cumulative cross-border fund flows year to date in 2022 (as of 260731)
KWEB (daily chart, as of 260802),The long position's take-profit level has moved up from around 25.5, as noted in the previous newsletter, to above 26.9.
_____________________________________________________
Four. US Equity Market
SPX (daily chart, as of 260802)
Ever since we turned bearish on SPX from 6.3, it has been firmly and precisely capped by the bull-bear line, Before a breakout, It is difficult for the US session to produce a major move in any meaningful sense. I mean that outside technology, it is also difficult for other sectors to sustain a genuine primary uptrend.
Figure: Positioning Map
Red indicates unanimous bearish positioning, green unanimous bullish positioning, and gray disagreement; a gold-rimmed dot indicates movement toward the upper right (as of 260802)
Top performers over the past week (as of 260802)
Figure: US Momentum Environment
Strong momentum above the green line; above the yellow line is where we can take shots. Below the red line is our disadvantage zone, where the market will tend toward random fluctuations (as of 260802).
The red line bounced; someone wasn’t willing to let it break decisively below the red line.
MEME (daily chart, as of 260802)
Take profit on the short and exit. Put the long stop at the lowest point of sentiment in this cycle. This is a very high-risk area, so don’t go long lightly. Actually, it’s best to wait for mutual confirmation; this trade is also an aggressive, early entry.
My trend lines are meant to help everyone get oriented. They mark points of qualitative change beyond changes in quantity. What you need to learn is to watch for qualitative change: it leads to wider price swings, and only wider swings bring profits.
...
Chart: Systematic funds’ selling elasticity on the downside is greater than their buying elasticity on the upside
Keep this in mind: according to Bank of America’s estimates, systematic strategies may continue selling about 770 billion dollars during a decline; during a sideways market, they may buy 330 billion dollars; and during a rally, they may buy 190 billion dollars. The S&P spot reference is around 7518, and the main CTA deleveraging trigger zone is around 7305–7326. “Elasticity” here means how much additional buying or selling is triggered in rule-based strategies for each move in a given direction. A decline simultaneously raises volatility, damages the trend, and lowers target positioning, so the selling response is far greater than the increase in buying during a rally. The amounts and levels shift with price, volatility, and time decay, so they are best treated as a flow map rather than fixed targets. The key point is that mechanical flows exist on both sides, with greater potential scale on the downside. If the index breaks below the trend trigger zone, renewed short positioning and CTA mechanical selling could create flows in the same direction. If cash buying continues and the rally no longer relies on short covering in futures, the market structure would be healthier.
In addition, according to GS PB, its clients have been net buyers of U.S. stocks for a fourth consecutive week, with net purchases of individual stocks reaching a record 82 billion dollars. Institutional buying was the strongest since 2020 12, retail investors also bought, while hedge funds sold for a third consecutive week.
...
As we enter a new month, 8 and 9 are usually not the most favorable months for stocks; 10 through 12 tend to be more positive. I hope everyone can continue to make rational, well-grounded decisions about when to enter and exit in the new month.
_____________________________________________________
Five. Foreign exchange: the Korean won, the U.S. dollar, and the Japanese yen
Continuing to track the Korean won as a point of cross-asset confirmation.
A decline in the Korean won can bring broader cross-asset issues (seeWeekly 260614）
USDKRW (daily chart, as of 260802), with KOSPI overlaid. The short take-profit level has moved up from above 1484 last week to above 1455 (below the T1 target).
In fact, the decline in KOSPI did begin in sync with the move in the Korean won. This is another cross-asset relationship that needs confirmation. As we’ve said many, many times on Planet:
Why does cross-asset logic matter? Because within a single market, you have no pricing anchor.
Those of you working in tech should not only follow the discussion of tech proxies in Part One, but also keep a close eye on the movement in KRW.
...
However, from this point on, the Korean won and KOSPI will not simply move in the same direction.
A moderate recovery in the stock market would unwind excessive currency hedges put in place earlier to protect against a stock-market decline. Companies converting export dollar revenues back into won, along with the repatriation of funds associated with the conversion of about 260 billion dollars in SK Hynix ADR, could also support the won. ADR is a depositary receipt listed overseas; when it is converted into locally listed Korean shares, shares and funds move between the two markets. Excluding equity flows, there could still be structural foreign-exchange inflows of about 120–130 hundred million dollars per month in the second half of the year.
Once KOSPI rapidly recovers its losses, pension funds often need to increase overseas assets or dollar hedges to restore their target allocation between domestic and foreign assets, creating fresh demand for dollars instead; the faster stocks rise, the further asset weights deviate from their targets, and the sooner rebalancing pressure may emerge in the currency market.
...
Chart: Why is the dollar falling?
DXY (hourly chart, as of 260802)
After breaking below dual support, 27 made several attempts with wick candles, then broke below the red line and the price range widened.
...
In fact, if you’ve been paying attention,
we were the only ones to give a clear warning of yen intervention:
Chart: Ari’s Weekly Wrap 2607227Read article
....
As flagged in last week’s weekly wrap, dollar-yen duly saw a squeeze last week. The timing of the collapse also matched expectations, occurring almost simultaneously with the AI trade led by U.S. semiconductor stocks. At the open, it gapped down SOXS 18.23%; a minute later, forced liquidations in yen trades began...
Chart: USDJPY overlaid with SOXS (1-minute chart)
The apparent scale of this round of intervention was about 8.45 trillion yen, or about 528 billion dollars; the New York Fed also conducted rate checks on USDJPY and EURJPY. A rate check asks for tradable or indicative prices. Its signal strength falls between verbal intervention and actual transactions; absent follow-up action, its marginal impact usually fades. The impact of selling foreign-exchange reserves on 30-year Treasuries is also less direct than one might imagine: reserve management favors highly liquid short- and intermediate-term assets, and intervention primarily changes the exchange rate and dollar liquidity—it does not dump long bonds onto the market all at once.
The Bank of Japan held its policy rate at 1%. The next rate hike may come in 10, followed by another in 2027 4, with the terminal rate around 1.5%. Such a gradual path is unlikely, on its own, to offset the U.S.-Japan interest-rate differential and residents’ ongoing allocation to overseas assets. The historical relationship between a stronger yen and global risk-off conditions has also weakened recently: when global risk aversion rises, capital no longer reliably flows into the yen.Yen longs after intervention therefore cannot simply be treated as a substitute position for shorting stocks.
Chart: Price action after this round of USDJPY intervention resembles that of 4 30
Chart: Comparison of USDJPY trends from 2026 4 30–5 5 and 7 30–31
The policy chain behind the yen is longer than a single intervention. Genuine coordinated U.S.-Japan intervention has typically occurred in crisis conditions, such as Japan’s financial crisis in 1998 or the Great East Japan Earthquake in 2011; the current situation is far from that threshold. The New York Fed’s rate check raised the signal’s strength, but still did not constitute coordinated action. As the limits on foreign-exchange intervention capacity become clearer, Japan has begun discussing using the Government Pension Investment Fund, GPIF, to increase domestic asset allocations, in the hope that pension funds will buy fewer overseas assets and reduce persistent yen selling. The primary responsibilities of pension funds are long-term returns and risk management; strategic allocation constraints mean they cannot rebalance without limit to meet short-term exchange-rate goals. Only after intervention and GPIF both show diminishing effectiveness might the government be forced to accept faster BoJ rate hikes. We have not reached that point yet.
Looking ahead, narrowing interest-rate differentials, capital repatriation, and sustained coordination would all need to occur together to change the yen’s long-term direction.
Without further policy action, USDJPY could still drift toward the middle of 160. If policy can push the exchange rate past around 155, exporter hedging and trend positions may change significantly. At present, USDJPY is Weakening, and the broad dollar is also Weakening. The yen rebound has addressed the recent path, but it has not declared the end of the dollar’s medium-term trend.
...
Chart: FX trade positioning
Red indicates unanimous bearish positioning, green unanimous bullish positioning, and gray disagreement; a gold-rimmed dot indicates movement toward the upper right (as of 260802)
_____________________________________________________
Six. China markets
Chart: Hang Seng Tech
HSTECH (daily chart, as of 260802)
Hang Seng Tech, For now, the level for turning bullish again remains unchanged: price needs to hold above 4910 at a minimum.
Closed precisely on the gold line.
Chart: Pair trades and narrative trades
Pair-trade performance (daily chart, as of 260802)
In mainland China trading, monetary rates, defensives, staples, chip spot markets, energy storage, resources, old-guard stocks, old energy, consumer electronics, stable demand, non-memory, software, and computing expansion outperformed monetary rates, cyclicals, services consumption, semiconductor capacity expansion, power generation, technology, new energy, chips, cyclical growth, memory, hardware, and interconnect bottlenecks.
Narrative-trade performance (daily chart, as of 260802)
Figure: Positioning Map
Red indicates unanimous bearish positioning, green unanimous bullish positioning, and gray disagreement; a gold-rimmed dot indicates movement toward the upper right (as of 260802)
Both bull markets and bear markets can make money; the market that cannot make money in is one without logic. Without logic, prices cannot unfold naturally, and without a natural price move, there is no trend. Profits come only from smooth trends, regardless of their length.
Abnormal-move monitoring for Basket (Top/Bottom 10) (1M, as of 260802)
The mainland’s official purchasing managers’ index PMI for 7 has fallen back into contraction territory: manufacturing 49.2, services 49.3, and construction 47.0. 50 is the dividing line between expansion and contraction; a reading below 50 means fewer businesses reported improvement than reported deterioration. The Politburo meeting focused on accelerating the implementation of existing policies and announced no new broad-based stimulus. The maximum fiscal resources already approved and available for deployment stand at approximately 7.5 trillion yuan. This represents available fiscal capacity, not 7.5 trillion yuan of new stimulus already put in place. Policy still has firepower, but current demand and construction activity will not automatically improve as a result.
Market prices have also not confirmed that a new China theme has emerged. Across 194 China thematic baskets and 1959 effective constituents covered through 7 31, the no_mainline state persists, meaning no theme has yet met the criteria for a leading theme in relative strength, trend persistence, and breadth across constituents. There are currently only 8 challengers and 1 fading themes: a challenger is a candidate close to the threshold for becoming a leading theme, but still lacking sufficient persistence or internal breadth; a fading theme is a former leader that remains relatively strong but has begun to lose momentum and participation. Data-center switches have moved from former leader to fading. Construction machinery, VLCC oil shipping, and white goods are forming, meaning their prices and internal structures are taking shape but have not yet been fully confirmed. Banks, PTA, refining and chemicals, copper resources, and potash are in the spark stage, showing only early signs of unusual activity. Banks have already entered Leading, and PTA has just entered Leading as well. This means their relative strength and momentum have moved into the leading quadrant, but persistence and breadth across constituents are still insufficient. Calling them confirmed leading themes now would overstate the current structure.
The global memory upcycle and the price structure of A-share technology stocks have diverged markedly. Long-term contracts, prepayments, and price-increase reports involving Samsung, MLCC, and the AI server supply chain are strong, but wafer manufacturing, CCL, semiconductor front-end equipment, data-center switches, and MLCC in the A-share market we monitor are mostly still in Weakening, with MLCC down about 31.9% over the past month. The improvement in global orders has not translated into relative price leadership for A-share technology stocks. The industry thesis may be sound, but the timing of the trade is still wrong;Only when technology themes exit Weakening, gains spread to more constituent stocks, and the improvement persists across several observation windows will fundamentals once again genuinely translate into a leading local theme.
_____________________________________________________
VII. Precious Metals
Figure: Gold
CMX Gold (daily chart, as of 260802)
Gold: still long
GDX (daily chart, as of 260802), long stop at the red line
Gold’s long-term rationale and short-term price action remain in conflict. Fiscal risks, policy credibility, and tail risks support holding convexity through options—that is, paying a limited premium for the potential of an accelerated rise in a crisis scenario. Yet gold remains heavy during the Nasdaq rebound, and short-term price action is not cooperating. CTA short positioning remains high, meaning trend-following strategies are still holding sizable gold shorts in accordance with their rules;
Gold has now moved out of the low-relative-strength zone and into Improving. Gold itself pays no interest, and rising real yields increase the opportunity cost of holding it relative to bonds; real yields continuing to rise while gold stagnates shows that this cost still outweighs the fiscal narrative.A cleaner confirmation would require all three to occur together: real yields stabilizing at a high level, the dollar stopping its advance, and gold continuing to improve.
Figure: Silver
CMX Silver (daily chart, as of 260802)
Silver and gold broke out almost in sync. Silver, too, remains long; stop at the red line.
A precise close at the gold line.
The precious-metals trade is the same story.
SIL (daily chart, as of 260802), long stop at the red line
_____________________________________________________
VIII. Alternative Assets
Xiao Huang (daily chart, as of 260802)
The stop was triggered; exit according to plan. To turn long again, price needs to break above the red line.
...
Bitcoin still looks heavy during the Nasdaq rebound. So-called high beta means that its price usually amplifies the ups and downs of overall risk assets more sharply; when the Nasdaq rebounds but Bitcoin fails to keep up, it indicates that the market's willingness to take marginal risk remains limited. Bitcoin has now moved from Low into Improving, but this is still a low-level recovery, far from a confirmed trend.
Miners are being re-embedded in the AI data-center financing chain. Some miners or powered-shell operators finance power and data-center assets through high-yield bonds, secured debt, loans, ABS, CMBS, and private credit; their tenants may be cloud providers or AI labs. Their equity risk thus extends to tenant credit, project delivery, lease backstops, refinancing, and concentration among the same group of cloud providers. Bitcoin prices, computing power, and electricity prices are only part of the picture.
The reason for emphasizing this here is the recent divergence; Bitcoin's own low-level recovery and the credit risks of miners' transformation are two different kinds of beta, and treating them together would obscure the true sources of risk.
Note: Please comply with the laws and regulations applicable in your region and allocate your personal assets within the bounds of compliance. This content is intended only for users in certain overseas jurisdictions.
Ninth, soft commodities (El Niño): cocoa, coffee, soybeans, raw sugar, and rice
Global food inflation has eased considerably, although fertilizer and supply shocks remain risks over the next six to nine months.Take Germany as an example: in 7, food inflation was only about 0.4% year on year, at odds with signals from some food manufacturers' surveys that they are preparing to raise prices.
Chart: Cocoa, back to long
Cocoa (daily chart, as of 260726)
Long-short-long-short-back to long. The point of swing trading is to keep up as much as possible with all abnormal price moves. An abnormal move is about common-sense if-then logic: if... the price continues to rise or fall, then where should it not go? If it does, then...
This is about the payoff, not the probability, although sometimes our win rate is not inherently low either. But focusing on anomalies is indeed our habit. Here, the price reached a level it should not have reached if it were going to keep falling, so we switched back to long, even if it might end in a loss. Interpret that as you will.
Quite often, stop-losses are a cost you have to pay to run this business.
Stop-loss: see the red line.
Chart: Coffee, breakout above the upper boundary
Coffee (daily chart, as of 260802)
The bottom-fractal reversal trade mentioned in the previous weekly newsletter: the rise has stalled, so for now do not move the take-profit level.
Chart: Soybeans, switched from long to short
Soybeans (daily chart, as of 260726)
Strictly follow last week's newsletter plan: switched from long to short; stop-loss: see the red line.
Chart: Raw sugar
Raw sugar (daily chart, as of 260802)
Still trapped in the structure we pointed out a few weeks ago...
Chart: Wheat
Wheat (daily chart, as of 260802)
Wheat, likewise, broke below the stop; see the red line.
Ten. Crude oil-related
The long stop was triggered; exited according to plan. If I go long again, I’ll continue to update.
...
In fact, there is a negative feedback loop between oil prices and policy that could fail at any time: policy pushes oil prices higher, and oil prices in turn force policy to moderate through inflation and the political cost of gasoline, thereby restraining further gains.When oil prices are low, the United States has more room to escalate pressure on Iran; near 100 dollars per barrel, political costs force policy to cool down, so in the past, statements have often pushed oil prices back down.
The regime shift occurs after inventories, straits, or refining bottlenecks begin to dominate. Before that, policy choices determine oil prices, making oil prices a “dependent variable” under the influence of policy; once supply is actually disrupted, oil prices constrain policy in turn and become an “independent variable” that policymakers have to accept.
The snapshot as of CFTC 7 21 shows that WTI rose about 40% over several weeks, while large speculators’ shorts remained near 22.8 ten thousand contracts, a five-year extreme. Gross longs—the total number of long contracts held by speculative accounts—actually fell from about 38 ten thousand to just above 30 ten thousand.The rally did not rely on a large influx of new longs, and shorts have not capitulated;If prices continue to rise, shorts being forced to buy back will create additional demand, providing fuel for a further squeeze. The data are lagged and cover futures accounts only; if diplomacy again suppresses the risk of physical disruption, these shorts may also remain in the market.
The pullback in outright crude prices—that is, the absolute price of a particular delivery month most often cited in the news—conceals the worsening physical situation. Brent has fallen about 11% from the previous week’s high, while flows through the Strait of Hormuz over the past seven days were only about 310 ten thousand barrels per day, back to the 5 low. In the Red Sea, Houthi actions have cut Saudi oil-tanker flows through the Bab el-Mandeb Strait by about half, to 150 ten thousand barrels per day, forcing at least 100 ten thousand barrels per day to be rerouted. The detour could extend the VLCC voyage by about 40 days. As a result, the number of voyages the same vessel can make in a year falls, effectively shrinking shipping capacity while raising logistics costs and narrowing the discounts available to Asian buyers. The fact that prices have stopped surging reflects the market’s still-high probability of a successful short-term negotiation; it does not mean supply has recovered.
The crude oil curve is a better reflection of current physical tightness than the flat price. Front-month spreads are the premium of nearby contracts over deferred contracts: the scarcer spot supply is, the more buyers are willing to pay for immediate delivery. Regional differentials reflect local supply and demand in a particular producing area, port, or grade of oil. Both already reflect the supply crisis, while the flat price and volatility are being held down by expectations that an agreement may be reached in the coming weeks. The cleaner long exposure now lies in refined products and natural gas, because they are harder to replace via gray-market shipping or alternative routes. The restart date for the Saudi refinery hit in the attack is reportedly around 8 15, further tightening refined-product supply. The pullback in Brent may therefore understate the energy inflation households actually face: the crack spread is the selling price of gasoline, diesel, or jet fuel minus the cost of crude oil input; when refining capacity or transportation is constrained, it can continue to strengthen even as the crude flat price falls. If the strait reopens quickly, cheap crude put skew means out-of-the-money puts are relatively inexpensive. Using options to go short caps losses at the premium while providing accelerating gains if oil prices plunge, offering more convexity than simply chasing a short position in futures.
...
There is another possibility: the risk around Hormuz could shift from a blockade to ongoing toll collection. Iran and Oman might try to frame fees as charges for navigational safety, traffic management, escort, emergency response, and environmental services, rather than as transit taxes that violate the passage principles of the United Nations Convention on the Law of the Sea (UNCLOS). There are precedents for various forms of service charges in the Turkish Straits, Danish Straits, and Russia’s Northern Sea Route. According to estimates, a VLCC round trip could cost about 26 ten thousand dollars; a Q-Max LNG vessel, about 13 ten thousand dollars, equivalent to around 0.02 dollars/MMBtu. The real obstacles are whether the international community would recognize the fees as legal, who would collect and enforce them, and whether payment would trigger US sanctions;
If settlement in dollars is the most efficient option, the fee arrangement could instead provide a diplomatic channel for US-Iran negotiations. It would not remove all supply immediately, as a blockade would; keep an eye on freight rates.
There is a sound basis for both moving forward and stepping back.
I’m grateful for everyone’s kind words, but I’m not important and have no intention of becoming famous; I prefer to keep my head down and work quietly. The content on our Planet is for internal use; we try not to share it externally.
Thank you all here. Many thanks to all our friends.
...
(End of full text)
Used together: the US session, A/H markets, and the commodities Planet, Only commit if you really need it I know everyone is supportive, but don’t waste money or spend it carelessly.
US stocks:https://www.meridianresearch.app/home?product=us-strategies(Xingzhi)
A/H markets:https://www.meridianresearch.app/home?product=asia-strategies(Harbour Dragon)
Commodity:https://www.meridianresearch.app/home?product=commodity-strategies(Earth's Core)
Note: All content in this document is for reference only. It does not constitute investment advice or serve as a basis for any laws, regulations, or regulatory policies. Investors should not use or rely on such information as a basis for decision-making.
Note: All content posted in the Planet is for sharing and reference only. Recipients are not considered our clients by virtue of reading content in the Planet. Nor does it constitute an invitation to anyone to sell or purchase securities or other investment products, or a basis for any transaction. Markets involve risk, and investing requires caution. All information and opinions in the Planet do not constitute investment advice. Past performance is not indicative of future results. Investors should carefully assess their own risk tolerance, make independent decisions, and assume full responsibility. Under no circumstances shall we be liable for any losses arising from the use of content in the Planet. Relevant data may be delayed or inaccurate, and we accept no legal responsibility for the completeness or accuracy of the information.
Note: All content in this document is for reference only. It does not constitute any investment advice, nor does it serve as a basis for any laws, regulations, or regulatory policies. Investors should not use or rely on such information as the basis for decision-making.