On Thursday, SPY fell 0.16% and QQQ fell 0.37%. But stretched over the past 5 trading days, SPY has still risen 3.62% and QQQ has risen 4.55%. Today the market didn’t drop much on the surface; what’s really tangled underneath are the yen, interest rates, bank credit, and AI’s cash burn. ETF Radar sees a “calm on the surface, deep transmission.”

First, look at the primary thread: BOJ and yen normalization.

There has been a lot of Japan-related news in the past two days. The core message is simple: the yen is still weak, Japan’s domestic strength is limited, and outsiders are even discussing whether to use Federal Reserve tools to help Japan stabilize dollar funding.

Put plainly, Japan now resembles a company that imports a lot of raw materials: when the local currency weakens, purchases become more expensive, and that pressure transmits to prices, corporate costs, and the financial system. But if Japan forcefully raises rates to save the yen, it could hurt domestic growth.

Why does the US market care? Because the market’s earlier worry wasn’t just the yen’s depreciation itself, but whether Japan, to defend the currency, would be forced to sell the US Treasuries it holds and convert them into dollars to extinguish the fire.

Imagine a large holder urgently selling Treasuries to raise cash: bond prices would be pushed down and yields would go up. Yields can be understood as the market’s measure of borrowing cost. If that measure rises, US equity sectors most sensitive to higher rates—like small caps and real estate—come under greater pressure.

The key in the recent reports is that the market is discussing a way to “first provide dollar liquidity to Japan.” Liquidity, in plain terms, means whether there is enough cash in the market to turn over. If Japan can borrow dollars first instead of immediately selling Treasuries, that large link—“yen crisis leading to a Treasury plunge”—would be largely cut off.

Thus, although the US large-cap market has been a bit hesitant, it hasn’t experienced a runaway drop. The fact that SPY rose 3.62% over the past 5 trading days carries a bit of that easing in the background.

But the second thread immediately pulls the tone back down: US banking and credit stress is still surfacing.

A striking set of headlines today fell on mortgages and real estate financing. One of the largest US mortgage institutions, United Wholesale Mortgage, was reported to have derivatives losses continuing to squeeze it; at the same time, there are federal-level pushes to regulate home equity investments; and with second-quarter mortgage, real estate, and homebuilder earnings rolling in, the market is re-focusing on “how tight high interest rates are squeezing real estate financing.”

The transmission chain here is easy to understand. If high interest rates persist, mortgage payments don’t fall, homebuyers become more cautious, developers sell homes more slowly, and institutions that live off financing are more likely to run into trouble. Further down the chain, commercial real estate, office loans, and mortgage-backed securities will be repriced.

Put bluntly, this is like the upstream of a river having less water. Upstream is cheap financing; downstream is real estate transactions, loan origination, and asset valuations. When upstream tightens, downstream all becomes leaner.

At the same time, another headline today moved the “not enough cash” theme from real estate to the tech giants.

Alphabet is reportedly planning to borrow another 200 to 250 billion USD. And this comes after it already issued 500 billion USD of bonds and 850 billion USD of equity this year. The reason is straightforward: AI infrastructure is extremely cash-intensive.

This is not just a matter of “tech companies expanding.” It shows that the most profitable, best-financed large companies still need to keep raising money to compete in the AI arms race.

In plain terms, building data centers, buying chips, and building power and network infrastructure is like constructing many high-speed rail lines at once: heavy initial capital expenditure, cash flows out first, and returns take a long time to manifest.

How will this transmit to the market? First, when rates are high, borrowing to expand is more expensive. Second, capital is more likely to concentrate in the “largest, strongest companies,” leaving smaller firms disadvantaged. Third, if the market starts to doubt that AI spending will pay back quickly, the semiconductor, cloud, and software chain will see divergence— not every “AI concept” can rise together.

So the various headlines today actually reflect the same underlying theme: the world is wrestling with “the cost of capital.” Japan fears exchange-rate and capital outflows, US real estate fears high rates tightening credit, and tech giants fear AI’s cash burn. The market not collapsing doesn’t mean the problems are gone; it only means pressure is still being transmitted in different corners.

There are countercurrents. On Thursday, energy and gas were strong: BNO rose 4.32% in a single day, and VDE rose 1.58%. That shows the market isn’t in full-on risk-off mode; it’s more like worrying about rates and credit while re-pricing geopolitical and energy risk.

Next, look at several most relevant ETFs.

First, SPY, the S&P 500 large-cap ETF.

Over the past 5 trading days, about a week, SPY actually rose 3.62%. ETF Radar’s model is biased positive for SPY over the next 1 to 3 months, with fairly consistent signals.

Its quantitative support: the first is that the CFTC COT net S&P position is still net short, but week-on-week it converged by 25,389 contracts to -297,476 contracts. COT net position, in plain terms, is the net directional stance of large players in the futures market after hedging. It is still net short, but the convergence indicates the shorts are less crowded.

The second is that SPY’s excess return in this theme window is +1.98%. Excess return, in plain terms, is the return above the benchmark, indicating that the easing signal has already been partly priced in by the market.

The third is the cash flow picture: since July 29, SPY’s net creations/redemptions are -97.24 billion USD, and over the past 5 days funds have also seen net outflows of -0.71% AUM. AUM is the total assets under management. Net outflow, in plain terms, means some capital is withdrawing; not everyone is chasing the rally.

So there is an important divergence here: SPY rose over the past week and the model is positive, but the cash picture is not hot. That suggests the large-cap rebound is more like “the level of worry has diminished” rather than “everyone is newly confident.”

Second, TLT, the long-term US Treasury ETF.

Over the past 5 trading days, about a week, TLT actually fell 0.28%. Looking forward, ETF Radar’s model is biased negative for TLT over the next 1 to 3 months, with fairly consistent signals.

The first quantitative evidence is that the CFTC net position for 30-year US Treasury futures is -389,522 contracts. Such a large net short means large players in the futures market are overall betting on pressure for long-term bonds.

The second is actual creation/redemption flows: since July 30, cumulative net outflows are -5.87 billion USD, about -1.43% AUM, and over the past 5 days funds have continued to see net outflows of -1.94% AUM. In other words, money is moving out of long-term Treasury ETFs.

The third is that TLT in this window has cumulatively fallen 4.32%, and in the past two days its excess return versus the theme is -1.04%. This indicates the market has begun trading on the idea that long-term rates are under pressure.

That aligns with today’s main thread. As long as the market worries about persistent high rates, sticky inflation, or unstable overseas funding, long-term bonds are vulnerable. Duration, in plain terms, is how sensitive a bond is to interest-rate changes; TLT has long duration, so when rates rise it tends to fall more.

Third, IWM, the US small-cap ETF.

Over the past 5 trading days, about a week, there isn’t a separate price table for IWM, but the system’s relevant window shows it is still short-term relatively strong, with an 8-day cumulative gain of 3.62%. As for the outlook, the model readings point to a negative bias for IWM, with multiple pieces of evidence aligned.

First, the Russell 2000 CFTC net position is -74,620 contracts, sitting at the 85th percentile of its own historical range. Percentile, in plain terms, means compared to its own history, current readings are tilted toward the short side.

Second, prediction markets show a 0.66 probability for a “rate hike again in 2026.” Prediction markets, in plain terms, are where many people bet real money on whether an event will occur. 0.66 means the market assigns a non-trivial chance to another hike.

Third, IWM’s net outflows since July 30 are -15.79 billion USD, or about -1.93% AUM, and over the past 5 days funds have continued to flow out by -1.28% AUM. Capital is leaving, which shows big money isn’t comfortable with small caps.

The divergence here is clear: short-term prices are still relatively strong, but the model is negative. The reason is simple: the model looks at the next 1 to 3 months, not just the past few days. As long as borrowing costs are high and small companies face weak financing and thin profits, they are more likely to be squeezed.

Lastly, ITA, the aerospace & defense ETF.

Over the past 5 trading days, about a week, ITA actually rose 4.98%. ETF Radar’s model is biased positive for ITA over the next 1 to 3 months, with fairly consistent signals.

Quantitatively, first, ITA has outperformed SPY by 10.78% within the theme window. Outperforming the broad market by that much indicates the market has priced in some defense premium.

Second, ITA has risen 4.38% over the past 8 days with a maximum drawdown of only -3.56%. Maximum drawdown, in plain terms, is the deepest decline from a peak during that period. A shallow drawdown suggests price resilience.

Third, at the ETF level since July 30 there have been net inflows of +1.12 billion USD, about +0.75% AUM, and over the past 5 days net inflows of +1.73% AUM. Add to that an options implied volatility of 31.74%, sitting at the 92nd percentile recently. Options implied volatility, in plain terms, is the options market’s expectation of future volatility; a high percentile means the market is pricing higher risk and hedging demand.

This thread doesn’t have a direct news trigger tied to today’s headlines; it’s more of an independently stronger medium-term stance. But it reminds us that while geopolitical and energy risks haven’t fully abated, defense remains a direction where the market is willing to keep some exposure.

If you remember only one sentence from today: the yen issue has not yet evolved into a Treasury shock, so large caps remain stable; but high rates, credit stress, and AI cash burn are pushing the cost of capital back onto center stage. ETF Radar’s picture is: large caps mildly stable, long bonds weak, small caps under pressure, and defense relatively strong. A final reminder: the above are system quantitative-model direction forecasts for the general market and do not constitute personal investment advice; consult a licensed investment advisor before investing.