Stock Market Week in Review: Bonds Break 5%, Agentic AI Arrives, Banks Tested

The Nasdaq 100 made a new high, volatility stayed subdued and Bitcoin recovered above $85,000. Yet one of the biggest conversations inside the 4 Pillar Trading Community this week was not about stocks at records. It was about the bond market.

Treasury yields were approaching levels not seen since 2007 while equities looked remarkably unconcerned. By the time of writing, on 28 September, the ten-year had moved through 5% to around 5.2%. That gap became the starting point for a week of discussion about interest rates, growth, bank profitability and the risks facing heavily indebted companies.

Through the week, mentors Edward Shek (Ed, to the room), Anthony Iser and others worked through the macro backdrop, while members shared research, challenged each other’s assumptions and talked through positions. Alongside that, the arrival of consumer agentic AI, questions about Oracle’s capital spending and early signs of stress in lower-quality credit gave the room plenty of chances to apply its macro-first framework.

One disclosure up front: this is an independent affiliate view from The 4 Pillar Trader, not the official 4 Pillar Trading account. It is drawn from the community’s own weekly review. Market figures, product news and event dates are as reported there and have not been independently checked against primary sources. Mentor and member views are theirs, described, not endorsed, and nothing here is buy or sell advice.

1. The bond market: why were equities ignoring rising yields?

The central discussion of the week was the sharp rise in government bond yields and the muted reaction from equities.

The community’s market review put the two-year Treasury yield at 4.76%, with a weak $70 billion five-year auction clearing at 5.033%. The ten-year approached 5%, close to levels last seen in 2007. Japan’s bond market was under pressure too, with the ten-year Japanese government bond yield at its highest since 1996.

Against that, Federal Reserve commentary stayed hawkish, the US dollar strengthened and the average 30-year mortgage rate reached 7.03%. The equity market shrugged.

Edward Shek: the market looks remarkably relaxed

Ed Shek described the equity reaction as “highly indifferent” and observed that investors looked “super relaxed, for now”. That last phrase captured a theme that ran through the week: the difference between what markets are doing today and the risks that could catch up with them tomorrow.

Jason McDonald shared a Bank of America strategist’s view that ten-year yields could reach 7% before seriously disrupting equities. He questioned it, suggesting stocks could run into real difficulty well before that level.

A member then brought in the counterargument. The Atlanta Fed’s running estimate for third-quarter real GDP growth stood at 5.1%. Add inflation, and nominal growth was tracking above 8%. That prompted a discussion of whether borrowing costs near 5% should be considered as threatening as they were in earlier cycles. In 2007, yields were similarly elevated, but nominal growth was considerably weaker.

The lesson the room drew was that rates cannot be assessed in isolation. Their bite depends on growth, inflation, credit conditions and the stage of the cycle. For traders, it reinforced the first principle of the 4 Pillar Trading approach: understand the macro environment before forming a view on individual stocks.

2. Bank earnings: why higher rates do not always help financials

With third-quarter bank earnings approaching, financials became the second big topic.

A member presented a sector analysis suggesting that large banks and regional lenders had already priced in a lot of optimism. The implication was that strong quarterly numbers might not be enough to drive further gains. Guidance would matter more.

The discussion drew out differing expectations across the majors. JPMorgan had pointed to mid-to-high-teens growth in investment banking and markets revenue, Bank of America expected investment banking fees to decline, and Goldman Sachs had flagged softer fixed-income, currency and commodities activity.

Anthony Iser on the yield-curve problem

Jason McDonald highlighted something that seemed to cut against the textbook relationship between rates and bank profits. Higher rates can widen lending margins, yet bank stocks were not getting the expected benefit.

One explanation was the flattening yield curve. The room noted that the gap between ten-year and two-year Treasury yields had narrowed from roughly 0.70 percentage points in February to about 0.25.

Anthony Iser took a follow-up question on bear flattening and bull steepening and walked through the mechanics. When rates rise, banks can sometimes lift lending rates faster than the rates they pay on deposits, which improves net interest margin. But the eventual effect depends on the shape of the curve, the economic cycle and the financial health of borrowers. A flatter curve squeezes lending profitability, and weaker consumer finances raise credit losses.

The takeaway was that “higher rates are good for banks” is not a thesis. It is a starting assumption that has to survive the curve, deposit costs and credit quality.

3. Agentic AI: Edward Shek on the next stage of artificial intelligence

AI stayed a major topic, but the conversation moved beyond semiconductor demand and data-centre construction.

Ed Shek flagged the arrival of consumer agentic AI as an important development following the launch of Meta’s Muse, as reported in the room. For much of the previous 18 months, investors had focused on enterprise agents and workflow automation. Consumer adoption, the discussion suggested, brings a different set of opportunities and competitive threats.

From answering questions to completing tasks

Jason McDonald shared examples of how consumers had reportedly used Muse in its first two weeks: booking haircuts, contacting customer support, cancelling insurance policies, ordering groceries and chasing airline compensation.

The distinction between a conventional assistant and an agent is that the agent carries out multistep tasks on the user’s behalf. For businesses, that matters. An agent that can compare prices, switch providers and complete purchases weakens any business model that depends on customers staying put. Amazon’s reported decision to block Muse from its platform became a talking point in its own right.

Which companies could benefit?

Ed shared a watchlist of companies with potential exposure to the growth of AI agents: infrastructure, enterprise software and cybersecurity names rather than the consumer agents themselves.

Ticker	Company
PLTR	Palantir
DDOG	Datadog
SNOW	Snowflake
NOW	ServiceNow
CRM	Salesforce
TEAM	Atlassian
DOCU	DocuSign
PANW	Palo Alto Networks
OKTA	Okta
TWLO	Twilio

A research watchlist as discussed in the community, not a recommendation to buy anything.

Ed’s broader point was that investors should look at companies positioned to benefit from agentic infrastructure regardless of which consumer-facing agent ends up winning. The discussion also considered potential disruption in travel booking, tax preparation, outsourced gig work and banking. If agents make it far easier to compare providers and switch accounts, companies that rely on customer inertia face more pressure.

For traders that produced two separate research questions. Which businesses benefit from wider AI adoption? And which established companies could see their existing model squeezed?

4. Oracle: is the AI investment cycle becoming too expensive?

The Oracle discussion looked at a different side of the AI boom: capital efficiency.

A member laid out a structured bearish thesis on the relationship between Oracle’s AI infrastructure ambitions and its growing financial commitments. The analysis accepted that demand was strong but raised concerns about rising capital expenditure, debt, lease obligations and a declining return on invested capital.

The same member tracked Oracle’s credit spread against the 30-year Treasury, reporting a widening from about 245 basis points in June to 283 by 22 September, with several Oracle bonds reportedly yielding around 8%. Those figures fed a wider question: will the returns from AI infrastructure justify the cost of building it?

Edward Shek flags a potential warning sign

Later in the week, a reported force majeure notice from Oracle concerning its Project Jupiter data centre in New Mexico drew attention. Bloom Energy, expected to supply power to the project, fell more than 5% in pre-market trading on the news, according to the review.

Ed Shek highlighted it as the kind of development that may look more significant in hindsight. The concern was not that demand for AI compute was disappearing. It was that the infrastructure needed to meet that demand was getting more expensive and harder to deliver.

Anthony Iser widened the point by separating revenue growth from investment returns. AI infrastructure companies may generate substantial revenue, but that does not mean the capital they are deploying will earn an attractive return. The open question was whether revenue would grow fast enough to justify the commitments.

Position sizing matters when trading a bearish thesis

The Oracle thread also produced a useful exchange on risk management.

The member running the bearish options position explained that the initial trade was about a third of the intended full size, with a plan to add if further negative developments confirmed the thesis. Another member questioned an estimated risk-reward of more than five to one, pointing to the difficulty of trading a company this widely followed.

The broader lesson was that an attractive theoretical risk-reward does not imply a high probability of success. Sizing has to allow for a thesis that takes longer than expected, or never plays out.

5. Credit markets and the K-shaped consumer

Another important contribution from Edward Shek looked at lower-quality credit and the uneven effect of restrictive monetary policy.

While broad credit markets looked calm, pressure was becoming visible among heavily indebted borrowers. Ed pointed to CCC-rated companies, particularly those that borrowed in 2021 and 2022 and are now approaching refinancing. The review put CCC yields at around 15%. For a business carrying a lot of debt, refinancing at those rates is a serious problem.

The discussion identified private-equity-backed technology, healthcare and consumer discretionary businesses as areas worth monitoring.

Why monetary policy hits borrowers differently

Companies with strong balance sheets that locked in long-term fixed-rate financing during the low-rate years may stay relatively insulated. Businesses with weaker balance sheets, floating-rate debt or a near-term maturity wall face a very different environment.

That unevenness connects to the K-shaped consumer theme the community returns to often. Higher-income households and financially resilient companies keep spending and investing, while lower-income consumers and highly leveraged businesses feel increasing pressure.

One member reported closing a put structure on OneMain Financial, a nonprime consumer lender, that had been a way of expressing a view on consumer credit deterioration. Anil Tewari’s watchlist for the following week included CarMax, Ally Financial and KeyCorp as further reads on consumer and bank credit conditions.

6. Other market themes discussed by the community

China’s technology sector. Anil Tewari was preparing a webinar on Alibaba and Tencent following Alibaba’s reported unveiling of its V900 AI chip. Jason McDonald disclosed long exposure to BABA and KWEB, while a member suggested KSTR as a more direct way to play China’s technology infrastructure build-out.

Pharmaceuticals and biotechnology. The room discussed the increasingly negative correlation between large pharmaceutical companies and AI-related semiconductor stocks. Raj Malhotra repeated his view that Eli Lilly had “basically become a hyperscaler”. Mentors also discussed using XBI rather than single biotech names when a trader lacks detailed company knowledge.

Tokenisation. The community highlighted the SEC’s reported 17 September innovation exemption for tokenised stocks, which brought SECZ and FIGR onto watchlists. The interest reflects the continued overlap between traditional markets and blockchain infrastructure.

Uranium. Jason McDonald explained the recent cooling in uranium miners by pointing to utility inventories. With utilities reportedly holding around three years of supply, there is little immediate pressure to sign new contracts.

Copper. Copper’s reported move to an all-time high kept the metal on the radar, with members continuing to track the theme against industrial demand.

7. Trading lessons: what members took from the mentors

Beyond the individual themes, some of the most valuable conversations were about process.

Anthony Iser on catalysts

A member asked whether a trade needs a catalyst before entry. Anthony explained that an ideal catalyst introduces new information that helps set a trade’s direction, timing and potential size, while acknowledging that genuinely identifiable catalysts are hard to find. Jason McDonald described catalysts as “icing on the cake” and said he may still enter a well-researched trade without one.

The distinction matters because a catalyst can improve timing, but the thesis has to stand on its own.

Connecting macro research with individual stocks

Anthony also addressed how the community’s top-down and bottom-up work fit together. Macro analysis identifies sectors and companies that may benefit from prevailing conditions. Company research then decides whether a particular stock is an attractive trade.

For the community’s preferred timeframe, he stressed that both need to line up. A favourable macro backdrop is not enough if the company’s valuation, earnings outlook or positioning contradicts the trade.

Reading between the lines in earnings season

One member described closing a short in General Mills after listening to the earnings call. The headline numbers looked weak, but management’s commentary suggested a more constructive outlook.

The example made the point that guidance, explanations and management’s confidence can change the reading of an otherwise disappointing report.

AI tools for research

Members also compared research workflows: a self-updating macro scorecard covering 19 indicators, AI-assisted sector dashboards, and data platforms such as FRED, Financial Modeling Prep, EODHD and Koyfin. Anthony released an updated 4 Pillar Trading market screener for members during the week.

8. What traders were watching: 28 September to 2 October

When	Event	Why the room cared
Tue 29 Sep, before the open	CarMax (KMX) earnings	Consumer credit conditions, vehicle affordability and financing trends
Wed 30 Sep, late US time	Bank of Japan Tankan survey	A key read for anyone watching Japan, the yen and global rates
Thu 1 Oct, before the open	Accenture (ACN) earnings	Enterprise technology spending, AI adoption and demand for IT services
All week	Ten-year Treasury yield	Whether it breaks above 5%, and whether equities keep ignoring it

Dates and timing are as given in the community review. Check them against an official calendar before trading around them.

9. Key takeaways from the week

• Macro context matters. Elevated bond yields mean different things depending on growth, inflation and credit conditions.
• Bank profitability is more than the rate level. The curve, deposit costs and borrower health all feed earnings.
• Agentic AI could reshape established industries. The opportunity extends well beyond the companies building consumer agents.
• AI capital efficiency deserves attention. Oracle’s debt, spending commitments and credit spreads raised real questions about the return on infrastructure investment.
• Credit stress can stay concentrated. Pressure on weaker borrowers may not show up in broad equity indices for a while.
• Process matters as much as ideas. Catalysts, sizing, earnings interpretation and the alignment of macro and company research ran through every thread.

10. Final perspective: looking beyond the index

The theme of the week was the contrast between calm equity indices and significant developments elsewhere in the financial system. Treasury yields were approaching historically important levels. Lower-quality borrowers faced increasingly expensive refinancing. Banks were heading into an earnings season where guidance could matter more than the print. And the rapid development of agentic AI was creating new questions about infrastructure spending, competitive advantage and disruption.

Inside the community, Edward Shek, Anthony Iser, Raj Malhotra and the other mentors and members approached those developments from different angles. The value was not in guessing which stock moves next. It was in examining the mechanisms behind market moves, challenging assumptions and connecting macro developments to individual trade ideas.

If that is the kind of room you want to learn in, the full review covers how the community works day to day, and the pricing page has the current membership options.

Frequently asked questions

Why were stocks calm despite rising bond yields?

The community discussed whether strong nominal growth was helping equities tolerate higher borrowing costs. Ed Shek highlighted the market’s apparent indifference, while Jason McDonald questioned how much further yields could rise before stocks felt real pressure.

Are higher interest rates good for banks?

Higher lending rates can improve profitability, but the outcome depends on funding costs, the yield curve and credit conditions. Anthony Iser explained why the wider environment matters when judging the link between rates and bank earnings.

What did Edward Shek say about agentic AI?

Edward Shek highlighted consumer agentic AI as an emerging theme following the reported Muse launch. He pointed to potential beneficiaries in enterprise software, infrastructure and cybersecurity, and to the risk for businesses that rely on customer inertia.

Why was the community discussing Oracle?

A member presented a bearish thesis focused on Oracle’s rising capital expenditure, debt, credit spreads and declining capital efficiency. Ed Shek and Anthony Iser contributed to the wider discussion about the cost and likely returns of AI infrastructure investment.

How does 4 Pillar Trading approach finding trades?

By combining macro research with company analysis. Anthony Iser explained that macro work identifies potential tailwinds, and stock-level analysis determines whether there is a viable trade.

This article is for educational purposes only and is not financial advice. Views attributed to mentors and members reflect discussions described in the community’s weekly review and may not represent their current positions. Trading involves risk, including loss of capital. Do your own research and speak to a qualified professional before making investment decisions. The 4 Pillar Trader is an independent affiliate and may earn a commission if you join through links on this site.