By Michael Santoli. Key takeaway: For investors worried about an AI rally collapse (a group that includes my "mystery broker" source), the bond market currently offers its best prices in years and can provide an asset buffer. Included below is an AI-generated model portfolio, drawing on past content from my TV appearances and columns.

Why Gen X seems most enthusiastic about AI's impact

Yogi Berra once said something that seemed illogical yet was quite brilliant: "It gets late early out there." That line originally described the late-summer shadows in left field at the old Yankee Stadium, and it applies just as well to assessing today's market.

The market keeps insisting the AI theme is "still in its early stages," and the public broadly believes economic momentum will remain strong, yet at the same time some late-cycle characteristics are already showing up. S&P 500 second-quarter operating profit growth exceeded 30%, and the market expects the third quarter to reach the same level; yet despite these strong results, the index sits only 2% above its June high. The S&P forward price-to-earnings ratio has compressed from 23 times a year ago to 19 times now. That signal can be read two ways: on one hand, investors are unwilling to pay a big premium for record profit margins and for growth that will slow next year; on the other hand, rising Treasury yields affect companies' overall cost of capital. The unemployment rate remains low, Americans spend what they earn, and consumer spending resilience is still decent.

However, real income growth has already begun to lag inflation, and households need to draw on savings to support spending. In recent weeks, corporate bond spreads have started to widen from historically extremely tight levels. A wave of tech debt issuance combined with Treasury volatility is testing investors' risk appetite. Of course, Federal Reserve policy has also shifted: from last year's precautionary rate cuts to rate hikes. The market expects several more hikes ahead. A strong economy can still absorb rate increases, but at the very least it signals that this economic expansion, already more than five years old, may be restrained.

Of course, every one of these signals can be explained as a mid-cycle adjustment, and that may indeed be the case. Markets are dynamic and do not run on a fixed clock, and "getting late" does not equal "about to end." When the government fiscal deficit reaches 6% of GDP while the private sector pours trillions of dollars into building new productive assets (large-scale capital investment of this kind has historically appeared mainly during government emergency stimulus periods), it is hard for the economy to fall into recession quickly. Shrinking valuations among AI leaders show, on one hand, that investors already recognize growth will eventually peak, and on the other hand, this also refutes the criticism that "AI stocks are in a bubble." Micron's stock price remains 15% below its June high; Nvidia's valuation carries a significant discount relative to the broader market, as well as to defensive, low-capital-expenditure Apple.

All else equal, this reflects that the market remains rational and that some risk has already been priced in. Market performance has become extremely divergent, making it hard to extract a clear macro signal. For several consecutive weeks, a large amount of historically rare data has appeared: on days when the S&P 500 closed higher, or was very close to a record high, a large number of individual stocks hit 52-week lows. Treasury yields kept setting new highs, further intensifying market disorder. Cyclical sectors came under pressure, and capital flowed into tech giants that are insensitive to interest rates and tied to the AI main line; the result is a rising index alongside many stocks hitting new lows.

At this stage of the market, index gains rarely come with strong broad participation, and declines are not entirely without resistance either. The S&P 500 hit a record high of 7,620 points on June 2, then moved sideways in a narrow range for two months and two days, until August 4, when it broke slightly above and set a new high. Two months and two days after that breakout, the index is again oscillating in a narrow range, and today's open is only about 0.5% away from another record high. But since August 4, when it first climbed above 7,700 points, the median share price of S&P 500 constituents, the Russell 2000 small-cap index, the bank index, and the equal-weighted consumer discretionary sector have fallen between 4% and 11%. Historically, this kind of turbulent internal action has sometimes released pressure far beyond what the headline index reflects, with late 2018 as a typical example.

That year was a midterm election year, and tightening financial conditions combined with weakening market internals meant that even with 20% earnings growth and no recession yet, the S&P 500 still fell nearly 20% by year-end. Of course, that was only a brief but severe market stress test. Valuations then expanded again, and the index climbed to new highs all the way until the COVID-19 pandemic struck. Dean Curnutt, founder and CEO of Macro Risk Advisors, laid out the dilemma facing risk-sensitive investors: "Treasuries, long called the 'risk-free asset,' now act more like a risk amplifier than a stabilizer in portfolios... Rising stock-bond correlation pushes up portfolio volatility; but falling correlation among individual stocks pushes volatility down, and pushes it very low. Over the past six months, the realized correlation of S&P constituents was only 4%. That incredibly low level is like giving index volatility a dose of 'Ozempic.'"

Of course, many people tolerate Ozempic, and their health improves as a result. The current market structure is especially friendly to passive index investors who do not screen holdings. The S&P 500 keeps adding weight to the strongest long-term growth theme and marginalizing sectors that are hit harder and cyclically fragile, thereby letting holders avoid deeper drawdowns. But one day, the AI tech sector's weight in the index will become excessively large. At that point, the S&P's powerful structural protection mechanism will instead expose investors to a crash cycle and large-scale capital losses. Facing this outlook, which is hard to time but carries risks that should not be underestimated, the bond market now offers rare multi-year value and can serve as an investment cushion.

Market temperature indicator

This indicator was compiled by John Kolovos and combines multiple data points to reflect both what investors say and how they actually trade. Kolovos interpreted the latest reading: "Affected by tightening financial conditions and rising cross-asset implied volatility, bullish sentiment cooled last week. Cyclical stocks are already oversold; the number of stocks hitting 52-week lows and 4-week lows, as well as the proportion of stocks whose relative strength index RSI has entered oversold territory, are typically seen only after the broader market has fallen 8%–10%. But the S&P 500 has pulled back only about 3% from high to low. Historically, this kind of oversold combination often points to better returns ahead."

Insider view

David Snyder, founder of Journey One Advisors and the source I have long cited as the "mystery broker," said bluntly that this is a late-cycle environment. Since last year, he has believed the long bull market has entered its final stage, a stage typically accompanied by a last sprint in tech stock momentum. In a LinkedIn post, he noted that high-beta stocks continue to outperform the broader market by a wide margin, a phenomenon that violates decades of financial theory and will eventually see a sharp mean reversion. As a journalist, I cannot trade individual stocks. And I am someone who adopts new things slowly and rarely uses AI tools.

So I asked a friend deeply involved in AI and investing to use AI to generate a long-short model portfolio, referencing all my TV remarks and column views. The final result was considerably more conservative than I expected, with a pragmatic, anti-hype positioning: long Microsoft, Chevron, and Coca-Cola; short Advanced Micro Devices, Micron, and Robinhood. I hope this virtual AI risk control has set strict stop-losses for those short positions. By stereotype, Gen X never readily embraces the "next big thing" sweeping society, but LinkedIn research shows Gen X professionals are the most enthusiastic about AI's impact, while the younger Gen Z is more cautious.

Perhaps Gen X has already established a firm career footing and does not need to fight for a place in a job market with scarce entry-level roles; or perhaps, in the eyes of this cohort now in their late 40s to early 60s, AI is just another round of technological change. One respondent put it this way: "Over the past thirty years, we have witnessed every major workplace transformation... This is just a new chapter." Let us hope so.

Closing thoughts

Historically, shorting at this time of year has often meant going against the trend. The chart below shows this year's Dow Jones Industrial Average trend, overlaid with Ned Davis Research's 2026 cycle composite chart, which combines annual seasonality, the four-year election cycle, and the decennial cycle pattern ending in 6. If you look only at historical statistics, now is the point when a tailwind period begins. But the disclaimers are crucial: first, note that the vertical axis scales of the two charts are different, so the composite chart is mainly about the shape of the intra-year trend, not the magnitude of gains or losses. More importantly, this year began with a huge divergence: the Iran conflict and the oil price surge disrupted the first-quarter rally that should have occurred. Understanding seasonality is useful, but as I often say, these historical patterns are the climate, not the short-term weather.