Bank of America strategist Michael Hartnett, who previously caught the tech selloff with precision, says his sell signal indicator has now been triggered off. Still, he makes clear that the end of a sell signal is not the same as a buy signal, and whether investors return to tech stocks in mid-November will hinge on the key variable of semiconductor price trends.
In his latest Flow Show report, Hartnett noted that the BofA Bull & Bear Indicator plunged from 8.1 to 7.4, falling below the 8.0 threshold and formally ending the sell signal triggered on May 26. Tactically, however, he remains defensively positioned and laid out two preconditions for returning to risk assets: a Fed rate hike on October 28 to restore credibility, and an easing of AI financing concerns. Neither has yet materialized.
Market expectations for a Fed rate hike in October have collapsed from more than 50% to just 14%. The AI credit market has just endured its worst week of the year: CoreWeave bonds have fallen into distressed territory, with its five-year CDS spread implying a roughly 63% probability of default. The Nasdaq 100 fell nearly 6% in a single week, its worst performance since "Liberation Day." Hartnett warned that if semiconductor prices decline, it would signal further tightening in AI credit and a peak in capital expenditure growth, at which point 2027 market returns would shift from capital appreciation to income-driven.
Sell signal ends, but buy signal has yet to trigger
The BofA Bull & Bear Indicator fell sharply by 0.7 percentage points this week to 7.4. Factors driving the decline included deteriorating breadth in global equity indices, hedge funds holding short positions in crude oil and S&P 500 futures, reduced long positions in gold futures, and weakening flows into high-yield and emerging market debt. The only factor pushing the indicator higher was continued net inflows into tech stocks.
Hartnett also noted that the "old version" Bull & Bear Indicator fell from 5.6 to 5.1, below its recent high of 8.0 on September 1. Both versions send the same message: the extreme optimism of the summer has come to an end.
Yet cooling sentiment does not mean money is truly leaving. BofA's breadth rule shows that a net 52% of global equity indices are in "oversold" territory (below their 50-day and 200-day moving averages), the highest since April 2025. But triggering a contrarian buy signal requires an oversold reading of a net 88%. Hartnett believes this threshold will be hard to reach in the near term unless the still "overbought" markets of the U.S., Taiwan, and Brazil fall another 6% to 10%.
Positioning remains highly crowded, tech overweight hits five-year high
While sentiment is cooling, risk exposure at the allocation level remains highly concentrated. The latest weekly EPFR data (through Wednesday) showed $64.4 billion in net inflows to equity funds, $11.8 billion into bonds, and $53.1 billion in outflows from cash—the tech selloff has left no obvious trace in flow data.
Equities now account for a record 66.4% of assets held by BofA's private clients, while cash has fallen to a historic low of 9.3%. For the full year, global equity fund net inflows are running at an annualized $1.3 trillion in 2026, with tech funds annualizing at $196 billion—more than 2.5 times the previous record. BofA's October fund manager survey shows investors' overweight in tech stocks is the highest since August 2020.
The Korean market presents the most striking divergence between positioning and price: the KOSPI has fallen 30% from its June high, yet after the August peak, only $7 flowed out for every $100 that flowed in. Hartnett noted that foreign investors were the main sellers this week, and with the buyback window nearing closure, the real withdrawal may not have begun yet.
AI financing cracks widen, real yields approach warning line
Hartnett's current defensive stance rests on three core pillars. First, the U.S. 10-year real interest rate is approaching 3%, a level historically associated with risk events such as the dot-com bust, the global financial crisis, and the SVB episode. Second, both the private equity ETF (PSP) and the business development company ETF (BIZD) have fallen back to first-half 2026 lows. Third, Oracle's five-year CDS has surged almost vertically, while overall investment-grade tech bond spreads have barely moved, indicating that the most highly leveraged borrowers are bearing the greatest pressure.
According to Goldman Sachs rates strategist David Curtin, the cost of AI infrastructure financing is one of the primary sources of market stress—for sub-investment-grade borrowers, a 6% to 8% all-in funding cost "bites faster" in a high-rate environment. Goldman data shows CoreWeave's 2031 bonds widened to more than 1,100 basis points over Treasuries at one point.
Consensus data cited by Hartnett shows hyperscaler capex expectations of $807 billion in 2026 and $1.16 trillion in 2027, while free cash flow is projected to turn negative at $63 billion this year and widen to negative $163 billion next year. Goldman data also shows that the semiconductor producer price index, after nearly three decades of decline, has recently spiked sharply.
Chip prices become the key variable for market direction in 2027
Hartnett offers his most critical forward-looking judgment of this cycle: semiconductor price trends will dominate the macroeconomic and market landscape into 2027. If chip prices sustain their rise, it would mean the U.S.-China AI arms race continues, expectations of implicit U.S. government support for bond and equity prices are maintained, investors return to tech stocks in mid-November, and the "short tech, long bonds" positioning reverses as 2027 begins.
Conversely, if semiconductor prices decline, it would mean further tightening in AI credit, a peak in AI capex growth, and the underlying logic supporting the current nominal GDP boom begins to unravel—U.S. nominal GDP has grown 63% from its COVID low to $33 trillion, with soaring chip prices a major driver. In that scenario, bonds would likely outperform, while any asset priced on the AI capex narrative would face clear pressure, and the core driver of 2027 asset returns would switch from capital appreciation to income.
Tactical defense, positioning in "peak yield" beneficiaries
On a practical level, Hartnett has chosen to increase allocations to assets that benefit from "peak yields," including small-caps, REITs, and the consumer discretionary sector, which BofA's October fund manager survey shows is the most underweight since July 2008. He explicitly stated he is not adding to tech stocks for now.
On longer-term asset allocation, Hartnett maintains overweight positions in gold, commodities, and emerging markets, while advising that the long-term "short bonds" trade and underweight China positioning should both be gradually reduced—a notable shift in his stance.
Hartnett also pointed to "price signals" showing that 355 S&P 500 constituents are more than 10% below their 52-week highs, 188 are more than 20% below, and 93 are more than 30% below. Beneath the seemingly calm index surface, damage at the individual stock level is already considerable, potentially creating value for investors patiently looking for entry points.
Comments