According to a report, JPMorgan's global markets strategy team presented a set of consistent evidence in its October 28 "Flows & Liquidity" report: over the past two weeks, retail capital has accelerated its inflow into equity funds, with leveraged equity ETFs recording the highest monthly capital inflow on record, and it is precisely this money that is supporting the bottoming of AI-related stocks.
The report, led by Nikolaos Panigirtzoglou, compressed its conclusion into a single sentence in its opening highlights: retail investors are pouring record funds into leveraged ETFs, supporting the bottoming of AI-related stocks; this retail impulse still has room to strengthen through year-end, especially through options; QQQ's short interest ratio has risen to its highest level since 2023, meaning institutional investors hold a short base waiting to be covered; tech companies' capital expenditures have risen sharply this year, but buyback resilience has exceeded expectations, providing additional background support for the stock market.
Leveraged ETFs: Space Created by the Drop, Then Filled In
Charts show that global equity fund flows have clearly accelerated over the past two weeks, led by leveraged equity ETFs, with this month's capital inflow into this category of products being "the strongest on record." The key lies in what happened before. The report notes that in prior weeks, the assets under management (AUM) of leveraged equity ETFs had seen further significant declines, with the AUM of leveraged storage stock ETFs falling relative to their underlying storage stock market capitalization to the lowest level since late May, erasing about two-thirds of the gains from the April-May rally. Low AUM levels precisely leave room for further short-term inflows — this is the direct basis for JPMorgan's judgment that "it can continue." In other words, this is not a chase without a pullback, but a re-accumulation after deleveraging.
Options: Retail's Next Wave May Come from Here
JPMorgan believes retail buying impulse still has room to strengthen through year-end, mainly through options. The chart below tracks the net amount of "open buy call options minus open sell" for small clients holding fewer than 10 contracts on single-stock options: this indicator has not reappeared since it collapsed during the deleveraging phase in June-July last year. That is, retail investors have already moved first in stocks and ETFs, but accumulation on the options side has not yet begun.
Institutions: Positions Already Reduced, Short Base Already Built
Institutions also have room. JPMorgan's broad equity positioning indicator, distilled from cross-asset positioning monitoring, shows that compared with the recent peak in mid-September, current positioning has fallen significantly — meaning speculative funds such as hedge funds also have room to "continue adding to tech stocks from here." A more intuitive corroboration is QQQ's short interest ratio: the chart below shows that QQQ's short interest as a percentage of its free float has risen to the highest level since 2023 (data as of October 27). Crowded shorts are themselves fuel — once tech stocks stabilize, covering demand will become one source of buying.
Buybacks: $1.55 Trillion, and "Unexpectedly Stable"
The report devoted considerable space to buybacks, because this is its core argument for judging that "background support remains." Despite tech companies significantly increasing capital expenditures this year, the tech sector's buybacks year-to-date have shown unexpected resilience; global buyback announcement amounts year-to-date have reached $1.55 trillion, which annualized at the monthly average pace would be about $1.7 trillion for the full year, higher than last year. Why can it be maintained? JPMorgan's explanation is financing conditions: the tech sector as a whole has been in a $40 billion financing surplus this year, and financing channels remain open. According to its credit team's observations, high-grade tech companies can currently access both public and private debt markets simultaneously, including unsecured debt, bank loans, data center project financing through SPVs, and GPU asset-backed debt. But the report also left a door open: the unsecured debt market is effectively closed to certain neoclouds (emerging cloud service providers) and highly leveraged tech companies. JPMorgan's stance is that such companies are neither key links in the AI ecosystem nor the main drivers of buybacks, so it is not yet fair to say that "financing constraints have already hit the core" — but if financing conditions tighten, this could change.
Momentum Funds: Treasury Shorts and Equity Longs Largely Closed, but Crowding Remains
JPMorgan also analyzed how trend-following funds (CTAs and other momentum investors) adjusted in recent weeks. JPMorgan used the z-score of momentum signals (adjusted by the ratio of 10-year to 6-month volatility). Bonds: momentum signals in 10-year U.S. Treasury, German bund, U.K. gilt, and Japanese government bond futures were at or near extremes at the start of the month, and then were mostly closed except for German bunds, with z-scores falling to the −0.3 to −0.4 range; German bunds remained deeper at about −0.7. What is truly crowded is 10-year OAT (French government bonds) versus German bunds, with a z-score of about −2.9, extreme by historical standards — momentum investors' shorts here are quite crowded, making profit-taking or mean-reversion signals easy to trigger, though the report notes this may first require a period of stabilization. Equities: the October semiconductor pullback brought equity momentum signals down from highs — the S&P 500 and Nikkei 225 fell from about +1.3 in early October to +0.9 and +0.4 respectively; MSCI Emerging Markets shifted from a small net long of about +0.5 to a small net short of −0.2. Within Asia, KOSPI fell sharply from early July and had briefly recovered to about +0.5 in late September, while MSCI Taiwan was about +1.3 in early October; after the semiconductor pullback, the two fell to 0 and +0.4 respectively; HSCEI was already slightly net short at the start of the month and is currently about −0.8. FX: the U.S. dollar's z-score against G10 currencies approached the extreme zone of +1.5 in early October and has now fallen back to about +0.8. Broken down, the dollar's long exposure is more driven by euro shorts — the euro's z-score once reached −2.1 in early October, still at −1.8 now, and most positions were built in September. This likewise implies risks of profit-taking or mean reversion. Energy: the z-scores of crude oil and TTF natural gas briefly fell after extremes in mid-September, but as concerns over renewed escalation of Middle East conflict resurfaced, both returned to more extreme positions — crude oil futures at +1.8 and TTF futures at +2.6; refined product futures remained at a high level of around +1.6. In one sentence, this section says: momentum funds have largely closed Treasury futures shorts and equity futures longs, but crowded positions remain in OAT versus German bunds, the euro, crude oil, and TTF.
Two Side Threads: Brazil and "Bitcoin Beats Gold"
After Brazil's second-round presidential election, retail inflows saw only a relatively modest reversal, while institutions reduced positions more quickly. CME's Ibovespa futures positioning proxy moved down further sharply after the second round; CME's Brazilian real futures positioning had been rising since "Liberation Day" in April 2025, and the post-second-round reduction largely erased the increase after the first round. On the retail side, looking at the EWZ (iShares MSCI Brazil) ETF: shares increased significantly after the first round and saw partial outflows after the second round, but price adjustments show portfolio weights have clearly been lowered. On the institutional side, looking at EWZ short interest ratio: it rose in mid-September and stayed high until around October 15, and rose slightly this week, consistent with the direction of unwinding net long equity futures. Bitcoin vs. gold: JPMorgan had previously argued that IBIT's higher short interest ratio relative to GLD constituted support for "bitcoin beats gold"; now IBIT shorts have fallen and GLD shorts have risen, and the gap between the two has almost disappeared, weakening the support from this single indicator. But other indicators still lean toward bitcoin: in options, hedging demand for bitcoin is more persistent and rising, while hedging demand for gold is declining; in futures, gold positioning has slipped slightly while bitcoin positioning remains high, and momentum traders are still adding gold shorts while maintaining relatively stable long exposure to bitcoin. There are also two changes unfavorable to gold — gold's return correlation with the S&P 500 has recently been higher than bitcoin's, meaning it behaves more like a risk asset and its diversification value has declined; bitcoin's 6-month volatility ratio versus gold fell to a record low of 1.36x in mid-September, and although it has rebounded to 1.5x, it remains in a downtrend and far below the historical average, meaning bitcoin's previous volatility disadvantage is narrowing.
Comments