Hedge Funds Load Semis; Software Only Covers
US Stocks · Options · News · Views
Semis and Memory Hold Up in a Broad Selloff
Monday’s tape was a broad selloff. SPX had about 367 declining stocks versus 133 advancing, and 9 of 11 sectors finished red. Volume picked up a little versus last week, but the important levels barely moved. The story is positioning and resistance, not one down day.
Semiconductors still showed relative strength. SOXX gained about 1.64%, with 19 components up and 11 down. Commerce Secretary Lutnick said the Trump administration opposes Apple using China-made memory chips, and that helped lift storage names. MU and SNDK completed stop-decline breakouts, but they still face pressure before new highs. A stop-decline is stabilization, not permission to chase.
Middle East headlines were messy: the 60-day U.S.-Iran deadline was said to have been extended; an Iranian official said policy is shifting from defense to full offense; Trump claimed a secret channel to the Islamic Revolutionary Guard Corps, which the Guard denied. Treasuries weakened and crude rose about 3.33%. This week is light on economic data and heavier on retail earnings. Q2 reporting is almost done; NVDA next week is the print that lets the market reassess S&P forward earnings.
Hedge-Fund Cash Comes Back Into Stocks
Goldman Sachs hedge-fund data through last weekend show net buying of U.S. stocks every session last week, the second-fastest pace in the past year. The bulk was single-stock long buying. Macro short covering was secondary. ETF shorts have now been covered for a sixth straight week. It looks like “buying America.” Split it apart and it is two trades: adding cash equities, and covering old ETF shorts.
Single stocks were about 70% of net buying, 2.1 standard deviations above the one-year average. Long buying versus short covering ran about 7.6 to 1. Covering did not do the heavy lifting; fresh longs did. Eight of 11 sectors were net bought. By dollar amount the biggest buys were information technology, communication services, health care, financials, and staples. The biggest sells were real estate, energy, and utilities.
Index and ETF products were about 30% of net buying, 1.0 standard deviation above the one-year average, with short covering versus long buying about 1.4 to 1. U.S.-listed ETF short interest fell for a sixth week, down about 3% and about 12% month over month, led by covering in small-cap, tech, and large-cap ETFs. Some of that was offset by adding shorts in real estate, utilities, and energy ETFs. Macro covering is not blanket risk-on; defensive ETF shorts are still being added.
IT was the largest net-buy sector, 1.2 standard deviations above the past year. Software buying was mostly short covering. Communications equipment and semiconductors plus equipment were long buying, and the semi sleeve was the largest subsector by dollars. Electronic equipment and instruments were long-short turnover. IT services were the largest net-sell sleeve as existing longs cut. Inside tech, chips and equipment are being added while IT services are being cut.
REITs had been net bought in seven of the prior eight weeks, then flipped to the largest net-sell sector, about -0.4 standard deviations versus one year. Shorting outran buying. Specialized, retail, and industrial REITs were sold hardest. Health-care and diversified REITs were still net bought. With hike odds still elevated, the complex is unloved; the health-care sleeve is the part that still holds up.
Software Books Stay Neutral; Longs Stay Cautious
Hedge funds’ net software allocation is about 4.5% of U.S. market cap. It started 2026 near 7%, was shorted down to 1.3% at the lows, and has bounced. That is still far from prior-cycle highs near 18%. Covering from 1.3% to 4.5% only says funds are no longer one-way short. It does not say they have reloaded a heavy software long. Fresh longs remain cautious. The book looks neutral.
Semiconductors and equipment stay overweight. Net allocation is about 10.4%, versus 6.8% at the start of the year and a mid-year peak of 14.1%. That is about four points off the hottest print, but still well above the January start and still historically high. Large funds still like semi upside. They are cautious adding software and, at least for now, are not aggressively re-shorting it. One sleeve overweight and one sleeve neutral also matches the tape: SOXX and IGV keep taking turns. Neither has a persistent lead.
Microsoft Digests Gains; IGV Still Caps at 107
MSFT pulled back into 469-487 and has now repaired the daily RSI overbought plus bearish-divergence setup. Repair is not a green light for a melt-up tomorrow. The prior rally was too fast and left loose supply. Digestion can be sideways, a grind, or a slow 0.2% to 0.5% drift lower. If the book is light and the view is still constructive, this is where to add. If the book is already heavy, hold and wait. The level did not change. The indicator did: it needed a reset, and it got one.
META dropped more than 3%, failed 632, and was sold back to the February 3 launch point. Weekly is neutral. Monthly still looks constructive for a longer horizon once a base is done, because overhead indicator room is ample. Daily is weak and needs time. Among the mega-caps, Meta’s AI monetization is the laggard, so CapEx, the earnings bridge, and new revenue paths still need work. Historically it cuts fast — spend, teams, whole units — and the market eventually takes the reset. On a six-month clock, a weak daily tape is not a surrender.
GOOG and AMZN saw a little more volume, but it is still light versus two weeks ago, and both remain rangebound. AMZN sold off on volume and is still holding 260. Lose 260 and it goes back to the original breakout level, which is a longer-term entry, not a thesis break. The post-breakout high is already in; this is a normal round-two pullback. Light-volume dips here change the entry, not the larger direction.
IGV stalled at 107. Last Friday it opened 106.4 and printed 106.75. Close is not through. Front-running resistance is a two-sided bet: a clean break is a better entry, a fail is a local high. The failed break leaves the tape weak near term, but the daily RSI divergence can use this pullback to reset for the next attempt. Hedge-fund software exposure is not aggressive — mostly covering — so the next tell is whether they add longs or add shorts. That will not show up in one or two sessions.
SOXX gained about 1.58% and swapped leadership with IGV again. The index is also under resistance, grinding higher around 559-560, still well short of 580. Until one of them clears, it is a coin flip which group leads the next break. MU and SNDK have completed stop-decline breakouts, which helps the floor; it does not erase the two heavier resistance bands above. Hedge funds still like semi upside from here. That is the large-book view, not a one-day candle guarantee.
QQQ now splits the old 693-724 zone in two. Lose 705-724 and the tape goes weak, with more downside likely and VIX liable to lift, though not necessarily through 22 right away. A real downtrend still needs 693. Aggressive dip-buyers who remain constructive on Nasdaq can use 705-724 — and it is an aggressive buy zone. Defend 705 before talking about 724. Lose 705 and cut offense; 693 is the confirmation line for a downtrend.