Semis Lead Again, Hedge Funds Still Won't Chase
US Stocks · Options · News · Views
Weak tape, strong chips, risk did not get worse
The major indexes closed lower on September 8 in New York. The Dow led the decline. After a confirmed strike on Saudi energy facilities that the Houthis claimed, Brent jumped to about $95 and energy finished higher. The equity tape split: SOXX sucked in capital for a second session, with 21 components up and 9 down. SPX had 356 decliners against 146 advancers. Only three of eleven sectors finished higher — energy, utilities, and REITs. XLV and XLF lagged. Strong groups stayed strong. Weak groups had no bid. That is a high-vol rotation tape, not a broad rally. Oil lifted energy. Semis pulled compute out of tech. The rest of the tape funded it.
The line to keep is not "semis are strong." Downside risk did not deteriorate further, and the bid is not large enough for a full risk-on tape. The two spots that looked dangerous a few sessions ago — Russell and SOX — both held. IWM held 288. SOX held its key area. Shorts stacked in those two spots became squeeze fuel and gave semiconductors a two-day lift. SPX is sitting in the middle of the range near the prior high, far from strong support below and short of the incremental capital a new leg higher would need. Active managers and retail are already heavy. The hedge funds that still have dry powder did not put it to work. Money rotated across sectors, with a squeeze on top. Risk is contained. Sentiment is not fully risk-on. Treat this as a new systematic bid and the next rotation will punish it.
Inflation expectations held. Households kept levering up
In the New York Fed's August survey, one-year inflation expectations stayed at 3.6%, the three-year reading eased from 3.3% to 3.2%, and the five-year held at 3%. The center held. The details did not. Gasoline expectations jumped 1.7 percentage points to 4.6%. Food, medical care, and rent expectations also moved higher. The labor outlook worsened: the mean probability that unemployment will be higher in a year rose 1.6 points to 44.4%, the highest since April 2020. The chance of finding a job if laid off fell to 45.4%. The chance of missing a minimum debt payment in the next three months rose 1.2 points to 13.2%. Credit feels harder to get, and households feel worse about their finances. Personal job-loss probability actually eased. Do not read 44.4% as "I am about to be fired." It is a bet that the unemployment rate goes up. Consumers are preparing for a worse macro tape.
Consumer credit is still expanding. July total credit rose $18.1 billion, above June's $14.6 billion and above the roughly $11.7 billion consensus. Revolving credit added $2.8 billion, taking card balances to $1.357 trillion. Nonrevolving credit — auto and student loans — jumped $15.3 billion, the largest monthly increase in more than three years, and pushed total consumer credit to about $5.19 trillion. The average rate on card accounts assessed interest is back at 22.15%. Not an all-time high. Expensive enough. Inflation expectations have not loosened, yet households are still borrowing at that rate. Spending can still be financed. The quality of that spend is getting worse. Demand is still there because leverage is still there, not because wallets got thicker. If Friday's CPI prints hot, gasoline expectations have already moved, and the Fed has less room to look away.
SOXX is still under the downtrend
SOXX tagged 534, opened near 530, and closed at 528. The falling trendline is still around 530–532 and drifts toward 530 into the next session. Until it breaks, algos, systematic flows, and a slice of technical longs stay on the sideline. These lines persist because failed tests teach the tape to sell the bounce there. Many constituents look like they already cleared their own lines. The index has not. Until the index goes, sector mood is still one spark short. That line remains hard near-term resistance. Whether the next few sessions bring a new compute story, a print, or a volume break in the ETF itself decides if this stays a hotspot or becomes a trend.
AMD closed through 498 on heavy volume, around 506, and is a hotspot again after months of quiet. 498 is the first gate — longs are building, which is good for holders. 528 is the second. Until 528 goes, do not treat this as a free run at new highs. It can still get sold back and chop here. If SOX gets rejected, AMD will have a hard time clearing 528. If 528 goes, overhead is thin and speculative flows will arrive faster. Split it: 498 confirms the hotspot. 528 confirms a shot at new highs. Trade the hotspot short-term. Wait for the index before calling a trend.
NVDA is still chopping and burning patience. It printed a near-term high the prior session, then sold that high and tagged last Thursday's low. SOX is rising and higher-beta names are being chased. The mega-cap is being used as a funding source. Price has not broken down. The tape is just fast. Fundamental books can wait. Two sessions ago SOX and Russell still looked like trouble. Two sessions later semis are the darling. In this tape, staying power comes from levels and size, not from chasing today's brightest bar.
Software cooled. Health care and financials have pullback levels
IGV ran to 110, failed the mid-range resistance, and snapped back to about 102. Defend 101. Lose it and the upside structure breaks, sending the ETF back into the 90.1–107 range. It stops being the hotspot. That is still a range, not a new short regime. OpenAI's new model and the AGI compute story put the infrastructure narrative back on, and money rotated from software into semis. Software is not broken. Relative performance got pulled. September and October are high-vol months. Hot groups change quickly. Today's leader can fade at the open. If the software book is meant to hedge the semi book, 101 has to still be there.
XLV tagged 176, then chopped, and sold again. Mid resistance sits at 173–175. 2027 sector growth for health care still screens above the SPX average. When nothing is working it can grind. When the tape has downside risk and institutions cannot go to cash, it is a minimum-exposure parking spot. The price is still chewing below resistance. Wait for a real pullback before treating it as an offense-defense add. Do not turn a defensive sleeve into an offensive one under 173–175.
XLF had been strong. Today's volume-down close landed near 57.3. 57.2 has not broken. It already flashed a warning. Lose 57.2 and the next spots are 56.4 and 53.8, a roughly 3–5% pullback. Hold it if that drawdown is acceptable; cut if it is not. Rates matter, but the November midterms matter more for regulation and the rate path. Who controls the chambers, and whether government splits, rewrites the policy premium. Polymarket is about 50/50. Guessing the split now is noise. Trade 57.2. That is the nearer decision this week.
Hedge funds at the 4th percentile. CPI and two prints this week
US hedge-fund net longs and net leverage both sit around the 4th percentile. The one-year fundamental long/short ratio is there too. Near zero on that scale is extremely light. They are not chasing the long side or the short side. It looks like a vacation. The books with ammunition are still. The books that are full cannot dump. Mutual funds, active stock-pickers, and retail are already heavy. Hedge funds have dry powder and will not use it. What is left of the bid just rotates among software, mega-cap tech, and semis. Rotation can be violent. A new bid has not shown up. That is why the index cannot trend while a sector can look like a bull market for two days.
Two storylines this week. On the data side, PPI prints, then Friday's August CPI, with FOMC next week. CME has a roughly 59–60% chance of a 25 bp hike — hawkish, not a cut. A hot CPI, with gasoline expectations already up, makes the Fed's job harder, lifts hike odds another step, and hits growth and semi multiples first. A mild print lets them sit still, and the rotation can run a few more days. Mild inflation does not automatically mean index highs. Hedge funds can stay on vacation.
The stock stories are ORCL and ADBE after the close on September 10. ORCL lives in IGV, but it trades the OpenAI tape, more like a software name wearing a semiconductor jacket. Fresh OpenAI news would pull it with the compute story, not necessarily with the software ETF. ADBE is the software large-cap most discussed as an AI disruption risk. It needs a hard print — revenue that shows the franchise is still growing. Salesforce plugged into OpenAI and made itself a tool layer. ADBE has not shown an equivalent answer. Options imply a 68% chance of finishing within $22 of $255, roughly 230 versus 270–280. Inside that band is an ordinary hard test. Outside it is a large earnings event. Holders should know those two rails first.
TSLA did the usual pullback-then-bounce. That is its tape, not today's main line.
Three rules. Semis: wait for SOXX to clear the falling trendline before treating AMD as a new-high setup. Until then it is a hotspot, not a trend. Software: IGV holds 101; lose it and it is a range, not a short trend. Positioning: risk did not get worse, and hedge funds did not arrive. You can trade the rotation. Do not mistake rotation for a new systematic bid. Friday CPI and the two after-close prints on the 10th are the windows that can rewrite this map. If two of three hit — hot CPI, SOX fails, XLF loses 57.2 — cut the rotation down.