Retail sentiment deteriorated further, with technology stocks and semiconductors becoming the main areas of position reduction

Overview

This report tracks retail trading activity, thematic fund flows, factor crowding, and options behavior during July 23 to July 29. The core conclusion is that retail risk appetite cooled noticeably, with single stocks being sold over multiple consecutive days, and technology- and semiconductor-related assets becoming the main areas of position reduction; at the same time, AI data centers, the Mag 7, and some growth themes still received selective buying.

Core views

Deteriorating retail sentiment is the main theme of this issue: overall flows are near the lowest levels seen over the past year, with especially pronounced selling pressure in single stocks. Technology ETFs, tech hardware, and the semiconductor chain faced the greatest pressure, which may be related to worsening retail single-stock P&L since the start of the year, fading momentum trades, and earnings-season uncertainty. Even so, retail investors continued to buy AI data center, electrification, Mag 7, and AI software/commercialization themes, indicating that capital is not exiting risk assets altogether but rotating out of technology trades with larger losses and higher crowding.

Analysis framework

The report cross-checks market positioning through dimensions including retail cash flows, net buying of single stocks and ETFs, thematic baskets, sector flows, factor crowding, implied correlation, social media heat, hedge fund shorts, and options trading activity. Using weekly and daily flows, z-scores, historical percentiles, thematic basket changes, and key single-stock events, it evaluates the relationship between retail behavior and broader market factor rotation.

Methodology notes

• Flow analysisRetail Flow Tracker

  Net retail buying of stocks and ETFs

  Measures changes in risk appetite using net retail buying amounts, historical percentiles, and z-scores for individual stocks and ETFs, while distinguishing between single-stock and ETF allocation directions.

• Factor positioningCrowding Analysis

  Crowding and de-crowding

  Measures position crowding through historical percentiles of factors such as momentum, semiconductors, low quality, speculative growth, and value, identifying areas that may be forced to de-risk as momentum fades.

• Thematic basketsThematic Basket Flow

  Fund flows into themes such as AI data centers, Mag 7, and growth

  Aggregates single-stock trading by thematic baskets to observe whether retail investors are still buying themes such as AI, electrification, data centers, and large-cap technology.

• Derivatives activityRetail Options Activity

  Retail options trading share and active names

  Uses retail share in options trading and the most active names to identify leveraged risk appetite, event trading, and potential sources of volatility.

• Event-driven analysisEarnings Stories

  Retail trading and stock-price reactions around earnings

  Combines earnings or earnings expectations for companies such as META, MSFT, QCOM, AAPL, and AMZN to explain retail selling or buying behavior during peak earnings season.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• US.META

  Earnings-related event and retail position-reduction target

  Strengths

  Revenue met expectations, and the low end of 2026 capex guidance was raised from $125B and narrowed to $130-145B, showing that AI and infrastructure investment remains strong.

  Weaknesses

  Next-quarter guidance was weak, the stock fell as much as about 10% after hours following earnings, and retail investors were net sellers on July 29.

  Comparison

  Compared with MSFT's post-earnings rise, META's short-term market reaction was weaker; within the Mag 7, it was one of the names sold by retail investors this week alongside AAPL and AMZN.

  Risks

  Higher capex may compress free cash flow expectations; if revenue guidance cannot match AI investment, valuation and retail sentiment may remain under pressure.

• NVDA

  The primary Mag 7 and AI trading target bought by retail investors

  Strengths

  Retail investors bought +$601M this week, and it remains a core beneficiary of flows into AI, data center, and Mag 7 themes.

  Weaknesses

  Semiconductor crowding has declined but remains elevated, and fading momentum may bring volatility.

  Comparison

  Compared with semiconductor-related names sold such as SNDK, MU, and AMD, NVDA still attracted significant net retail buying.

  Risks

  If semiconductor de-crowding continues, NVDA may be affected by sector beta and momentum-factor volatility.

• AAPL

  Large-cap tech hardware name reduced ahead of earnings

  Strengths

  Analysts believe revenue and earnings results may prove more resilient than investors fear.

  Weaknesses

  Retail investors were net sellers of -$169M this week; after average price increases of about 20% for iPad and Mac, demand elasticity has become a key observation point.

  Comparison

  Along with META and AMZN, it was a Mag 7 member sold by retail investors this week, while NVDA, GOOGL/GOOG, TSLA, and MSFT were bought.

  Risks

  Higher memory costs are driving price increases; if near-term consumer response is weaker than expected, hardware revenue expectations may be affected.

• MSFT

  Earnings name driven by cloud growth and AI capex

  Strengths

  Cloud business sales rose +43% YoY, the stock rose about 3% after hours following earnings, and retail investors bought $23.4M on July 29.

  Weaknesses

  Quarterly capex rose +70% YoY to $41B, reflecting high capital spending intensity.

  Comparison

  Compared with META's decline due to weak guidance, MSFT's cloud growth performance was more supportive of the market reaction.

  Risks

  If high capex cannot continue to translate into cloud and AI revenue growth, valuation may face reassessment.

• QCOM

  Data center transition and earnings-event name

  Strengths

  Revenue beat expectations, the company expects FY2027 non-handset revenue growth to accelerate to 60%, and it is pushing into the data center market.

  Weaknesses

  Earnings were below expectations, and retail investors sold $6.8M on July 29.

  Comparison

  Unlike the traditional handset-chip dependence story, the company is trying to expand into data centers, but in the short term it is still affected by weaker-than-expected earnings.

  Risks

  Execution risk in data center expansion and earnings coming in below expectations may limit valuation recovery.

• Tech ETFs and SOXL

  Technology ETF outflows and leveraged semiconductor sentiment indicator

  Strengths

  They remain highly liquid tools for expressing risk appetite toward technology and semiconductors.

  Weaknesses

  Technology ETFs posted the second-largest weekly outflow on record, with SOXL being one of the main drivers, while crypto ETFs such as BITO were also reduced.

  Comparison

  ETFs overall still saw net inflows, but technology ETFs and crypto ETFs clearly moved against that trend with outflows.

  Risks

  Outflows from leveraged or high-volatility ETFs may amplify short-term price pressure on semiconductors and momentum factors.

Key data

• Total retail fund flow this week$2.2BBelow the past 12-month weekly average of $6.8B.
• Net ETF buying+$3.7BRetail investors continued to prefer ETFs over single stocks.
• Net single-stock buying-$1.4BSingle stocks saw clear net selling, including four consecutive days of single-stock selling.
• Single-stock flow in the technology sector-$2.2BTechnology was the largest net-selling sector outside the Mag 7.
• Main Mag 7 buysNVDA +$601M；GOOGL/GOOG +$377M；TSLA +$121M；MSFT +$33MRetail investors still selectively bought some large-cap technology stocks.
• Main Mag 7 sellsAAPL -$169M；META -$26M；AMZN -$22MDivergence emerged within large-cap technology around earnings season.
• Semiconductor crowding91.8%ileDown from 99%ile, reflecting semiconductor de-crowding.
• Low-quality factor crowding94.8%ileCrowding in high-risk factors remained elevated.
• Speculative growth factor crowding98.0%ileIndicating that high-risk growth trades remained crowded.
• Value factor crowding88.7%ileValue factor crowding was also relatively high.
• META event dataOn July 29, retail investors sold $2.2M, with a z-score of -0.4; shares fell as much as about 10% after hoursThe company narrowed its 2026 capex guidance to $130-145B, revenue met expectations, but next-quarter guidance was weak.
• MSFT event dataOn July 29, retail investors bought $23.4M; cloud business sales rose +43% YoY; quarterly capex increased +70% YoY to $41BThe stock rose about 3% after hours following earnings.
• QCOM event dataOn July 29, retail investors sold $6.8M; FY2027 non-handset revenue growth is expected to accelerate to 60%The company is trying to reduce dependence on its handset business through the data center market.

Impact & implications

For investors, this report suggests that short-term risk comes not only from fundamental earnings themselves, but also from portfolio de-risking pressure driven by crowded momentum trades, semiconductor deleveraging, and rising implied correlation. If selling pressure in technology and semiconductors continues, it may keep weighing on high-beta and high-momentum assets; however, retail buying remains in AI data centers, the Mag 7, and some growth themes, indicating that thematic trading has not broken down completely, and the more critical question ahead is whether earnings can support expectations for capex, cloud growth, and AI commercialization.

Risks

• Fading momentum and month-end rebalancing may continue to amplify volatility in high-beta, high-momentum, and high-risk factors.
• Concentrated selling in tech hardware, semiconductors, and crypto ETFs may trigger further de-crowding.
• If cloud growth, returns on AI capex, or consumer demand come in below expectations during peak earnings season, divergence among large-cap technology stocks may intensify.
• Implied stock correlation has risen from low levels, which may raise portfolio risk estimates in risk models and push systematic or quantitative portfolios to reduce net exposure.
• Crowding in low-quality and speculative growth remains high, and if risk constraints tighten, related assets may face forced position reduction pressure.

What to watch

• Whether retail single-stock selling continues, especially in tech hardware and the semiconductor chain.
• Whether META's forward revenue outlook, AI capex, and return-on-investment narrative improve.
• Growth in cloud businesses, capex guidance, and retail margin changes at MSFT and AMZN.
• Demand elasticity for iPad, Mac, and future iPhone 18 after AAPL's price increases.
• Whether semiconductor crowding continues to fall from elevated levels, and whether core AI names such as NVDA can still absorb buying.
• Whether changes in implied correlation and momentum beta exposure trigger further de-risking by quantitative portfolios.
• With retail options trading share at a high level, event-driven volatility in TSLA, NVDA, MU, AMZN, SNDK, META, AMD, AAPL, SPCX, and MSFT.