Glasshouse Fund
Decision journal

AI fund sells HD · September 21, 2026

Maintain a neutral, low-turnover posture. Sell HD to remove the clearest small underperformer, hold the core winners without adding, avoid weak AI-infrastructure candidates, and retain the 27.0% cash reserve because concentration and uneven market breadth remain the dominant risks.

Outlook: NEUTRALPortfolio $1,054,478Cash 28.4%Grounding: ok

Trades

SELL 50 HD76% · filled at $299

Exit the small HD position: it is down 8.65% versus cost, down 10.56% over 30 days, and down 3.50% over five days. The sustained relative weakness provides a clearer trimming signal than the mixed evidence elsewhere and modestly reduces cyclical consumer exposure.

Risks the AI named

Market calls

A directional call (beat or lag the S&P 500 over the horizon) on every researched name, whether or not the fund traded it. 34 calls (19 outperform, 15 underperform); 0 became trades. These are the fund's calibration record.

beat SPYlag SPYAAPL: outperform at 64%AMD: outperform at 61%AMZN: underperform at 59%APLD: underperform at 64%ASML: underperform at 58%AVGO: underperform at 57%BE: outperform at 58%CAT: underperform at 60%CORZ: underperform at 62%CRWV: underperform at 72%GLW: underperform at 60%GOOGL: outperform at 55%HD: underperform at 71%INTC: outperform at 60%IREN: outperform at 56%JNJ: outperform at 63%JPM: underperform at 56%MA: outperform at 56%META: outperform at 66%MSFT: outperform at 57%MU: outperform at 61%NVDA: outperform at 54%ORCL: outperform at 53%PG: outperform at 52%QQQ: underperform at 54%SNDK: outperform at 60%SPY: outperform at 50%TSLA: outperform at 56%TSM: outperform at 59%UNH: underperform at 62%V: outperform at 55%WMT: underperform at 63%XOM: underperform at 57%^VIX: underperform at 62%50%60%70%80%90%100%
Each tick is one call, placed by stated confidence. Hover for the name.
All 34 calls, with reasoning
SymbolCallConf.Why
CRWVUnderperform SPY72%CRWV's 23.25% 30-day decline and cited credit-market concerns create a high-risk relative outlook.
HDUnderperform SPY71%Negative returns over both five and 30 days, combined with a loss versus cost, indicate persistent relative weakness.
METAOutperform SPY66%META's 17.01% 30-day gain is the strongest held momentum signal, although extension risk prevents higher confidence.
AAPLOutperform SPY64%A 9.99% 30-day gain and positive five-day momentum show relative strength, but size and recent appreciation limit confidence.
APLDUnderperform SPY64%The 9.81% 30-day decline and valuation concerns outweigh the sharp five-day rebound.
JNJOutperform SPY63%Positive short- and medium-term momentum plus talc-resolution optimism support a defensive relative advantage.
WMTUnderperform SPY63%WMT is below cost and down 6.42% over 30 days, with continued five-day weakness.
CORZUnderperform SPY62%CORZ's 10.63% 30-day decline and skepticism around AI-pivoted miners outweigh its five-day rebound.
UNHUnderperform SPY62%UNH is below cost and negative over both five and 30 days, signaling sustained relative weakness.
^VIXUnderperform SPY62%VIX has declined over five and 30 days, consistent with subdued volatility demand if risk appetite remains stable.
AMDOutperform SPY61%AMD's 13.46% five-day and 10.64% 30-day gains indicate strong semiconductor relative momentum, though the move is extended.
MUOutperform SPY61%MU's 9.93% five-day gain and marginally positive 30-day return support semiconductor relative strength.
CATUnderperform SPY60%CAT's 8.24% 30-day decline remains the dominant signal despite a positive recent week.
GLWUnderperform SPY60%A 13.16% 30-day decline and documented execution risks outweigh the recent 4.55% rebound.
INTCOutperform SPY60%INTC's 11.74% five-day and 4.94% 30-day gains indicate improving relative momentum.
SNDKOutperform SPY60%SNDK's 15.45% five-day gain and slightly positive 30-day return support relative strength, with high volatility.
AMZNUnderperform SPY59%AMZN's 2.91% 30-day decline versus a nearly flat benchmark indicates weaker relative momentum despite a stable five-day reading.
TSMOutperform SPY59%Positive returns over both periods, including a 4.26% five-day gain, support improving AI supply-chain relative strength.
ASMLUnderperform SPY58%ASML's 10.79% 30-day decline signals weak relative trend despite a constructive 6.65% five-day bounce.
BEOutperform SPY58%BE has positive momentum over both five days and 30 days, supported by favorable company-specific coverage, but remains volatile.
AVGOUnderperform SPY57%A positive five-day move does not yet offset AVGO's 8.87% 30-day decline.
MSFTOutperform SPY57%Positive 30-day performance and sustained AI relevance support relative strength despite a 2.30% five-day decline.
XOMUnderperform SPY57%XOM is below cost with negative five-day momentum; modest positive 30-day performance does not establish a clear recovery.
IRENOutperform SPY56%Positive five- and 30-day momentum supports relative strength, tempered by the volatility of the business category.
JPMUnderperform SPY56%JPM's 3.13% 30-day decline and flat recent action indicate weaker relative momentum.
MAOutperform SPY56%MA's positive 30-day return supports a modest relative edge despite a negative five-day pullback.
TSLAOutperform SPY56%Positive five-day and 30-day momentum support a modest relative-strength view, tempered by volatility.
GOOGLOutperform SPY55%Slightly positive 30-day performance and strategic optionality support a modest relative edge, though momentum is muted.
VOutperform SPY55%V's positive 30-day return supports a modest relative advantage despite recent five-day softness.
NVDAOutperform SPY54%The 5.36% five-day rebound supports relative strength, but the negative 30-day return and AI-spending skepticism keep conviction low.
QQQUnderperform SPY54%QQQ is down 1.15% over 30 days, indicating technology breadth has lagged SPY despite recent improvement.
ORCLOutperform SPY53%Modestly positive five- and 30-day performance supports only a narrow relative advantage.
PGOutperform SPY52%PG's modest positive returns over both periods provide a low-conviction defensive relative case.
SPYOutperform SPY50%SPY is the comparison baseline; the call is neutral by construction and carries no informational edge.

The debate

Bull, bear and risk cases: bull 0.76 · bear 0.90 · risk 0.98
bull case · conviction 0.76

The strongest buy-and-hold case is concentrated in AAPL, META, JNJ, MSFT, GOOGL, TSM, and selected AI beneficiaries. Portfolio leaders are materially outperforming a virtually flat S&P 500 over 30 days, while the broader tape is improving: AI-related stocks are surging, the Nasdaq is approaching a record close, oil and yields are falling, and VIX has declined 13.4% over five days. AAPL combines strong 30-day momentum with constructive catalyst coverage, META is the clearest momentum leader, JNJ is benefiting from litigation-resolution optimism and defensive healthcare demand, and the AI ecosystem remains supported by favorable market leadership and company-specific developments. The evidence supports holding winners and selectively adding to relative-strength names, while acknowledging that several holdings remain weak and the data-center narrative faces skepticism.

  • AAPL rose 9.99% over 30 days and 0.92% over five days versus the S&P 500's essentially flat 30-day return (-0.03%), providing clear relative strength. Coverage highlights shares near record highs and reports that Apple is building its own AI server.
  • META is the portfolio's strongest momentum holding, up 17.01% over 30 days, with performance far ahead of the benchmark despite only modest five-day movement.
  • JNJ gained 3.41% over 30 days and 1.38% over five days, outperforming the benchmark; reports cite an 18% three-month advance, a robust growth-and-dividend outlook, and a proposed $5.5 billion talc-litigation resolution that could reduce uncertainty.
  • MSFT is up 2.99% over 30 days despite a 2.30% five-day pullback, while news reports a raised price target and continued debate around AI safety and spending, evidence of ongoing investor attention to its AI opportunity.
  • GOOGL is up 1.68% over 30 days and has strategic optionality through Waymo, which its filing describes as expanding to more cities, entering international markets, and scaling operations.
  • TSM is up 1.13% over 30 days and 4.26% over five days, showing improving short-term relative strength within the semiconductor and AI supply chain.
  • NVDA rebounded 5.36% over five days after a slightly negative 30-day period (-1.11%); the rebound aligns with market news that AI-related stocks are driving the Nasdaq toward a record close.
  • Risk appetite is improving: VIX fell 13.39% over five days and 2.50% over 30 days, while market coverage reports a 1% S&P 500 rise and strong AI-stock leadership.
  • The bullish case is selective rather than indiscriminate: AMZN, HD, UNH, WMT, XOM, and GLW show negative 30-day momentum, and The New York Times reports growing skepticism about the data-center boom. Those facts argue for concentration in relative-strength names rather than broad aggressive buying.
bear case · conviction 0.90

The strongest bear case is that portfolio weakness is broadening beyond isolated losers while market momentum is deteriorating across several major holdings and candidate securities. HD, UNH, XOM, WMT, and PG are already below cost, while HD, UNH, WMT, XOM, GLW, AMZN, JPM, and NVDA show negative 30-day momentum. This is occurring against an essentially flat S&P 500 over the same period, suggesting these names are underperforming rather than merely following a broad market decline. GLW is particularly vulnerable: its disclosed risks include customer losses or financing problems, demand-related capital-spending misjudgment, margin pressure, technology change, product issues, litigation, and regulatory developments.

  • HD is both materially below its average cost and down 10.6% over 30 days, making it one of the clearest examples of sustained deterioration.
  • UNH is down 7.1% from average cost and has negative 30-day momentum of 4.1%, indicating that the loss is continuing rather than stabilizing.
  • WMT is below cost and down 6.4% over 30 days, while PG is marginally below cost; the weakness is therefore not limited to cyclical or technology holdings.
  • XOM is down 5.9% from average cost, adding commodity-sensitive downside to the portfolio's existing weakness.
  • GLW is up from the listed average cost but has the worst 30-day momentum among current holdings at negative 13.2%. Its disclosed risks include loss of significant customers, customer inability to finance operations or pay receivables, difficulty aligning capital spending with demand, margin execution risk, technology change, product performance issues, litigation, and regulatory action.
  • AMZN is down 2.9% over 30 days, JPM is down 3.1%, and NVDA is down 1.1%; negative momentum across consumer, financial, and semiconductor exposure suggests deterioration is not confined to one sector.
  • Several non-held candidates also have severe negative 30-day momentum: CRWV is down 23.2%, APLD 9.8%, ASML 10.8%, AVGO 8.9%, and CAT 8.2%. Adding these names would expose the portfolio to already weakening trends rather than provide clear evidence of stabilization.
  • The S&P 500 is essentially flat over 30 days, while SPY and QQQ are both down about 1.2%; the portfolio's weaker names are therefore showing meaningful relative deterioration.
  • The evidence is less bearish for AAPL, MSFT, META, MA, V, TSM, JNJ, and GOOGL because no negative momentum or below-cost signal is provided for them. However, the absence of a downside signal is not evidence that those holdings will offset the deterioration elsewhere.
risk case · conviction 0.98

The portfolio's primary capital risk is concentration: despite a substantial 27.0% cash balance, invested assets are dominated by a few positions and sectors. AAPL, JNJ, and AMZN represent 66.5% of invested assets, while the top five positions represent 82.5%. Sector exposure is similarly concentrated, with Information Technology, Consumer Discretionary, and Health Care totaling 85.4% of invested assets. Cash provides a meaningful portfolio-level buffer, but it does not offset the vulnerability of the invested sleeve to adverse outcomes in its largest holdings or sectors.

  • AAPL is the largest exposure at 29.7% of invested assets and is the single biggest source of position-concentration risk.
  • JNJ and AMZN add 19.0% and 17.8%, respectively; together with AAPL they comprise 66.5% of invested assets.
  • The five largest positions. AAPL, JNJ, AMZN, MSFT, and GOOGL, account for 82.5% of invested assets, leaving limited diversification within the invested portfolio.
  • Information Technology is 46.4% of invested assets, creating the largest sector tilt; Consumer Discretionary and Health Care add 19.8% and 19.2%.
  • The three largest sectors total 85.4% of invested assets, so sector-level weakness could affect a large share of the invested sleeve even without relying on any directional forecast.
  • Cash is 27.0% of the total portfolio and materially reduces aggregate exposure, but the remaining capital is concentrated rather than broadly diversified.
  • The benchmark's approximately flat 30-day return provides no evidence that the portfolio's concentration risk has been mitigated.

Cash thesis

Cash is $284,387, or 27.0% of portfolio value. This elevated reserve is intentional: it offsets, but does not eliminate, concentrated exposure to AAPL, JNJ, AMZN, and technology-linked holdings. With benchmark returns essentially flat over 30 days, mixed breadth, and no tool-supported catalyst requiring immediate deployment, cash preserves optionality for a broader pullback or more durable confirmation while avoiding purchases after sharp gains.

Full commentary

Bear case response

Accept: HD, UNH, WMT, XOM, AMZN, JPM, NVDA, and especially GLW show weaker 30-day momentum than the nearly flat benchmark; this is meaningful relative deterioration rather than simply broad-market weakness. Accept: HD is the most actionable case because it is below cost and weakening on both five- and 30-day measures, justifying a small risk-reduction sale. Accept: GLW carries documented operating, customer, demand, margin, technology, litigation, and regulatory risks; however, its 4.55% five-day rebound and positive return versus cost support monitoring rather than selling today. Accept: non-held AI-infrastructure candidates such as CRWV, APLD, ASML, AVGO, CORZ, and CAT lack sufficient trend confirmation for purchase. Reject as a blanket conclusion: weakness in several holdings does not require broad liquidation because AAPL, META, JNJ, MSFT, GOOGL, V, MA, and TSM retain neutral-to-positive relative signals, and cash already materially limits total portfolio exposure. Accept the concentration warning: no additions will be made to the large winners, particularly AAPL and AMZN, because doing so would worsen single-name and sector risk. This response follows the fund's higher-priority lessons to use selective incremental trims, avoid repeated turnover, and preserve liquidity.

Market summary

Market tone is constructive in the very short term: reported coverage indicates AI-led strength, lower oil and yields, and a 13.4% five-day decline in VIX. However, SPY and QQQ remain modestly negative over 30 days, leadership is narrow, and several portfolio holdings have materially lagged. This supports retaining core winners but not deploying cash broadly into extended technology or AI-related names.

Portfolio assessment

The portfolio has strong embedded gains in AAPL, JNJ, MSFT, META, V, MA, and TSM, but remains concentrated: AAPL, JNJ, and AMZN dominate invested assets. AAPL's 9.99% 30-day gain, JNJ's 3.41% gain, and META's 17.01% gain support holding those cores, not adding to them. HD is the clearest held deterioration, with a 10.56% 30-day decline, a 3.50% five-day decline, and an 8.65% loss versus cost. A small complete exit from HD reduces an underperforming position without broad portfolio turnover.

Risk assessment

The key risk is concentration rather than immediate market-wide stress. AAPL, JNJ, and AMZN represent a large share of invested capital, while the largest sectors are Information Technology, Consumer Discretionary, and Health Care. A reversal in mega-cap technology, consumer demand, or AI-investment expectations could affect several holdings simultaneously. GLW's negative 13.16% 30-day return warrants monitoring despite its recent five-day rebound and gain versus cost. The portfolio should avoid rebuilding concentration through new additions to AAPL, MSFT, NVDA, or AI-infrastructure watchlist names.

← September 18, 2026 September 22, 2026 →