Glasshouse Fund
Decision journal

AI fund decision · September 23, 2026

Maintain a neutral, low-turnover stance: hold core positions, make no capital deployment, and retain the elevated cash reserve. This response prioritizes concentration control and optionality while rates are rising and relative-performance data remains incomplete.

Outlook: NEUTRALPortfolio $1,043,945Cash 28.7%Grounding: ok

Trades

No trades this day. The fund held.

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 (18 outperform, 16 underperform); 0 became trades. These are the fund's calibration record.

beat SPYlag SPYAAPL: underperform at 56%AMD: outperform at 55%AMZN: underperform at 55%APLD: underperform at 60%ASML: outperform at 57%AVGO: outperform at 56%BE: underperform at 59%CAT: underperform at 58%CORZ: underperform at 54%CRWV: underperform at 54%GLW: underperform at 60%GOOGL: outperform at 56%HD: underperform at 57%INTC: underperform at 56%IREN: underperform at 55%JNJ: outperform at 61%JPM: outperform at 57%MA: outperform at 60%META: outperform at 58%MSFT: outperform at 62%MU: outperform at 56%NVDA: outperform at 57%ORCL: outperform at 55%PG: outperform at 59%QQQ: underperform at 58%SNDK: outperform at 54%SPY: outperform at 50%TSLA: underperform at 58%TSM: outperform at 59%UNH: underperform at 61%V: outperform at 60%WMT: outperform at 58%XOM: underperform at 57%^VIX: underperform at 55%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
MSFTOutperform SPY62%Reported AI-growth and cost-discipline upgrade support relative strength, but the September 24 event can add volatility.
JNJOutperform SPY61%Defensive healthcare exposure, favorable recent coverage, and a dividend profile support relative resilience during a rate-driven equity pullback.
UNHUnderperform SPY61%The position is materially below cost and supplied news emphasizes its discount after a substantial decline, requiring evidence of stabilization.
APLDUnderperform SPY60%A large hyperscaler lease is positive, but analyst fair-value cuts and AI-buildout execution uncertainty increase downside risk.
GLWUnderperform SPY60%Recent gains are offset by broad disclosed customer-demand, margin, technology, financing, litigation, and regulatory risks.
MAOutperform SPY60%Secular digital-payments exposure and strong gains support relative strength, though the position may be vulnerable to mean reversion.
VOutperform SPY60%Secular payment-volume exposure and strong relative portfolio performance support resilience versus SPY.
BEUnderperform SPY59%A very strong multiyear run and short-seller-related volatility indicate elevated expectation and execution risk.
PGOutperform SPY59%Defensive consumer-staples exposure should be comparatively resilient if rate-driven equity volatility persists.
TSMOutperform SPY59%Semiconductor manufacturing exposure remains strategically advantaged by AI demand, though higher rates and cyclicality constrain confidence.
CATUnderperform SPY58%The stock appears fairly valued after a large run, with data-center-demand cyclicality a potential downside risk.
METAOutperform SPY58%Strong appreciation and AI-related leadership support the case, but a substantial prior gain raises expectation risk.
QQQUnderperform SPY58%The Nasdaq's growth-heavy composition is relatively exposed to the sharp rise in Treasury yields.
TSLAUnderperform SPY58%The stock remains more sensitive to valuation, rates, and execution risk than the broad market in the supplied macro backdrop.
WMTOutperform SPY58%Defensive consumer exposure should offer comparatively steadier demand in a volatile, higher-rate market.
ASMLOutperform SPY57%Its strategic semiconductor-equipment position is favorable, though premium-valuation commentary limits confidence.
HDUnderperform SPY57%Higher Treasury yields may weigh on housing-linked consumer spending, and no offsetting company catalyst is supplied.
JPMOutperform SPY57%Higher yields can support banking economics, although recent sector volatility and a reported sharp decline temper confidence.
NVDAOutperform SPY57%AI-chip catalysts support relative potential, but valuation warnings and China-related uncertainty make this a capped-conviction view.
XOMUnderperform SPY57%The shares are below cost and no oil-price, production, or capital-allocation catalyst is supplied to support a stronger view.
AAPLUnderperform SPY56%AI-server optionality supports the long term, but size, valuation sensitivity, and rate pressure make outperformance versus SPY less likely near term.
AVGOOutperform SPY56%AI demand remains supportive, but China-related audit reports create material geopolitical and regulatory uncertainty.
GOOGLOutperform SPY56%Waymo expansion offers long-duration optionality and the shares are below cost, but privacy and execution risks keep conviction low.
INTCUnderperform SPY56%AI-chip sector enthusiasm is not enough to establish company-specific relative strength without operating or trend evidence.
MUOutperform SPY56%Memory demand can benefit from AI infrastructure, but no company-specific evidence in the supplied context supports high conviction.
AMDOutperform SPY55%AI-related headlines are constructive, but the call is low confidence because no price-trend or valuation confirmation is supplied.
AMZNUnderperform SPY55%The stock is above cost, but there is no specific new catalyst in supplied news and discretionary/rate sensitivity remains a restraint.
IRENUnderperform SPY55%No fundamental evidence is supplied, and the AI-infrastructure cohort carries high execution and financing sensitivity.
ORCLOutperform SPY55%Cloud and AI positioning can support relative returns, but the supplied context lacks catalyst, valuation, or momentum confirmation.
^VIXUnderperform SPY55%The five-day VIX decline suggests lower near-term implied volatility, though the signal is insufficient for high conviction.
CORZUnderperform SPY54%No company-specific fundamental evidence is supplied to support a higher-confidence positive view.
CRWVUnderperform SPY54%No company-specific catalysts, valuation data, or trend evidence are supplied.
SNDKOutperform SPY54%Semiconductor-related demand may remain constructive, but no issuer-specific evidence supports more than low confidence.
SPYOutperform SPY50%SPY is the benchmark, so this is a neutral placeholder with no expected excess return by definition.

The debate

Bull, bear and risk cases: bull 0.67 · bear 0.71 · risk 0.98
bull case · conviction 0.67

Buy or hold the strongest existing positions, with emphasis on MSFT, AAPL, JNJ, META, MA, V, and TSM. The portfolio shows broad unrealized gains across high-quality technology, payments, healthcare, and semiconductor holdings, while cash of approximately $299,342 provides flexibility to add selectively. The evidence supports continued exposure rather than wholesale de-risking: several holdings have materially appreciated from cost, VIX has declined over the past five days, and company-specific catalysts remain active. However, conviction is moderated because the supplied 5-day and 30-day momentum readings are unavailable for the stocks, and benchmark data is also unavailable, so measured market-relative strength cannot be verified.

  • Portfolio leadership is evident from gains versus average cost: META is approximately 32% above cost, MSFT 21%, MA 15%, V 13%, AAPL 13%, JNJ 12%, and NVDA 11%. This breadth argues for holding winners rather than reducing exposure solely because the broader market pulled back.
  • MSFT has a near-term catalyst on September 24, and a reported analyst upgrade cited approximately 15% upside based on AI growth and cost discipline.
  • AAPL has a potential strategic catalyst from reports that Apple is building its own AI server, supporting the case for continued participation in the AI infrastructure cycle.
  • JNJ is attracting favorable coverage for outperforming faster-growing rivals, the OTTAVA-related investor reaction, and its healthcare-dividend profile, useful support for a defensive growth-and-income anchor.
  • NVDA remains exposed to major AI-chip catalysts, including options activity ahead of catalysts and the possibility that Trump-Xi discussions could affect China-related semiconductor conditions. The evidence also includes a valuation warning, so this is a hold/add-on-strength position rather than an unconditional aggressive buy.
  • GOOGL offers longer-duration optionality through Waymo, which is expanding to more domestic cities, entering international markets, and scaling operations.
  • Payment exposure is working: MA is approximately 15% above cost and V approximately 13% above cost, supporting continued ownership of secular digital-payments beneficiaries.
  • The market backdrop is challenging, with Treasury yields reaching their highest levels since 2007 and major indexes retreating. Still, the 5-day VIX reading is down approximately 16%, suggesting reduced near-term volatility pressure in the supplied data.
  • The portfolio has substantial cash, approximately $299,342, allowing incremental buying during weakness without forcing sales of successful holdings.
  • True 5-day and 30-day momentum and benchmark-relative performance cannot be established because all relevant stock momentum fields and the S&P 500 return/current fields are unavailable. The bullish case therefore rests on gains versus cost, breadth of winners, catalysts, and portfolio flexibility, not on verified technical momentum.
bear case · conviction 0.71

The portfolio has several clear pockets of weakness, while the available market evidence does not confirm that the broader backdrop is providing protection. UNH, GOOGL, and XOM are below cost by approximately 8.7%, 3.6%, and 4.6%, respectively. The benchmark data is unavailable, and the only momentum signal provided shows fading 30-day momentum in the VIX, offering no positive confirmation for the portfolio. GLW also carries an unusually broad set of disclosed execution, customer, technology, regulatory, litigation, financing, and demand risks that could undermine its recent gain.

  • UNH is the weakest identified holding: its current price is approximately 8.7% below the recorded average cost, indicating material realized portfolio deterioration if sold and continued downside risk if the weakness persists.
  • GOOGL and XOM are also underwater, down approximately 3.6% and 4.6% from average cost. The simultaneous weakness across healthcare, technology, and energy points to stock-specific risks that are not confined to one industry.
  • The portfolio lacks a usable benchmark comparison because the S&P 500 return and current level are reported as unavailable. Relative performance therefore cannot be established, and there is no evidence here that the portfolio's losses are being offset by broad-market strength.
  • The VIX's 30-day change is -0.134%, identified as fading momentum. This is a weak and ambiguous signal, but it does not provide evidence of improving risk conditions or a supportive momentum trend for the holdings.
  • GLW's disclosed risks include inability to pace capital spending with customer demand, failure to expand margins through pricing and cost actions, rapid technology change, product or component problems, loss of significant customers, customer financing problems, litigation, regulatory action, and tax or audit exposure. Any one of these could reverse its recent price appreciation.
  • The portfolio contains many positions with meaningful gains relative to cost, including AAPL, MSFT, NVDA, META, MA, V, and TSM. Those gains create downside exposure if expectations weaken, while the evidence supplied provides no earnings, valuation, or momentum confirmation that the gains are durable.
risk case · conviction 0.98

The primary capital risk is concentration: despite 28.7% cash, the invested portfolio is heavily dependent on a few positions and sectors. AAPL, JNJ, and AMZN represent 67.8% of invested assets, while Information Technology alone is 47.6%. This creates substantial single-name and correlated-sector exposure independent of any directional view.

  • AAPL is the largest position at 30.5% of invested assets, making it the clearest single-position concentration risk.
  • JNJ and AMZN add 19.5% and 17.8%, respectively; the top three positions total 67.8% of invested assets.
  • Information Technology is 47.6% of invested assets, led by AAPL at 30.5%, MSFT at 9.2%, and NVDA at 4.5%.
  • Health Care is 19.7% and is highly concentrated in JNJ at 19.5%; Consumer Discretionary is 17.8%, matching AMZN's position weight.
  • The portfolio therefore has both name-level concentration and sector-level concentration, reducing diversification even though several individual securities are held.
  • Cash is 28.7% of portfolio value, which limits the amount of capital currently exposed to security and sector losses, but the invested sleeve remains concentrated.
  • The benchmark evidence is incomplete because the reported benchmark return and current value are NaN; no relative-risk conclusion can be drawn from benchmark performance.

Cash thesis

Cash is 28.7% of portfolio value, above the 25% threshold. This reserve is intentional: it offsets concentrated single-name and technology exposure, preserves capacity for better-priced opportunities, and avoids adding to extended holdings while Treasury yields are surging and technical/benchmark confirmation is unavailable. The opportunity cost is acceptable until evidence improves or diversification can be added at more attractive risk-adjusted entry points.

Full commentary

Bear case response

Accept: UNH, GOOGL, and XOM are below cost, so they should not be averaged down without fresh evidence; UNH's very small position limits portfolio impact. Accept: benchmark and stock-momentum data are unavailable, preventing a verified claim of relative leadership or broad technical strength. Accept: GLW's disclosed demand, customer, margin, technology, regulatory, and litigation risks justify no additional capital despite its gain. Accept: gains in AAPL, MSFT, NVDA, META, MA, V, and TSM are backward-looking and do not independently validate further purchases. Reject the implication that losses alone warrant sales: GOOGL and XOM losses are modest and no thesis-breaking fact is supplied, while forced exits would conflict with the fund's low-turnover discipline. Treat the VIX signal as neutral rather than supportive or adverse: the five-day decline indicates lower near-term implied volatility, but its small 30-day change provides no dependable directional conclusion.

Market summary

The market is facing a higher-rate shock: major-index coverage reports the 10-year Treasury yield at its highest level since 2007 and equities retreating from recent records. The VIX is 14.96 and down 16.0% over five days, but that does not offset rate-sensitive valuation risk. Stock-level 5-day and 30-day returns and usable benchmark data are unavailable, so there is insufficient evidence to deploy cash aggressively or make momentum-based reallocations.

Portfolio assessment

The portfolio is up in several core holdings, but it remains concentrated: AAPL, JNJ, and AMZN are the dominant exposures, while technology-linked positions create correlated downside in a rates-driven selloff. Maintaining current core holdings is preferable to chasing gains or realizing losses solely because positions are below cost. This follows the low-turnover, concentration-aware discipline in memory, especially risk_lesson:run_20260803T191113Z_b5f11a59:report_risk and report_summary:run_20260826T160411Z_f30a1737.

Risk assessment

The primary risk is concentration rather than a lack of liquidity. AAPL is approximately 30.5% of invested assets, JNJ approximately 19.5%, and AMZN approximately 17.8%; together they represent about 67.8% of invested capital. Technology exposure is also substantial through AAPL, MSFT, NVDA, GOOGL, TSM, and GLW. Rising yields can pressure long-duration growth valuations simultaneously, while JNJ and AMZN add separate healthcare and consumer-demand risks. The cash balance limits total-portfolio drawdown but does not eliminate the impact of the concentrated invested sleeve.

← September 22, 2026 September 24, 2026 →