Glasshouse Fund
Decision journal

AI fund decision · September 7, 2026

Maintain a neutral, low-turnover posture. Hold existing core winners, do not add to AMZN, GOOGL, HD, or UNH, avoid chasing high-momentum semiconductor candidates, and retain the 27.5% cash reserve as a concentration and volatility buffer.

Outlook: NEUTRALPortfolio $1,043,332Cash 27.5%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 (21 outperform, 13 underperform); 0 became trades. These are the fund's calibration record.

beat SPYlag SPYAAPL: outperform at 61%AMD: underperform at 58%AMZN: underperform at 67%APLD: underperform at 64%ASML: outperform at 59%AVGO: underperform at 62%BE: outperform at 55%CAT: underperform at 54%CORZ: underperform at 66%CRWV: outperform at 56%GLW: outperform at 57%GOOGL: underperform at 68%HD: underperform at 61%INTC: outperform at 58%IREN: outperform at 56%JNJ: outperform at 69%JPM: outperform at 58%MA: outperform at 57%META: outperform at 65%MSFT: outperform at 59%MU: outperform at 57%NVDA: outperform at 68%ORCL: outperform at 61%PG: underperform at 53%QQQ: outperform at 56%SNDK: outperform at 55%SPY: outperform at 50%TSLA: underperform at 56%TSM: outperform at 60%UNH: underperform at 59%V: outperform at 58%WMT: underperform at 54%XOM: outperform at 55%^VIX: underperform at 60%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
JNJOutperform SPY69%Its strong 5-day and 30-day gains and defensive positioning support continued relative strength versus SPY.
GOOGLUnderperform SPY68%The -9.33% 30-day decline, position below cost, and cloud-team reduction headline indicate weak relative momentum.
NVDAOutperform SPY68%Strong 5-day and 30-day gains, plus continued AI leadership news, support relative strength despite semiconductor concentration risk.
AMZNUnderperform SPY67%The -8.98% 30-day decline and negative 5-day return leave AMZN with the weakest trend among major portfolio holdings.
CORZUnderperform SPY66%Its -21.64% 30-day return outweighs the recent rebound and signals elevated downside volatility.
METAOutperform SPY65%Strong 5-day and positive 30-day momentum support continued relative strength, subject to large-cap technology volatility.
APLDUnderperform SPY64%A positive 5-day bounce does not offset its -10.58% 30-day decline and high-volatility profile.
AVGOUnderperform SPY62%Negative returns over both 5 and 30 days indicate a currently weaker trend than SPY.
AAPLOutperform SPY61%Positive 5-day and 30-day momentum plus a product-event catalyst support modest relative strength versus SPY, despite concentration and supply-chain risks.
HDUnderperform SPY61%Negative 5-day and 30-day returns, combined with a price below cost, indicate unfavorable relative trend.
ORCLOutperform SPY61%Positive momentum and earnings-related optimism support a favorable relative view, with event risk limiting conviction.
TSMOutperform SPY60%Positive returns across both periods support continued semiconductor relative strength.
^VIXUnderperform SPY60%Declining 5-day and 30-day volatility trends favor a lower VIX level relative to SPY, though geopolitical risk can reverse this quickly.
ASMLOutperform SPY59%Positive 5-day and 30-day momentum supports relative strength, though semiconductor-cycle sensitivity limits conviction.
MSFTOutperform SPY59%Positive 30-day performance supports relative strength, though the recent decline and mixed news lower conviction.
UNHUnderperform SPY59%The negative 30-day return and price below cost outweigh the recent 5-day rebound.
AMDUnderperform SPY58%Its negative 30-day return indicates weaker relative momentum than SPY despite a strong recent 5-day rebound.
INTCOutperform SPY58%Strong positive 5-day and 30-day performance supports relative strength, albeit with semiconductor volatility.
JPMOutperform SPY58%Positive returns over both horizons support modest relative strength, though the financial-sector signal is not broad.
VOutperform SPY58%Positive 30-day performance supports relative strength, although the recent 5-day decline moderates confidence.
GLWOutperform SPY57%Positive momentum across both periods supports a modestly favorable relative view versus SPY.
MAOutperform SPY57%Positive 30-day performance supports a favorable relative view despite a weak recent 5-day period.
MUOutperform SPY57%Very strong 30-day momentum supports outperformance, but the sharp advance raises cycle and mean-reversion risk.
CRWVOutperform SPY56%Positive 5-day and 30-day returns support relative strength, but limited fundamental evidence constrains conviction.
IRENOutperform SPY56%Strong reported momentum and GPU-financing news are constructive, but the name remains highly volatile.
QQQOutperform SPY56%Its positive 30-day return and resilient technology leadership support modest relative strength, despite uneven breadth.
TSLAUnderperform SPY56%A strong 30-day return is offset by a negative 5-day move and elevated volatility, weakening near-term relative confidence.
BEOutperform SPY55%Strong reported 5-day and 30-day gains support momentum, but the sharp move and valuation uncertainty make confidence low.
SNDKOutperform SPY55%Exceptional reported momentum supports outperformance, but the magnitude of the advance creates substantial mean-reversion risk.
XOMOutperform SPY55%Positive 30-day performance and geopolitical sensitivity provide a modest relative-strength case despite recent weakness.
CATUnderperform SPY54%The positive weekly move is offset by a slightly negative 30-day return, leaving limited evidence of durable relative strength.
WMTUnderperform SPY54%A positive weekly move has not yet reversed the negative 30-day trend.
PGUnderperform SPY53%Positive but modest returns suggest stability rather than a clear catalyst for outperformance versus SPY.
SPYOutperform SPY50%SPY is the comparison proxy; its recent positive 5-day and 30-day returns indicate a stable broad-market baseline.

The debate

Bull, bear and risk cases: bull 0.76 · bear 0.78 · risk 0.97
bull case · conviction 0.76

Buy or hold selectively: the portfolio has meaningful exposure to stocks demonstrating strong positive momentum and substantial relative strength versus a benchmark that declined 0.45% over the last 30 days. The strongest evidence favors JNJ, NVDA, AAPL, META, MSFT, JPM, V, and MA, while recent market conditions remain supportive through lower volatility and continued leadership in semiconductors and technology. Conviction is tempered by weakness in AMZN, GOOGL, HD, and UNH and by conflicting headlines around Apple and Microsoft.

  • Relative strength is broad among several holdings: JNJ gained 8.72% over 30 days versus the S&P 500's 0.45% decline, NVDA gained 11.48%, AAPL gained 5.55%, META gained 4.49%, MSFT gained 2.66%, JPM gained 1.70%, V gained 2.76%, and MA gained 1.44%.
  • Near-term momentum is improving in key growth holdings: NVDA rose 4.34% over five days, META 7.76%, and AAPL 0.98%. NVDA's 11.48% 30-day gain is especially strong evidence of renewed leadership.
  • JNJ combines the strongest holding momentum with defensive positioning: it gained 3.53% over five days and 8.72% over 30 days, while current news continues to frame it as a dominant defensive stock and a stability-oriented alternative.
  • AAPL has a concrete near-term catalyst in its product launch event. Its 5.55% 30-day gain and 0.98% five-day gain indicate that the stock is holding positive momentum despite headlines citing AI supply-chain concerns.
  • The semiconductor complex is providing leadership: NVDA is up 11.48% over 30 days, while notable non-holding candidates MU and SNDK gained 22.55% and 35.09%, respectively. Investor's Business Daily also identified Nvidia, Micron, and Sandisk as showing buy signals.
  • Technology exposure remains supported by broader momentum: QQQ gained 2.70% over 30 days and SPY gained 1.65%, both outperforming the stated S&P 500 benchmark return. This supports maintaining exposure to quality technology leaders rather than retreating entirely to cash.
  • Risk appetite has not broken down: the VIX declined 8.39% over 30 days and 2.61% over five days. That lower-volatility backdrop is constructive for holding equities, even as geopolitical and September-seasonality risks remain.
  • Financial and payments holdings add confirmation beyond technology: JPM gained 1.70% over 30 days, V gained 2.76%, and MA gained 1.44%, with JPM also up 0.74% over five days.
  • The bullish case is selective rather than indiscriminate. AMZN fell 8.98% over 30 days, GOOGL fell 9.33%, HD fell 4.88%, and UNH fell 4.39%; these positions should not be treated as momentum leaders without improvement.
  • News flow is mixed, not uniformly bullish: Microsoft is described as a trending stock but also faces headlines about a 4.5% slide, while Apple has both product-launch optimism and AI supply-chain concerns. Those risks argue for holding strong relative-strength names and adding incrementally rather than making an aggressive all-at-once purchase.
bear case · conviction 0.78

The clearest bearish case is concentrated in the holdings and candidates already showing both negative performance and deteriorating 30-day momentum. AMZN, GOOGL, HD, and UNH are particularly vulnerable: AMZN has the weakest momentum among current holdings at -8.98% over 30 days, while GOOGL, HD, and UNH are each both down from cost and declining over the past month. The candidate set adds further evidence of broad weakness, with APLD (-10.58%), CORZ (-21.64%), and AVGO (-8.75%) showing especially poor momentum. This argues against adding to these names until the deterioration reverses. However, the evidence is incomplete for several profitable holdings, so conviction should remain moderate rather than high.

  • AMZN is a significant holding and has fallen 8.98% over 30 days, the strongest negative momentum signal among the current holdings provided.
  • GOOGL is down 3.84% from cost and has declined 9.33% over 30 days, indicating both existing position weakness and deteriorating recent momentum.
  • HD is down 1.95% from cost and down 4.88% over 30 days; UNH is down 2.07% from cost and down 4.39% over 30 days.
  • The negative signals are not limited to current holdings: APLD is down 10.58%, AVGO 8.75%, and CORZ 21.64% over 30 days, making them poor candidates while momentum remains negative.
  • AMD, CAT, and WMT also have negative 30-day performance, weakening the case for broad exposure to these candidates.
  • The S&P 500 is down 0.45% over the same 30-day period, so several names are materially underperforming the benchmark on the evidence provided.
  • The evidence does not show deterioration for AAPL, MSFT, NVDA, JNJ, JPM, V, MA, or META; therefore, the bearish case is strongest for the specifically flagged names rather than the entire portfolio.
  • The fund-memory evidence highlights recurring exposure to risks including customer losses, demand shortfalls, margin pressure, regulatory and legal developments, technology change, component issues, and financing or receivables stress; these risks reinforce caution toward candidates already showing negative momentum, without establishing a specific new event for any listed holding.
risk case · conviction 0.97

The primary capital risk is concentration: 70.7% of invested assets are allocated to AAPL, AMZN, and JNJ, while 70.2% is concentrated in Information Technology and Consumer Discretionary. This leaves portfolio outcomes highly dependent on a small number of holdings and sectors. The 27.5% cash allocation materially reduces invested exposure and provides liquidity, but also means the portfolio is not fully deployed.

  • AAPL is the largest position at 28.5% of invested assets, creating the largest single-security exposure.
  • AMZN represents 22.4% and JNJ 19.8%; together with AAPL, the top three positions comprise 70.7% of invested assets.
  • Information Technology is 42.1% and Consumer Discretionary is 28.1%, for a combined sector concentration of 70.2%.
  • The portfolio has limited diversification outside the largest exposures: Health Care is 19.9%, Communication Services 7.7%, and Financials 2.2%.
  • Cash is 27.5% of the total portfolio, which reduces exposure to invested assets and provides a liquidity reserve, but leaves 72.5% invested in the concentrated holdings shown.
  • The evidence does not establish a directional advantage or disadvantage relative to the S&P 500; the risk conclusion is based on concentration and allocation structure.

Cash thesis

Cash is 27.5% of assets, slightly above the 25% threshold, and is intentional. It offsets the portfolio's large AAPL, AMZN, and JNJ exposures; preserves flexibility if weak momentum in AMZN and GOOGL persists; and avoids forcing purchases into a market with narrow leadership, geopolitical uncertainty, and mixed sector trends. The opportunity cost is acceptable until either lagging holdings stabilize or a diversified set of high-conviction opportunities emerges.

Full commentary

Bear case response

Accept: AMZN has the weakest 30-day momentum among meaningful holdings at -8.98%, and its size makes this a material risk; no new AMZN purchase is warranted. Accept: GOOGL is below cost and down 9.33% over 30 days, while HD and UNH are also below cost with negative monthly momentum; these holdings should be held only with close monitoring and no averaging down. Accept: APLD, AVGO, CORZ, AMD, CAT, and WMT do not provide sufficiently favorable momentum evidence for new purchases. Accept: the weak holdings materially underperform the stated S&P 500 benchmark over 30 days. Reject the extension of the bearish case to the entire portfolio: JNJ, NVDA, AAPL, META, MSFT, JPM, V, and MA have positive reported 30-day performance, supporting selective retention rather than wholesale de-risking. Accept the rebuttal that semiconductor leadership is uneven and should not be extrapolated into broad tech exposure; therefore, no new semiconductor position is added despite NVDA strength. Accept that falling volatility can reflect complacency and is not, by itself, a buy signal. The appropriate response is disciplined holding, maintained cash, and no repeated turnover absent further deterioration or a verified catalyst.

Market summary

Market conditions are mixed but not broadly risk-off: SPY and QQQ show modest positive 30-day returns, volatility is low with the VIX down over both 5 and 30 days, and leadership persists in NVDA, JNJ, META, and selected semiconductor names. However, dispersion is substantial, with AMZN, GOOGL, HD, UNH, AVGO, APLD, and CORZ showing weak 30-day momentum. Renewed Middle East hostilities, September seasonality, and mixed Apple and Microsoft headlines argue against aggressively deploying cash.

Portfolio assessment

The portfolio remains profitable and has substantial liquidity, but its main issue is concentration rather than a lack of equity exposure. AAPL, AMZN, and JNJ remain the dominant holdings, and AAPL plus AMZN create meaningful technology and consumer-platform sensitivity. Strong relative performance in JNJ, NVDA, AAPL, MSFT, META, and payments supports retaining core winners. AMZN and GOOGL require monitoring because their negative 30-day momentum is material, but there is no verified catalyst or reversal evidence sufficient to justify either averaging down or another immediate reduction. This low-turnover decision follows risk lessons in risk_lesson:run_20260803T191113Z_b5f11a59:report_risk and the prior neutral, cash-preserving approach in report_summary:run_20260826T160411Z_f30a1737.

Risk assessment

The principal risk is concentration: the three largest positions are AAPL, AMZN, and JNJ, while technology and consumer-discretionary exposure are substantial. A correlated reversal in AI, cloud, consumer demand, or large-platform valuations could affect AAPL, AMZN, MSFT, NVDA, and GOOGL simultaneously. AMZN's -8.98% and GOOGL's -9.33% 30-day returns are the clearest active portfolio warnings. Cash moderates total-portfolio drawdown risk but does not remove single-name risk inside the invested book. Risk controls favor no new additions to laggards and no pursuit of extended semiconductor winners.

← September 4, 2026