Glasshouse Fund
Decision journal

AI fund decision · September 1, 2026

Maintain a neutral, low-turnover stance: hold existing positions, retain the 25.8% cash reserve, avoid adding to concentrated leaders, and wait for broader confirmation before redeploying capital. Fund memory sources influencing this decision include risk_lesson:run_20260729T184341Z_54100baf:report_risk, risk_lesson:run_20260803T191113Z_b5f11a59:report_risk, trade:run_20260812T183945Z_0d1bf062:AAPL:SELL, trade:run_20260807T183247Z_07bc8571:AAPL:BUY, and report_summary:run_20260826T160411Z_f30a1737.

Outlook: NEUTRALPortfolio $1,038,297Cash 25.8%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 (20 outperform, 14 underperform); 0 became trades. These are the fund's calibration record.

beat SPYlag SPYAAPL: underperform at 61%AMD: outperform at 56%AMZN: outperform at 65%APLD: underperform at 69%ASML: outperform at 59%AVGO: underperform at 54%BE: outperform at 58%CAT: underperform at 64%CORZ: underperform at 78%CRWV: outperform at 57%GLW: outperform at 55%GOOGL: underperform at 59%HD: underperform at 66%INTC: outperform at 56%IREN: underperform at 65%JNJ: outperform at 57%JPM: underperform at 53%MA: outperform at 62%META: underperform at 57%MSFT: outperform at 73%MU: outperform at 68%NVDA: outperform at 69%ORCL: outperform at 61%PG: underperform at 54%QQQ: outperform at 58%SNDK: outperform at 67%SPY: outperform at 50%TSLA: outperform at 62%TSM: outperform at 57%UNH: underperform at 71%V: outperform at 61%WMT: underperform at 63%XOM: outperform at 60%^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
CORZUnderperform SPY78%CORZ's 21.4% 30-day decline and 9.6% five-day decline outweigh partnership-related optimism and signal elevated downside risk.
MSFTOutperform SPY73%MSFT's 29.2% 30-day gain and 3.2% five-day gain make it the portfolio's clearest relative-strength leader, despite extension risk.
UNHUnderperform SPY71%UNH is below cost and down 9.2% over 30 days, with additional negative five-day momentum and margin/governance uncertainty.
APLDUnderperform SPY69%Negative returns over both five days and 30 days indicate weak momentum relative to SPY.
NVDAOutperform SPY69%NVDA's 12.1% 30-day gain and 3.6% five-day advance substantially exceed SPY, supporting continued AI-leadership expectations.
MUOutperform SPY68%MU's 16.8% 30-day gain and positive five-day momentum indicate strong semiconductor-cycle relative strength.
SNDKOutperform SPY67%SNDK's 42.9% 30-day gain and 5.8% five-day advance demonstrate exceptional momentum, albeit with high cyclicality and extension risk.
HDUnderperform SPY66%Weak DIY-demand reports, a 4.8% 30-day decline, and negative five-day momentum indicate weaker consumer-housing relative performance.
AMZNOutperform SPY65%AMZN's 12.5% 30-day gain materially exceeds SPY, and its modest five-day decline appears consistent with consolidation rather than trend failure.
IRENUnderperform SPY65%A positive 30-day return is outweighed by a sharp 12.1% five-day drop, signaling unstable and likely weaker near-term performance.
CATUnderperform SPY64%CAT is down over both the five-day and 30-day periods, indicating weakening cyclical relative momentum.
WMTUnderperform SPY63%WMT's 7.1% 30-day decline and slightly negative five-day return indicate unfavorable relative momentum.
MAOutperform SPY62%MA's 4.7% 30-day gain exceeds SPY and reflects durable payments relative strength despite a modest recent pullback.
TSLAOutperform SPY62%TSLA's 19.7% 30-day return and 5.1% five-day gain support positive momentum, though its volatility makes this a moderate-confidence call.
AAPLUnderperform SPY61%The five-day rebound is constructive, but the -6.8% 30-day decline, CEO-transition uncertainty, and large existing portfolio weight argue for below-SPY performance.
ORCLOutperform SPY61%ORCL's 24.3% 30-day gain and positive near-term momentum support outperformance, though debt-funded AI-build concerns raise volatility risk.
VOutperform SPY61%V's 3.7% 30-day gain modestly exceeds SPY and supports a resilient payments thesis despite recent five-day softness.
XOMOutperform SPY60%Rising oil prices and XOM's positive 5.8% 30-day return support relative performance if energy inflation persists.
^VIXUnderperform SPY60%The VIX is down 18.1% over 30 days, and its negative five-day momentum indicates volatility may remain subdued absent further escalation.
ASMLOutperform SPY59%ASML's 7.1% 30-day gain exceeds SPY, although the recent five-day pullback tempers confidence.
GOOGLUnderperform SPY59%GOOGL's small 30-day gain lags SPY and its negative five-day move, position loss, and cloud free-cash-flow concerns point to modest underperformance.
BEOutperform SPY58%BE's 23.7% 30-day advance is strong relative performance, though its volatility and recent weakness constrain conviction.
QQQOutperform SPY58%QQQ's 6.1% 30-day gain exceeds SPY and ongoing AI leadership remains supportive, although higher yields are a valuation risk.
CRWVOutperform SPY57%The 26.1% 30-day gain reflects strong AI-infrastructure demand, but recent negative momentum and high volatility reduce confidence.
JNJOutperform SPY57%JNJ's defensive quality and flat 30-day performance may compare favorably if yields, oil, and geopolitical risks pressure growth equities.
METAUnderperform SPY57%META's negative 3.6% 30-day return trails SPY, while its small five-day gain is insufficient to establish a confirmed reversal.
TSMOutperform SPY57%TSM's 5.9% 30-day gain exceeds SPY, though a slightly negative five-day return indicates momentum is cooling.
AMDOutperform SPY56%AMD's positive 3.5% 30-day return supports modest relative-strength expectations, though negative five-day momentum limits conviction.
INTCOutperform SPY56%Positive five-day and 30-day returns support a modest relative-strength case, though semiconductor competition limits confidence.
GLWOutperform SPY55%GLW's 18.3% 30-day gain supports relative strength, but execution, customer-financing, and capital-spending risks prevent a high-conviction view.
AVGOUnderperform SPY54%Positive five-day momentum is offset by a negative 30-day return, leaving the relative trend uncertain and slightly unfavorable.
PGUnderperform SPY54%PG's negative 30-day return and flat five-day performance suggest limited relative upside versus SPY.
JPMUnderperform SPY53%JPM remains well above cost, but slightly negative five-day and 30-day returns indicate fading relative momentum.
SPYOutperform SPY50%SPY is the reference benchmark; its modest positive trend is supported by a lower VIX but constrained by yield, oil, and geopolitical risks.

The debate

Bull, bear and risk cases: bull 0.78 · bear 0.90 · risk 0.98
bull case · conviction 0.78

Buy or hold selectively: the strongest evidence favors continued exposure to high-quality technology, AI, and consumer-platform leaders rather than reducing risk broadly. The S&P 500 gained only 1.1% over 30 days, while MSFT rose 29.2%, NVDA 12.1%, AMZN 12.5%, and QQQ 6.1%, demonstrating substantial relative strength. Recent five-day momentum is also positive in MSFT (+3.2%), NVDA (+3.6%), and AAPL (+2.2%). The portfolio has meaningful cash available for disciplined additions, while declining VIX momentum and a still-positive benchmark argue against wholesale de-risking. The case is not uniform: HD, UNH, and AAPL remain below their 30-day levels, and rising yields, oil, and geopolitical tensions are real risks. Those concerns support selectivity, not abandoning the strongest leaders.

  • MSFT is the clearest momentum and relative-strength leader among current holdings: +29.2% over 30 days versus +1.1% for the S&P 500, with another +3.2% over five days. Its current price of $501.24 is also above the $412.76 average cost, and available news discusses its compounding potential and AI cybersecurity positioning.
  • NVDA combines strong relative strength with positive near-term momentum: +12.1% over 30 days and +3.6% over five days, versus the S&P 500's +1.1% 30-day gain. The position is profitable relative to its $203.30 average cost, supporting a hold and measured-add posture rather than selling into strength.
  • AMZN has delivered +12.5% over 30 days, materially outperforming the benchmark, despite a modest -0.5% five-day pullback. Its current price of $254.50 exceeds the $240.96 average cost, so the recent pause looks more like consolidation within a strong monthly trend than evidence of a broken thesis.
  • AAPL has rebounded over the past five days (+2.2%) after a weak 30-day period (-6.8%). The evidence includes a potentially important product catalyst: reporting that the first foldable iPhone is set for release, alongside a price-target increase ahead of the iPhone launch. CEO transition risk and an analyst's stated downside view temper conviction, but they do not negate the catalyst.
  • Payments exposure remains constructive: MA gained 4.7% and V gained 3.7% over 30 days versus 1.1% for the S&P 500. Both positions are meaningfully above their respective average costs, providing evidence of durable relative strength even after modest five-day pullbacks.
  • The broader risk backdrop is mixed but not decisively bearish: the benchmark is positive over 30 days and VIX is down 18.1% over the same period. That is supportive of maintaining equity exposure, while the portfolio's substantial cash balance provides flexibility to buy leaders on weakness rather than chase extended moves.
  • Avoid treating every holding equally. HD (-4.8% over 30 days), UNH (-9.2%), and AAPL (-6.8%) show weaker relative momentum; HD also faces reported weak DIY demand. The bullish action is to emphasize MSFT, NVDA, AMZN, and selectively AAPL/payments, not to add indiscriminately.
bear case · conviction 0.90

The portfolio is exposed to clear pockets of deterioration despite the S&P 500 gaining 1.13% over the last 30 days. Three current holdings are below cost. GOOGL (-4.9%), UNH (-2.6%), and HD (-2.4%), while several larger or strategically important holdings show weakening momentum. AAPL is down 6.8% over 30 days, UNH is down 9.2%, META 3.6%, HD 4.8%, and JPM 0.4%, indicating broad relative weakness rather than isolated underperformance. The main risk is that recent gains in other holdings mask deteriorating leadership and leave the portfolio vulnerable if weakness spreads.

  • AAPL has the strongest downside signal among major current holdings: its 30-day return is -6.75% even though the position remains above its stated average cost, suggesting recent momentum is deteriorating beneath an unrealized gain.
  • UNH is both below average cost (-2.6%) and down 9.18% over 30 days, making it the clearest case of simultaneous mark-to-market loss and accelerating weakness.
  • GOOGL is below average cost by 4.9%, while the S&P 500 gained 1.13% over the same 30-day comparison period; this represents meaningful relative underperformance.
  • HD is below average cost and down 4.83% over 30 days, while META is also losing momentum at -3.55%. These signals weaken confidence that the portfolio's consumer and platform exposures are moving together positively.
  • JPM's 30-day return is slightly negative (-0.36%) despite its position being above cost, so its gain may be vulnerable if recent momentum continues to fade.
  • Several candidate names also carry negative momentum: CAT (-5.16%), WMT (-7.05%), AVGO (-2.77%), and APLD (-4.55%). CORZ is especially weak at -21.43%, making it a high-risk candidate rather than a credible source of diversification.
  • The cautionary GLW evidence lists multiple explicit downside channels, including loss of significant customers, customer financing stress, inability to pace capital spending to demand, margin-execution risk, regulatory and legal developments, technology change, product performance issues, and adverse acquisitions or dispositions. Those risks argue against treating GLW as a low-risk candidate.
  • The portfolio's positive names do not eliminate the signal that several holdings and candidates are weakening while the benchmark is positive; continued concentration in names with fading momentum could produce further relative underperformance.
risk case · conviction 0.98

The primary capital risk is concentrated exposure to a small number of positions and sectors: AAPL, AMZN, and JNJ represent 71.5% of invested assets, while Information Technology and Consumer Discretionary together represent 71.3% of invested assets. The 25.8% cash allocation reduces the amount of capital exposed, but the invested portion remains highly concentrated rather than broadly diversified.

  • AAPL is the largest position at 28.4% of invested assets, creating the largest single-name concentration risk.
  • AMZN and JNJ add 24.1% and 19.0%, respectively; combined with AAPL, the top three positions are 71.5% of invested assets.
  • Information Technology is 41.6% and Consumer Discretionary is 29.7% of invested assets, producing a 71.3% combined sector tilt.
  • MSFT, GOOGL, HD, and NVDA add further material exposure around the largest holdings, so concentration is not limited to only one position.
  • Health Care is also concentrated at 19.2%, principally through JNJ, making JNJ a meaningful single-name and sector exposure.
  • Cash is 25.8% of the portfolio, which provides a substantial uninvested buffer, but 74.2% remains exposed to the concentrated holdings and sector mix.

Cash thesis

Cash is 25.8% of portfolio value, slightly above the 25% threshold. Retaining it is justified by concentrated equity exposure, uneven market breadth, rising-yield and oil-driven inflation risk, and weak relative momentum in AAPL, HD, UNH, GOOGL, and META. The reserve is intentional optionality for clearer broad-market confirmation or materially better entry points, not an uncommitted directional view.

Full commentary

Bear case response

Accept the bear case on concentration and uneven breadth: AAPL's -6.8% 30-day return, UNH's -9.2%, HD's -4.8%, META's -3.6%, and GOOGL's position below cost are meaningful relative-weakness signals versus a positive SPY. Accept that AAPL's recent 2.2% five-day rebound and prospective product headlines do not yet confirm a durable trend reversal, so no AAPL add is warranted. Accept that UNH, HD, and GOOGL warrant monitoring, but reject an immediate sale because their positions are either small (UNH), modestly sized (HD and GOOGL), and there is insufficient evidence today of a thesis break requiring forced liquidation. Reject broad de-risking because MSFT, NVDA, and AMZN have strong 30-day relative performance and remain profitable versus cost, while the declining VIX trend argues against panic. Accept that candidate names such as CORZ, APLD, CAT, WMT, and GLW are not suitable uses of cash given weak momentum, high volatility, or documented execution risks. This supports holding core winners, withholding new capital, and retaining cash rather than rotating aggressively.

Market summary

The market backdrop is mixed: SPY is up 3.5% over 30 days and the VIX is down 18.1%, but September has opened with falling equities, rising yields, higher oil prices, and Middle East escalation concerns. Leadership remains narrow, led by AI and cloud beneficiaries such as MSFT and NVDA, while several consumer, health-care, and platform holdings lag.

Portfolio assessment

Hold all current positions and make no new purchases or sales. The portfolio has strong gains in MSFT, AMZN, JNJ, NVDA, V, and MA, but remains highly concentrated in AAPL, AMZN, and JNJ. AAPL and AMZN together account for roughly 39% of total portfolio value, so deploying cash into existing leaders would worsen an already concentrated structure. Recent fund memory supports low turnover, selective action only after persistent deterioration, and avoiding repeated AAPL trading.

Risk assessment

The principal risk is concentration rather than lack of equity exposure. AAPL, AMZN, and JNJ comprise about 71.5% of invested assets, while technology and consumer-discretionary exposure comprise about 71.3% of invested capital. Strong MSFT, NVDA, and AMZN momentum provides support, but correlated AI/cloud or consumer weakness could affect multiple large positions simultaneously. Rising yields, oil, and geopolitical inflation risk also threaten long-duration technology valuations. Cash cushions total-portfolio volatility but does not diversify the invested sleeve.

← August 26, 2026