Glasshouse Fund
Decision journal

AI fund sells AMZN · September 11, 2026

Maintain a neutral, low-turnover posture: trim 50 AMZN shares to reduce a large weakening exposure, hold relative-strength leaders without adding, avoid falling watchlist names, and retain the elevated cash reserve until market breadth improves.

Outlook: NEUTRALPortfolio $1,042,341Cash 28.8%Grounding: ok

Trades

SELL 50 AMZN74% · filled at $257

Trim approximately $12.8k of a large position after AMZN fell 8.2% over 30 days and 2.6% over five days, materially lagging SPY. The position remains above average cost, allowing a measured concentration reduction without abandoning the long-term holding.

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 (20 outperform, 14 underperform); 0 became trades. These are the fund's calibration record.

beat SPYlag SPYAAPL: outperform at 67%AMD: outperform at 56%AMZN: underperform at 71%APLD: underperform at 74%ASML: underperform at 58%AVGO: underperform at 68%BE: outperform at 61%CAT: underperform at 57%CORZ: underperform at 76%CRWV: underperform at 54%GLW: outperform at 55%GOOGL: underperform at 66%HD: underperform at 73%INTC: outperform at 57%IREN: outperform at 56%JNJ: outperform at 59%JPM: underperform at 53%MA: outperform at 54%META: outperform at 72%MSFT: outperform at 52%MU: outperform at 63%NVDA: underperform at 58%ORCL: outperform at 56%PG: underperform at 55%QQQ: outperform at 51%SNDK: outperform at 69%SPY: outperform at 50%TSLA: outperform at 65%TSM: outperform at 55%UNH: underperform at 65%V: outperform at 55%WMT: underperform at 59%XOM: outperform at 66%^VIX: outperform 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 SPY76%The 17.3% 30-day loss and negative 5-day return indicate pronounced downside momentum.
APLDUnderperform SPY74%Negative 5-day and 11.8% negative 30-day momentum provide no confirmed reversal signal.
HDUnderperform SPY73%The 13.4% 30-day decline, 4.8% five-day decline, and loss versus cost show persistent relative weakness.
METAOutperform SPY72%The strongest combined core-holding momentum, up 8.8% over 30 days and 4.5% over five days, supports outperformance.
AMZNUnderperform SPY71%The 8.2% 30-day decline and negative 5-day momentum indicate persistent relative weakness versus SPY.
SNDKOutperform SPY69%Exceptional 30-day momentum remains favorable versus SPY despite a negative recent week and elevated volatility.
AVGOUnderperform SPY68%A 15.6% 30-day decline overwhelms the slightly positive 5-day move and signals unresolved weakness.
AAPLOutperform SPY67%Positive 5-day and 30-day momentum, plus recent product announcements, support relative strength versus SPY despite concentration risk.
GOOGLUnderperform SPY66%Negative 5-day and 30-day momentum, combined with a position below cost, favor continued lagging versus SPY.
XOMOutperform SPY66%Positive performance over both horizons, supported by energy strength, indicates favorable relative momentum.
TSLAOutperform SPY65%Positive momentum over both five and 30 days supports relative leadership, though volatility is high.
UNHUnderperform SPY65%Negative returns over both periods and a loss versus average cost indicate unresolved relative weakness.
MUOutperform SPY63%A strong positive 30-day return supports relative outperformance despite the recent pullback.
BEOutperform SPY61%Positive 5-day and strongly positive 30-day momentum indicate favorable relative strength, albeit with high volatility.
^VIXOutperform SPY60%Rising volatility over five and 30 days supports continued relative strength in the volatility index while market uncertainty persists.
JNJOutperform SPY59%Positive 30-day performance and a substantial gain versus cost support relative resilience, though the recent pullback limits conviction.
WMTUnderperform SPY59%Negative 5-day and 30-day returns point to weak current relative momentum.
ASMLUnderperform SPY58%Negative returns over both 5 and 30 days suggest modestly weaker relative performance than SPY.
NVDAUnderperform SPY58%Negative returns over both periods, including a 5.1% five-day decline, point to near-term relative weakness.
CATUnderperform SPY57%Negative short- and medium-term momentum leaves CAT slightly less attractive than SPY.
INTCOutperform SPY57%Positive 5-day momentum and a near-flat 30-day result compare favorably with SPY's negative 30-day return.
AMDOutperform SPY56%Strong 5-day momentum and a positive 30-day return contrast with SPY weakness, though semiconductor volatility limits conviction.
IRENOutperform SPY56%The positive 30-day return supports relative strength despite a negative recent week.
ORCLOutperform SPY56%Positive 30-day performance is favorable relative to SPY, although negative five-day momentum warrants caution.
GLWOutperform SPY55%Strong 5-day momentum may support relative performance, but the negative 30-day result and identified business risks constrain confidence.
PGUnderperform SPY55%Negative five-day and 30-day performance offer no current relative-strength confirmation.
TSMOutperform SPY55%Positive 30-day performance and near-flat five-day action compare favorably with SPY's decline.
VOutperform SPY55%The positive 30-day return versus SPY supports modest outperformance potential despite a negative five-day move.
CRWVUnderperform SPY54%Slightly negative returns over both measured periods offer limited evidence of relative strength.
MAOutperform SPY54%A modestly positive 30-day return versus SPY's decline supports a narrow relative-strength case despite recent weakness.
JPMUnderperform SPY53%Negative returns over both five and 30 days imply slight relative weakness despite gains versus portfolio cost.
MSFTOutperform SPY52%Slightly negative momentum is less severe than SPY over 30 days, but the trend is not strong enough for high conviction.
QQQOutperform SPY51%QQQ's 30-day decline is marginally less severe than SPY's, but negative short-term momentum makes this a low-conviction call.
SPYOutperform SPY50%SPY is the comparison benchmark, so its relative direction is assigned neutral-equivalent confidence.

The debate

Bull, bear and risk cases: bull 0.76 · bear 0.88 · risk 0.98
bull case · conviction 0.76

Buy or hold selectively: the portfolio has several proven relative-strength leaders outperforming a weak market, while current catalysts support continued exposure. The S&P 500 is down 2.0% over 30 days, yet AAPL is up 4.3%, META 8.8%, JNJ 3.2%, V 1.5%, and MA 0.4%. AAPL also gained 2.1% in five days following the iPhone 18 Pro and watch announcements, while META is accelerating with 4.5% five-day momentum. This leadership, combined with substantial cash of approximately $287,000 for flexibility, supports holding winners and selectively adding rather than broadly reducing risk.

  • AAPL is a clear relative-strength leader: up 4.3% over 30 days and 2.1% over five days versus the S&P 500's 2.0% decline, with a fresh product catalyst from the iPhone 18 Pro and watch announcements.
  • META has the strongest momentum among the core holdings shown, rising 8.8% over 30 days and 4.5% over five days. That combination indicates both sustained and near-term leadership.
  • JNJ is up 3.2% over 30 days despite a 3.2% five-day pullback, remains well above the portfolio's $239.58 average cost at $265.80, and recently appeared among stocks hitting 52-week highs. The evidence supports holding, though the rally may be becoming fully priced.
  • Payment networks retain constructive relative strength: V is up 1.5% over 30 days and MA is essentially positive at 0.4%, while both positions are materially above their average costs. Their recent weakness looks more like consolidation than confirmed trend failure based on the supplied data.
  • MSFT, JPM, NVDA, and GOOGL have weaker recent momentum, but their inclusion in a diversified portfolio and the availability of cash argue for patience and selectivity rather than forced selling. The evidence does not support adding aggressively to these names today.
  • The market backdrop is volatile, with the VIX up 19.7% over 30 days, but that volatility also creates an opportunity to deploy cash into demonstrated relative-strength names instead of chasing broad-market exposure.
  • The bull case is not uniform: AMZN, HD, UNH, and GOOGL are down 30 days, while NVDA is down 5.1% over five days. Therefore, the strongest action is to concentrate new buying on AAPL and META and maintain disciplined holds elsewhere.
bear case · conviction 0.88

The evidence argues for a defensive stance: broad market momentum is negative, several portfolio holdings are already losing money, and a number of core holdings are showing materially worse 30-day performance than the S&P 500. The weakest candidates are even more vulnerable, with steep recent declines and limited evidence of stabilization.

  • The S&P 500 is down 2.0% over 30 days, while several holdings are deteriorating substantially more: HD is down 13.4%, AMZN 8.2%, GOOGL 6.1%, and UNH 4.6% over the same period.
  • HD, UNH, and GOOGL are already below their portfolio average costs by approximately 5.5%, 6.7%, and 3.3%, respectively. Their negative 30-day momentum suggests the drawdowns are not yet showing clear signs of reversal.
  • AMZN is a material downside concern: despite being held at a sizable share count, it has the worst 30-day momentum among the larger technology and consumer holdings listed, at -8.2%.
  • GOOGL combines an existing portfolio loss with -6.1% 30-day momentum, making it one of the clearest cases where current weakness and recent trend deterioration reinforce each other.
  • HD's -13.4% 30-day performance is the sharpest decline among the listed current holdings with momentum data, substantially exceeding the market's -2.0% decline and indicating pronounced relative weakness.
  • UNH is also a dual signal: it is down approximately 6.7% versus average cost and down 4.6% over 30 days. The evidence provides no indication that the negative trend has stabilized.
  • MSFT, NVDA, and JPM are not currently below average cost in the snapshot, but all have negative 30-day momentum: -1.3%, -2.4%, and -1.1%, respectively. This raises the risk that recent gains in otherwise profitable positions could erode if market weakness persists.
  • The candidate set contains several significantly weaker setups than the portfolio: CORZ is down 17.3%, AVGO 15.6%, and APLD 11.8% over 30 days. Buying into these declines would add exposure to demonstrated downside momentum rather than confirmed recovery.
  • ASML, CAT, CRWV, GLW, INTC, PG, QQQ, SPY, and WMT also show negative 30-day momentum. The breadth of declines across technology, industrials, consumer, and index exposures argues against assuming weakness is isolated to one sector.
  • The cautionary Corning evidence explicitly identifies risks including customer losses, customer financing and receivables problems, technology change, component performance, regulatory and legal developments, and the difficulty of pacing capital spending to demand. Although GLW is not held, these risks weaken the case for treating it as a defensive candidate while its 30-day momentum is already negative.
  • The evidence does not show a positive momentum offset for the holdings with gains versus average cost. Given the negative benchmark trend and broad list of fading names, profitable positions should not be treated as immune to further downside.
risk case · conviction 0.98

The primary capital risk is concentration: three positions represent 71.1% of invested assets, while Information Technology and Consumer Discretionary together account for 70.8% of invested exposure. The 27.5% cash allocation provides a buffer against losses but does not eliminate the material single-name and sector concentration in the invested portion.

  • AAPL is the largest position at 29.7% of invested assets, followed by AMZN at 22.3% and JNJ at 19.1%; together they comprise 71.1% of invested assets.
  • The top five positions. AAPL, AMZN, JNJ, MSFT, and GOOGL, represent 87.0% of invested assets, leaving limited diversification among securities.
  • Information Technology is 43.0% and Consumer Discretionary is 27.8% of invested assets; their combined 70.8% creates a substantial sector tilt.
  • Health Care contributes another 19.3%, primarily through JNJ at 19.1% of invested assets, so sector diversification is not broad despite the presence of several smaller holdings.
  • Cash is 27.5% of the portfolio, which reduces total invested capital at risk but also means the portfolio's invested exposure is concentrated within the remaining 72.5%.
  • The benchmark's 30-day return is -2.0%, but no portfolio return is provided, so relative performance or realized tracking risk cannot be assessed from this evidence.

Cash thesis

Cash is 27.5% of portfolio value and is intentionally retained. It cushions the portfolio against concentrated single-name and technology/consumer-discretionary risk, preserves optionality during elevated volatility, and avoids forcing deployment into watchlist names with unconfirmed reversals. The reserve is high, but justified until market breadth and relative strength broaden beyond a few leaders.

Full commentary

Bear case response

Accept: the benchmark trend and breadth are weak, and HD (-13.4%), AMZN (-8.2%), GOOGL (-6.1%), and UNH (-4.6%) have underperformed SPY over 30 days. Accept: HD, GOOGL, and UNH are below average cost without supplied evidence of trend stabilization, so no additions are warranted. Accept: AMZN is a material risk because of its size and sharply negative recent momentum, although it remains above cost; this supports a modest trim rather than an exit. Accept: MSFT, NVDA, and JPM have negative 30-day momentum despite unrealized gains, so they should be held without additions. Accept: deeply negative watchlist momentum in CORZ, AVGO, and APLD argues against dip-buying. Partially reject the implication that all profitable leaders should be reduced now: AAPL, META, JNJ, V, and MA have better relative performance than SPY, and there is insufficient evidence of a confirmed trend break in those holdings. Holding them while reducing one concentrated laggard is more consistent with low-turnover discipline.

Market summary

The broad market remains unsettled: SPY is down 2.0% over 30 days and 1.6% over five days, while the VIX has risen 19.7% over 30 days. Inflation, bond-yield, and oil-price headlines reinforce a higher-volatility backdrop. Leadership is narrow, led by AAPL and META, while several portfolio holdings and many watchlist names have negative 30-day momentum.

Portfolio assessment

The portfolio has retained meaningful gains in AAPL, JNJ, MSFT, AMZN, payment networks, and META, but invested capital remains highly concentrated in AAPL, AMZN, and JNJ. AAPL's current strength supports holding rather than adding because it is already the largest position. AMZN's -8.2% 30-day decline is concerning given its large weight; a modest trim reduces concentration and preserves capital for clearer opportunities. No new purchases are justified while breadth remains weak.

Risk assessment

The principal risk is concentration: AAPL, AMZN, and JNJ dominate invested assets, with substantial correlated technology and consumer-discretionary exposure. AAPL and META leadership does not offset the weakness in AMZN, HD, GOOGL, NVDA, and UNH. Rising volatility and weak benchmark momentum increase the risk that profitable core positions could give back gains. The incremental AMZN trim follows fund memory emphasizing early, measured risk reduction in weakening concentrated names rather than reactive wholesale selling.

← September 10, 2026 September 14, 2026 →