When: Post-Q3 stock reclaims $1,012 (50 SMA)

Then: Restore toward full weight on 50 SMA reclaim

TRIM 25-30% OF POSITION GRADUALLY AT 913-952; MAINTAIN 70% CORE POSITION WITH STOP AT 970

• 13.86 trailing P/E, 16.9% ROE, 187B cash, and 100B tangible book provide bounded downside floor that prevents unlimited impairment
• Q2 26 revenue of 20.3B (up 39% YoY), NII of 3.95B (up 27% YoY), and 42.2% operating margin reflect genuine operating leverage, not just tax-rate effects
• Even with normalized tax rates, GS is earning north of 16 dollars per share on a run-rate basis; at 12.4x earnings this is not a peak multiple
• MACD histogram compressing from negative 9.18 to negative 3.01 and ATR dropping 23% from 29.42 to 22.70 signal selling exhaustion, not continuation
• Polymarket pricing 78% probability that GS leads the OpenAI IPO represents a real binary catalyst not yet fully priced
• Technical breakdown has occurred but selling pressure is fading per multiple indicators; trim into rallies not vacuums
• Sentiment at 5 out of 10 with zero retail and zero institutional coverage means next data point will move stock violently

→ vs aggressive: Single high-conviction timing bet around Q3 earnings catalyst is speculation, not risk management; what if earnings come in line or better? What if NII is stronger than expected, AFS marks were already telegraphed, or credit provisioning is lower than feared?

→ vs aggressive: Fortress balance sheet does matter; a bank earning 16.9% ROE on 2.13T in assets does not trade at six times earnings indefinitely; aggressive analyst conflates technical correction with fundamental impairment

→ vs aggressive: 12 times forward earnings cuts both ways; most sell-side analysts will cut estimates by only 5-15%, not 30%, leaving forward earnings north of 60 dollars; at 12x that you are at 720 with tangible book supporting equity

→ vs neutral: The 'do nothing' crowd is equally dangerous; sentiment at 5 out of 10 with zero coverage means the next data point will move the stock violently, so you do not want to be a full holder when consensus catches up to the technical breakdown

Trim 25 to 30 percent of GS position in tranches toward 913 to 952 zone over multiple sessions; stop at 970; core position preserved at 70 to 75 percent of benchmark weight; re-entry rule on sustained close above 952; additional trim on confirmed breakdown below 881 with volume.

• Selling pressure is exhausted: MACD histogram improved from negative 9.18 to negative 3.01, ATR compressed from 29.42 to 22.70, lower Bollinger Band flattening and rising
• Q2 2026 was a peak capital markets quarter (record equity tape, OpenAI IPO process ramping, elevated IB activity) not a steady-state earnings level; revenue grew 39 percent YoY, NII grew 27 percent, operating margin expanded to 42 percent but Q3/Q4 may be lower
• OpenAI IPO is a real asymmetric catalyst with 78 percent probability and 2028 resolution, not a fully priced tail event
• Fortress balance sheet provides bounded downside (no zero, no 4x earnings) but does not prevent 10 to 15 percent further decline; price went from 1148 to 897 with balance sheet intact
• Pre-earnings quiet is not actionable directional signal; does not indicate institutional positioning direction
• Post-earnings re-entry rule: sustained close above 200 SMA at 952; post-earnings additional trim rule: breakdown below 881 toward lower Bollinger Band at 857
• Stop at 970 above the 200 SMA defines structural invalidation of bearish view

→ vs aggressive: Treating multiple bearish datapoints as certainty rather than higher probability creates narrative not analysis; aggressive analyst wants to sell down here but exhaustion signals indicate scale out into bounces because demand is returning

→ vs aggressive: Calling OpenAI IPO a 2028 tail event that's already in the multiple understates the asymmetric upside; 78 percent probability on a marquee tech IPO with 2028 resolution is a real binary catalyst being ignored while asymmetric downside is emphasized

→ vs conservative: Using balance sheet strength as a shield against price reality is misleading; bounded downside does not mean stock cannot go down another 10 to 15 percent; stock went from 1148 to 897 with fortress balance sheet in place

→ vs conservative: Anchoring on peak Q2 2026 earnings as steady-state is the mirror image of aggressive analyst's error; if Q3 revenues come in at 17 to 18 billion instead of 20 billion the conservative framework also breaks down

→ vs conservative: Overweighting bullish signals because they fit the constructive narrative is the mirror image of aggressive analyst's dismissal of bullish signals; honest read is mixed signals arguing for measured execution not aggressive positioning in either direction

→ vs conservative: Pre-earnings quiet tells us nothing actionable about Q3 being good or bad or institutional positioning direction; reading directional implications from silence is overfitting

→ vs neutral: Both analysts miss explicit post-earnings decision rules for both upside and downside outcomes; the neutral analyst fills this gap with re-entry above 952 and additional trim below 881