When: Close below $204 on volume >15M, OR Q3 miss with guidance reset, OR securities fraud inquiry escalation with material disclosure/indictment

Then: Exit to zero

SELL

• IBM is $31 below the 200-day SMA with a confirmed structural downtrend—three failed attempts to reclaim $238 over six weeks
• Volume collapse from 22.5M to 3.9M shares (80% reduction) signals distribution into resistance, not selling exhaustion—buyers refusing to show up at resistance
• Negative tangible book value of $54B means common shareholders receive nothing in liquidation; $3.6B quarterly cash decline is structural, not cyclical
• Sub-one current ratio creates real short-term refinancing pressure despite $12B trailing free cash flow
• The $240M Together AI deal represents only ~0.4% of revenue—a press release, not a transformative catalyst
• The 13.1% post-earnings drift was mechanical mean reversion from oversold conditions, not market pricing AI dominance
• MACD remains at -1.56 (below zero) after 18 positive histogram sessions—recovery signature, not new momentum
• RSI recovered from 29.8 to 51.7 then rolled back to 46.5; the higher RSI low with flat-to-lower price is an unconfirmed bearish setup risk
• The August 26 NVIDIA earnings is an unpriced binary risk that could disappoint and drag AI infrastructure names down
• Reward-to-risk is closer to 2:1, not 3:1, once downside honestly extends to $204-$199.19 (12-14%) instead of the Bollinger lower band at $220
• ATR compressed to $7.86 vs 30-day average of $11.50—current calm understates volatility expansion risk
• Proposed trim of 25-40% into resistance is a measured derisking playbook, not panic; holding full and adding on weakness is cowboy position management

→ vs aggressive: Volume collapse from 22.5M to 3.9M is not 'selling exhausted'—it's buyers refusing to show up and sellers distributing into resistance at $238

→ vs aggressive: The $240M Together AI deal is 0.4% of revenue, not transformative; August 26 NVIDIA print is an unpriced binary risk, not a guaranteed catalyst

→ vs aggressive: $262 upside target is just the rolling 200 SMA in a confirmed downtrend; real upside requires full trend reversal to $280-$300 zone

→ vs aggressive: Downside extends to $204 (July panic low) and $199.19 (absolute low), not $220—making the reward-to-risk closer to 2:1, not 3:1

→ vs aggressive: MACD at -1.56 is still below zero; 18 positive histogram sessions is oversold recovery, not new bull momentum

→ vs aggressive: Adding on weakness with no defined stop below $220 is 'cowboy position management' that turns manageable drawdowns into portfolio-defining losses

→ vs aggressive: The $10.8B pension obligations and negative tangible book value are structural realities, not equivalent to Berkshire's optional profit-generating insurance float

BUY

• 15% trim before NVIDIA print on August 26, retain 85% of position
• Stop at $223 (approximately 1 ATR below current price)
• MACD histogram positive for 18 consecutive sessions; RSI recovered from 29.8 to 51.7, now at 46.5
• ATR compressed at $7.86 versus 30-day average of $11.50
• FCF yield 5.5%; dividend yield 2.87%
• Dividend coverage 1.9x ($12 billion trailing FCF / $6.26 billion annual dividends)
• Institutional sentiment mildly bullish at 6.1 out of 10
• Together AI deal valued at $240 million (0.4% of revenue)
• Key resistance at $238; key support at $204 (panic low) and $199.19
• After NVIDIA print: add 5-10% if price breaks above $238.50 on volume > 6 million shares
• Exit remaining position fully if price breaks below $220 with expanding volume
• Volume contraction from 22.5 million shares to 3.9 million shares
• Total debt $65 billion against $7.2 billion cash
• $170 billion of negative tangible book value attributable to treasury stock from buybacks
• Fed funds pricing at 86% probability of zero cuts in 2026
• Recession probability at 8%
• Reward-to-risk asymmetry corrected to approximately 1.3 to 1.5:1 (not 3:1 as aggressive analyst claimed)

→ vs aggressive: Volume contraction could be distribution into resistance, not supply exhaustion; neither interpretation has higher evidentiary weight

→ vs aggressive: Sellers do exhaust at lower prices (capitulation), and July 14 gap-down is a textbook example of that dynamic

→ vs aggressive: Berkshire Hathaway comparison overreached—IBM's $65 billion debt against $7.2 billion cash creates refinancing pressure unlike Berkshire's optional float capital

→ vs aggressive: Reward-to-risk math cherry-picked using $220 Bollinger band instead of structural support at $204/$199.19, collapsing true asymmetry to 1.3-1.5:1 from claimed 3:1

→ vs aggressive: Volatility compression at current ATR of $7.86 is not a directional signal; using it to size positions ignores that 30-day average is $11.50

→ vs conservative: 25-40% trim too aggressive given MACD bullish for 18 consecutive sessions and RSI recovery from 29.8 to 51.7

→ vs conservative: Together AI deal dismissal too dismissive—strategic positioning as credible AI inference vehicle for enterprise customers has cumulative significance alongside TD SYNNEX, quantum roadmap, and US Open

→ vs conservative: Negative tangible book mischaracterized—$170 billion of the reduction is treasury stock from buybacks (capital return), not operational asset destruction or distress

→ vs conservative: Two-tranche execution plan adds operational complexity without clearly improving risk-adjusted outcome; creates execution risk from multiple small decisions