When: Q3 13F filing (mid-November)

Then: Confirms or refutes Berkshire trimming overhang

REDUCE EXPOSURE SIGNIFICANTLY; avoid reload until structural reversal confirmed; hedge remaining exposure

• Trim direction is correct but one-third (33%) is insufficient given convergence of three signals: deteriorating fundamentals relative to peers, broken technicals (200 SMA violation), and credit cycle normalization
• Stop loss at $335.63 on a position trading at $321.80 is effectively a re-entry trigger, not a protective stop; recommend no re-entry until structural change rather than tactical bounce
• 34% ROE is built on cyclical lows in net charge-offs; a 100 basis point normalization on the $225B loan book equals $2.25B pre-tax impact (~18% of TTM net income)
• Buffett trimming at ~19x earnings is a synchronous signal with technical breakdown and Q2 earnings miss (nearly 6% below consensus)
• Q2 earnings confession indicates the peer gap is structural rather than cyclical, not a temporary setback
• Polymarket recession odds at 8% represent the kind of consensus comfort that historically precedes credit normalization
• Recommend 40-50% trim instead of 33% to manage tail risk from coordinated three-direction thesis assault
• Remaining AXP exposure should be hedged through paired trades or sector-level puts given 1.05 beta amplifies idiosyncratic credit risk
• Re-entry bar: close above 200 SMA on improving breadth with RSI divergence and Q3 earnings that close gap to peers
• Cost of over-trimming is manageable forgone upside; cost of under-trimming into credit cycle rollover (15-20% drawdown) is not acceptable for stability-focused portfolio

→ vs aggressive: Reload mentality at $305 is not conservative; hoping for a second chance at the same name with the same risk profile is trading behavior, not prudent capital management

→ vs aggressive: Credit cycle tail risk deserves far more weight than acknowledged; 100bps normalization on charge-offs = $2.25B pre-tax impact (~18% of TTM earnings), which the market is not pricing

→ vs neutral: Fair value at 16x forward earnings is not a floor when the multiple is being defended by decelerated earnings that missed consensus by nearly 6%

Reduce exposure by 33% via trim at $321.80; maintain remaining position; re-enter on tiered price levels ($335, $345) or confirmed close above $340.74; no hedges needed as trim serves as the hedge

• 33 percent trim is the right magnitude—meaningful risk reduction while preserving upside participation if the bear case is wrong or delayed
• Q2 delivered record revenue of $19.64 billion and record EPS of $4.53 against a peer miss—this is a franchise firing on all cylinders but facing multiple compression, not a thesis breakdown
• Loan book grew 1.6 percent quarter over quarter and net interest income stabilized around $4.6 to $4.7 billion per quarter
• The 12 percent upside to 335 versus 16 percent downside to 305 asymmetry the aggressive case sells ignores time, holding cost, and the conditional probability of a tactical mean reversion bounce given lower Bollinger Band tag with RSI at 35
• Bounce probability of 40 to 50 percent over the next two to four weeks materially alters the risk-adjusted math on the trim
• A stop at $335.63 is too tight as a re-entry trigger because $335.63 sits on the 200 SMA which acts as a magnet rather than a clean break level—recommend wider zone at $340.74 or confirmed close above the 50 SMA
• Polymarket shows 93 percent probability of zero Fed cuts in 2026, structurally positive for AXP's net interest income on the cardholder loan book
• Polymarket has recession odds at 8 percent, which does not support the conservative analyst's worst-case 100 basis point charge-off scenario as the central case
• The affluent customer base historically demonstrates more resilience than subprime or prime card books during prior credit normalizations
• Q3 earnings in October is the real decision point; everything between now and then is positioning around an outcome the data does not yet allow to be called with high confidence

→ vs aggressive: Reload mentality is a market timing call dressed up as risk management—the 'asymmetric prize' framing assumes $305 is a foregone conclusion floor and ignores the possibility of a two-quarter basing pattern, sideways chop, or the stock ripping back to $345 before ever touching $305

→ vs aggressive: The $335.63 stop is not an 'invitation to be shaken out'—it functions as a re-entry trigger if the technical structure reverses, which has value by pre-committing to adding exposure on confirmed strength rather than hoping for a lower entry that may never come

→ vs conservative: A 100 basis point charge-off rate move is not supported by current data—NCOs are creeping higher directionally but this implies a charge-off rate back near 2019 to 2020 cycle peaks, requiring either a recession or credit-specific blowup that the data is not yet flagging

→ vs conservative: The 'no re-entry until structural turn' instruction is a recipe for permanent underweight in a name that has compounded at 470 percent over a decade; a bar requiring confirmed full reversal above all moving averages is prohibitively high and locks in opportunity cost

→ vs conservative: Adding puts on top of a 33 percent trim on a name with a 1.05 beta is double-counting the risk and creates drag on remaining exposure during a window where the bounce risk is real