TSMC reports into an AI panic where even good news gets sold — short any earnings pop that fades

Verdict: a fade worth waiting for — but it needs an 11% bounce before it can fire

This idea is a conditional, not a forecast: it only shorts if an earnings relief rally stalls above the Bollinger (20) midline — about $602 for SMH and $453 for TSM, both roughly 11% above the current closes — and right now nothing is armed, so the only sensible action is to wait. The strongest point for the trade is the tape context: Bloomberg (October 14, 2026) framed TSMC's results as facing a "high bar" to calm AI jitters, and Reuters (October 15, 2026) reported Asian AI trades crumbling after hot US CPI, which is exactly the sell-the-pop environment the rule set monetizes. The strongest point against is that TSMC itself is the bull case — FY2024 revenue up 33.9% to TWD 2.89 trillion, net income up 36%, free cash flow up 197.9% to TWD 870 billion, and a trailing dividend of $3.85 per share growing about 33% a year — so a genuine re-ignition of the AI trade skips the position entirely and leaves the thesis with nothing. Provisional ownership data through the September 30, 2026 period shows net insider open-market buying of roughly $370K across 161 holders, which does not support the bearish narrative either. The backtest is real but thin — 92.7% cumulative return over 13 trades in five years at a 53.8% win rate and 13.5% maximum drawdown, with exits filled on daily bars (so exit quality is coarse) and no robust nearby-parameter setup established because the sensitivity evaluation ran out of time. If TSMC's print and guidance lift price through the midline and RSI (14) — now 17.1 on SMH and 19.8 on TSM — recovers above 55 and holds, the verdict flips to avoid.

Trade now

Nothing is actionable today: the setup is waiting, not failing. The strategy's entry asks for price **above** the Bollinger (20) midline before it will arm, and both tickers sit well below it. SMH closed at $542.26 against a midline of about $602 — roughly $59.5 (about 11%) short. TSM closed at $408.07 against a midline of about $453 — roughly $44.7 (about 11%) short. The RSI portion of the condition is already satisfied on the downside (SMH RSI (14) at 17.1, TSM at 19.8, both far under the 55 threshold), and the MACD crossover leg is flagged as near on both tickers. In other words, the market has already sold off hard — the fade-the-pop entry is designed for the *next* bounce that stalls, not the current downtrend. What "wait" means concretely: a rally back through the midline on the relevant daily chart (roughly $602 for SMH, $453 for TSM) combined with a fresh RSI (14) cross back above and then below 55, plus a MACD line crossing below its signal line. Until price reclaims that midline, no entry condition can complete, and per the thesis a strong TSMC earnings reaction that reignites the AI trade simply means the trade is skipped. If armed, the risk mechanics are fixed: a 2.4% stop against a 4.8% target, a 2:1 reward-to-risk ratio, and positions capped at 25% of the book each. On the completed 60-month backtest, the SMH-led rule set produced a 92.7% return across 13 trades with a 53.8% win rate and a 13.5% maximum drawdown; the 12-month window added a 4.5% return over 5 trades at a 60% win rate. Exits in that backtest were filled on daily bars rather than intraday data, so treat the reported drawdown and win rate as coarse. One caveat for calibration: parameter-sensitivity evaluation ran out of its time budget, so no robust nearby-parameter setup was established — trade the published rules as written, not adjusted variants.

Selling the AI relief rally: the setup that argues for fading TSMC's pop

The idea's core claim is behavioral, not fundamental: TSMC reports into a market that is actively liquidating AI positions, so good news gets sold. That context is well documented in the cited tape — Bloomberg reported on October 14, 2026 that TSMC results face a high bar to calm global AI stock jitters, and Reuters reported on October 15, 2026 that Asian markets sold off as AI trades crumbled after hot US CPI data. When the marginal buyer has become a forced seller on macro data, the initial earnings pop is increasingly sourced from short covering rather than conviction buying, which is precisely the pattern a fade-the-pop rule set is designed to monetize. The backtest evidence supports the bearish read on event-driven strength. Over the 60-month window, the strategy produced 13 completed trades with a 53.8% win rate and a 92.7% cumulative return, with a maximum drawdown of 13.5%. On the shorter evaluation windows, the 24-month pair returned 20.7% with a single trade, and the 12-month pair returned 4.5% across 5 trades with a 60% win rate. Note that all three windows were traded (status: traded), so this is realized evidence rather than an untriggered watch-list setup. The pattern held across multiple regimes rather than depending on one lucky episode. The macro torque behind the thesis is real in the fundamentals. TSMC's FY2024 revenue grew 33.9% to TWD 2.89 trillion…

Scores

• Conviction score breakdown: 51
• Thesis support: 72
• Trade readiness: 30
• Risk quality: 62
• Backtest evidence: 55
• Fundamentals trend: 38

Watch items

• SMH — Close vs Bollinger (20) midline
• SMH — RSI (14)
• SMH — MACD (12,26,9) vs signal line
• TSM — Close vs Bollinger (20) midline
• TSM — RSI (14)
• TSM — Next earnings report
• TSM — Trailing 12-month dividend per share