TSMC's massive profit isn't enough for greedy investors — short the semiconductor rally

Verdict: the TSMC rejection story is real, but the short setup never armed — wait for confirmation

The verdict: this is a credible short thesis trapped behind an entry that has never once fired, so the call is **wait**, not short. The strongest point for the idea is behavioral — per the Bloomberg report of July 17, 2026, TSMC's 77% profit jump failed to lift the stock and the selloff spread globally, exactly the 'good news fully priced' signal the thesis needs. It is reinforced by net open-market insider selling of roughly $565.4 million at NVDA and $153.2 million at AMD in the filings covering the June 30, 2026 period (note this is a delayed reporting-period disclosure, not a current tally). The strongest point against is that the fundamentals undercut the 'decay' story: NVDA printed $96.7 billion of free cash flow with a 60.4% operating margin and 76.3% return on equity last fiscal year, AMD's net margin actually improved 6.4 points to 19.9% in the quarter ended June 27, 2026, and TSM shows modest net insider *buying* of about $370,000. On the mechanics, the setup is not live: SMH closed at $568.53, still above its $563.60 20-day average, with RSI at 54.0 and ADX at just 4.8 — all far from the required bearish band. Across 60 months of evaluated daily bars (1,234 of them) the entry never triggered, and the parameter work returned no robust recommendation. NVDA is closest: it closed at $218.29, below its $220.74 20-day average, with RSI at 43.9 — but its ADX of 10.0 is half the required level, and the on-balance-volume gate cannot currently be evaluated. What would flip the verdict: a confirmed SMH close below its 20-day average with RSI inside 30–50 and ADX above 20 would arm the setup and change the action from wait to entry. Conviction breakdown follows in the table; composite scoring is computed server-side.

Trade now: the short is armed in theory, but SMH hasn't broken anything yet

The short setup on SMH is not live yet, and the market is moving the wrong way for it. SMH last closed at $516.13, which is $34.19 above the 20-day simple moving average of $481.94 — the entry needs a close at or below that average. RSI (14) sits at 66.2, well above the 50 ceiling, so the momentum part of the entry is also far from triggering. Two of the entry conditions are already satisfied: RSI is comfortably above 30, and ADX (14) at 32.1 is above the 20 floor. The on-balance-volume check cannot be evaluated right now, so treat it as an unknown rather than a met condition. If the entry does fire, the risk framework is fixed: a hard stop at a 2.4% loss on the position, a take-profit at 4.8%, and a forced exit after 15 trading days if neither is hit. That gives roughly 2-to-1 reward-to-risk per position, with no single position allowed to exceed 20% of the book. On current prices, a 2.4% stop from the $482 area would put invalidation near $470, close to the $469.61 resistance level in the level map — meaning the stop sits just below real structure rather than in open air. What 'wait' means concretely: do nothing today. Watch for SMH to close back below its 20-day average while RSI (14) stays between 30 and 50 and ADX holds above 20. Given SMH is 11.2% below its range high but still 227.9% above its range low, the pullback needed is meaningful, not a one-day dip. The idea's own thesis — that TSMC's blowout quarter failed to lift chip stocks because valuations are stretched — argues the selloff resumes; the trend measures say it has not started yet. If the thesis is right, the entry will come to you; forcing a short here means shorting an uptrend with no confirmation. One calibration note: no robust parameter setup was established for this strategy. The research author requested a bounded optimization to surface evaluable history, but it did not produce a nearby-parameter recommendation, and no variants were applied to the live rules — so the published thresholds should be taken as written.

Stretched Valuations Meet Insider Selling: The Short Case Has Fuel

The core of this idea — that even record earnings can't justify today's chip valuations — is anchored in the TSMC story: per the Bloomberg report from July 17, 2026, TSMC failed to impress despite reporting a 77% profit jump, and the selloff spread from Asia to US futures. That narrative sits on top of genuinely extraordinary fundamentals. NVIDIA's most recent fiscal year (ended January 25, 2026) shows $215.9B of revenue growing 65.5% year over year, a 71.1% gross margin, a 60.4% operating margin at the 98.7th percentile of its sector, and a 76.3% return on equity. When a company that dominant only *meets* expectations and the stock falls, the thesis argues the good news was already fully priced — and the market's reaction supports that reading. The ownership data adds teeth to the bearish read. Across the most recent 13F-style reporting period covering 2026-06-30, insider activity at both AMD and NVDA shows net open-market selling: roughly $565.4 million of net insider selling at NVDA and $153.2 million at AMD. Executives at the two flagship names of the AI trade reducing exposure near highs is consistent with the thesis that informed parties see limited upside at these prices. Fundamentals at AMD also give a cautionary angle even as margins improve. AMD's Q2 2026 results (period ended June 27, 2026) improved sequentially — net margin rose to 19.9% from 13.5%, gross margin to 53.8% — but the full-year picture still shows a 12.5% net margin against a 34.3% revenue growth rate in FY2025, meaning valuation, not earnings quality, is doing the heavy lifting in the stock. Finally, SMH itself is a concentrated vehicle: its top ten holdings account for about 69.9% of the fund, with NVDA at 17.8% and TSM at 9.2%. A valuation-driven unwind in just a few names transmits directly into the ETF, so a…

Scores

• Conviction score breakdown: 49
• Thesis support: 65
• Trade readiness: 40
• Risk quality: 60
• Trigger proximity: 25
• Fundamentals trend: 55

Watch items

• SMH — Price vs 20-day SMA
• SMH — RSI (14)
• SMH — ADX (14)
• SMH — Price vs 10-day EMA
• NVDA — ADX (14)
• TSM — RSI (14)
• NVDA — Insider net open-market flow
• AMD — Net margin (Q/Q change)