As the 10-year yield broke above 5%, the market punished long-duration, capital-heavy AI hardware names whose demand depends on cheap financing, yet cybersecurity stocks rallied through the same sell-off because their subscription revenue is recession- an

Verdict: the rotation thesis is intact, but the entry has not fired — wait on NVDA's ADX

The idea's core claim — that security software offers rate-resistant AI exposure while capital-heavy chips do not — is genuinely supported by the fundamentals: CRWD grew fiscal 2026 revenue 111.8% to $4.81B with $1.31B in free cash flow, and NVDA itself runs a 55.6% net margin and 76.3% return on equity, so neither leg lacks financial ballast. The strongest point against is that Palo Alto's latest quarter deteriorated sharply — net margin swung 22.6 points from +16.7% to -5.9% — and insiders at all three names were net open-market sellers totaling roughly $854M in the June 30, 2026 filing period (a delayed ownership record, not a current one). The strongest point for the trade is its disciplined rule structure: the 60-month backtest on this exact system produced 13 trades with a 61.5% win rate and a worst drawdown of about 18.8%, and the system never chases overbought names like CRWD (RSI 70.6) or PANW (RSI 65.2). Right now no entry has fired: NVDA is closest, at $213.9 with RSI (14) of 42.1 already satisfied, but its ADX (14) sits at 15.5 versus the required level above 18. The verdict flips on one number — a sustained move of NVDA's ADX above 18 would arm the setup; insider selling extending into the next filing period alongside further PANW margin erosion would argue for skipping it entirely.

Trade now: NVDA is the only name near a trigger — and it is one indicator away

The rotation is a pullback-buying system across CRWD, NVDA, and PANW: it wants each stock to trade below its 50-day average at some point during the day, then close back above it, with RSI (14) at or below 45 and ADX (14) above 18. As of the latest close, none of the three has fired — but they are at very different distances. NVDA is the closest. It closed at $213.90, about $1.04 below its 50-day average of $214.94, and its RSI (14) of 42.1 already satisfies the sub-45 condition. The missing piece is trend strength: ADX (14) sits at 15.5 versus the required level above 18, roughly 2.5 points short. CRWD and PANW are far away — CRWD closed at $241.36 with RSI at 70.6, and PANW at $375.65 with RSI at 65.2, both deeply overbought rather than pulling back, so any entry there requires a meaningful pullback first. Waiting means sitting flat on all three names and letting NVDA's ADX do the deciding; there is no partial position to take today. Risk on any entry is capped two ways: a fixed 2.5% stop on the position, plus support-based and resistance-based exits, with take-profit set at 5% — an effective reward-to-risk of roughly 2-to-1. Position sizes are capped at 25% per name under a fixed-risk sizing scheme. The evidence base is a completed backtest on this exact rule set: over a 60-month window it produced 13 trades with a 61.5% win rate and a worst drawdown of about 18.8%, and the shorter windows (24 and 12 months) show the same system traded profitably in recent conditions. That supports waiting for the conditions rather than forcing a position today.

Security Software's Cash Machines Are Built for Higher-for-Longer

The idea's core claim — that the market wants AI exposure in a form that survives rates above 5% — is supported by the fundamentals themselves. CrowdStrike delivered $4.81B in fiscal 2026 revenue, up 111.8% year over year, which puts it at the 89th percentile among 788 Information Technology peers for growth. Its free cash flow of $1.31B ranks in the 98th percentile of the sector. That combination — hyper-growth revenue paired with best-in-class cash generation — is precisely what a higher-for-longer environment rewards, because the business does not need cheap financing to keep expanding. Palo Alto Networks reinforces the security-software side of this thesis. For the fiscal year ended July 31, 2025, it generated $9.22B in revenue and $3.47B in free cash flow — placing it in the 99th percentile for cash flow among 791 sector peers. Even with a recent quarterly stumble, it holds a 73.4% full-year gross margin and a 13.5% operating margin, which comfortably beats most software peers and gives it room to fund growth without external capital. On the hardware side, NVIDIA is undeniably strong on paper — $215.9B revenue, up 65.5% year over year, with a 55.6% net margin and a 76.3% return on equity, both elite by any measure. But that strength is exactly what the thesis correctly identifies as the wrong exposure in this regime: it is capital-heavy, dependent on sustained hyperscaler capex cycles, and structurally more sensitive to rate and demand shocks than a subscription security franchise. The thesis is not arguing NVIDIA is weak — it is arguing the security names offer the same AI-adjacent narrative with a fundamentally different balance-sheet character. That distinction is grounded in the numbers, not just narrative. The backtest evidence backs the directional preference. Over 60 months on daily bars, the CRWD pair generated a 253.5% return with a 61.5% win rate across 13 trades, while the more recent 24-month window still delivered a 26.7% return with a 62.5% win rate across 8 trades. That is not a one-quarter fluke — the strategy has been consistently profitable across two materially different market regimes. The 12-month window is thinner (3 trades, 2.8% return) but still positive, which is the key question for a rotation thesis: does the recent regime…

Scores

• Conviction score breakdown: 61
• Thesis support: 72
• Trade readiness: 58
• Risk quality: 62
• Backtest evidence: 60
• Fundamentals trend: 55

Watch items

• NVDA — ADX (14)
• NVDA — Close vs 50-day EMA
• CRWD — RSI (14)
• PANW — RSI (14)
• CRWD — RSI (14)
• NVDA — Insider net open-market value
• CRWD — Gross margin
• PANW — Net margin
• NVDA — Quarterly dividend ex-date