Mag7 Desk Note #30

Executive Takeaway

Nothing on this tape pays for beta. Hold the NVDA/AMZN core, keep MSFT static into RSI 78, own GOOGL as valuation not trend, leave the three sub-EMA20 names as stubs, and express any new risk through long optionality where IV sits below realized across all seven.

Positioning

• No directional edge: all seven NEUTRAL, nothing above 1.01x RVOL. The EMA20 split is a sizing input, not a signal.
• NVDA and AMZN are the only intact, non-extended trends, and both clear comps (AMZN +7.5pp vs WMT/COST; NVDA ~11pp ahead of AMD, level with AVGO).
• MSFT has the best structure (+14.0%, +10.1pp vs GOOGL/AMZN) and the worst entry at RSI 78.1 — hold, don't add.
• AAPL/META/TSLA stay stubs; all three lag their peer sets, TSLA at the 15th percentile, META at the 26th.

Watch

NVDA holding its +7.7% EMA20 trend into earnings — the only name with intact structure and non-trivial conviction.

What Changed

Tape, options, COMPS, insider and short-interest data are unchanged; every headline is [SEEN], so nothing is newly published. The one real delta is the vs-SPY column repopulating with mixed-session provenance and flipping signs. Claude flagged it but minimized it; OpenAI and Gemini did not surface it at all.

Portfolio

All three books are now effectively the same: ~27% cash, NVDA $210 and AMZN $190–195 together roughly half of NAV, MSFT capped at $60, GOOGL $90, $10 stubs in AAPL/META/TSLA. Claude is marginally more invested ($215 cash, AMZN $195); OpenAI and Gemini are identical. The risk here is concentration, not cash discipline — and GOOGL's weight in every book contradicts COMPS, which has it lagging META/MSFT by 6.0pp.