• Type: refresh (subtype: hypothesis-scoring) — first numeric scoring of A/B/C against the post-debut tape
• Source: code-computed precompute summaries (summaries{space,ai-infra,ai-power,ai-infrastructure,ai-scan,semis,memory,etf-ideas,geopolitical,holdings-stocks,drone-defense,spacex-s1-supply-chain}.json`, 2026-06-16 close) + 30d (3 bars: 6/12, 6/15, 6/16).

Calendar note: README/calendar labels today as "T+4" counting calendar days from the 6/12 debut. By trading sessions this is T+2 (6/12 debut → 6/15 → 6/16). This entry uses the trading-session count ("T+2") for the price read; both refer to the same date. The 2026-06-16 close is the second full session after debut.

Event truth vs price truth (discipline up front): Everything below is price action only. None of it proves the IPO caused any cohort move — markets move for many reasons, and the scan window overlaps a broad mega-cap drawdown that predates and is independent of SPCX. Where the tape is consistent with a hypothesis I say so; I do not claim causation.

SPCX trajectory since debut (the anchor fact):

• Offering (assumed, per S-1/A priced cover): $135. First print 6/12 open $150.
• 6/12 close $160.95 (intraday high $176.52) — +19.2% vs offering.
• 6/15 close $192.50 — +42.6% vs offering.
• 6/16 close $201.80 (intraday high $225.64) — +49.5% vs offering.
• Debut→6/16: +25.4% over two sessions; strong, accelerating bid, no sell-the-news fade in SPCX itself. Volume 519M (debut) → 256M → 315M.
• summary fields: SPCX rsi/vsSma20 = null (insufficient history; only 3 bars — no SMA20/RSI computable yet); pctFrom52wkHigh -10.57% is meaningless at 3 bars and should be ignored.

Hypothesis scoring

Hypothesis A — Liquidity drain → NOT-FIRED (at cohort level). The cleanest pre-identified drain tell (set 2026-06-04) was semis / memory rolling below SMA20 in the post-IPO window. It did the opposite:

• memory (8 names): 0/8 below SMA20, median vsSma20 +10.3%, median 7D +13.4% — the most-stretched cohort melted up through the IPO window (WDC +25% 7D / vsSma20 +26.7%, STX +17%, MU +14%, DRAM +16%). This is the precise opposite of the drain signature.
• semis (15 names): only 5/15 below SMA20, median vsSma20 +1.1%, median 7D +7.9%. The five below-SMA20 names (NVDA −2.7%, AVGO −9.2%, QCOM −4.0%, ON −1.3%, MXL −5.3%) are explainable by name-specific weakness (NVDA/AVGO mega-cap drawdown) while ARM (+23.7% 7D), AMKR (+15.7%), AMAT (+14.0%), MRVL (+14.5%), INTC (+16.8%) ripped. No coherent cohort roll-over.
• AI-power / BTC-miner cohort (the "what gets sold" list): all green. CRWV +4.8% 7D (vsSma20 +10.4%, strong-up), CORZ +3.2% (vsSma20 +4.9%), IREN +6.7% (vsSma20 −0.2%), APLD +8.3% (vsSma20 +4.7%). CRWV — the lone pre-IPO softness and most Musk-orbit-exposed name — recovered into the IPO rather than leading a drain.
• NBIS +15.0% 7D / +32.6% 30D, RSI 65.2, vsSma20 +15.5% — strongest name in the whole complex, ran straight through the event.
• Rotation pair-watch did NOT fire either way: gas/value trio flat-to-down (DTM +0.2% 7D, KMI +1.2%, AR −1.6%), so no visible "money landing in laggards" leg; and the SMA20 drain leg didn't fire, so the paired rotation tell is null on both sides.
• Verdict: at T+2 the AI-infra silicon cohort is firmer, not weaker. The mechanical-flow drain thesis (always the lower-conviction sentiment-mediated leg) shows no tape support. NOT-FIRED — consistent with the 2026-05-26 base-rate revision that the $75B raise is below the SPY-level noise threshold.

Hypothesis B — SPCX-as-new-AI-infra-comp → INDETERMINATE (lean: hot SPCX lifting, not compressing, the cohort).

• The pre-stated tell was directional correlation with SPCX, not absolute direction. SPCX ripped +25% over two sessions; the AI-infra cohort moved with it, not against: NBIS +15% 7D, IREN +6.7%, APLD +8.3%, CRWV +4.8%, CORZ +3.2% all green into the same window.
• This is consistent with the "SPCX prices hot → cohort lifts on relative-cheapness" branch of B and inconsistent with the "cheap anchor → multiple compression" branch. I see no sign of NBIS/CRWV compression vs SPCX — the opposite if anything.
• BUT this is co-movement in a risk-on space/AI tape, not identified comp re-rating. No analyst-note / fund-positioning language captured this window; correlation-vs-common-beta is unresolvable from two sessions of price. INDETERMINATE on the mechanism; the directional sign is favorable to the cohort, not threatening. Needs the deep-dive's analyst-language sweep to move off INDETERMINATE.

Hypothesis C — Halo, then air out → CONFIRMED (the air-out leg held; no debut-day reversal bounce).

• The pre-IPO sell-the-news unwind (flagged 6/4) continued through and after the debut rather than bouncing on IPO excitement:
  • LUNR 7D −9.7% (30D −30.5%, vsSma20 −30.1%, RSI 38.3), RDW −10.0% 7D (vsSma20 −26.5%), FLY −5.1% 7D (30D −29.6%, vsSma20 −25.8%), ASTS −7.4% 7D (vsSma20 −18.8%), PL −5.0% 7D (30D −32.2%, vsSma20 −29.9%), RKLB −1.3% 7D (but 30D −20.2%, vsSma20 −14.8%).
  • All six speculative halo names are now well below SMA20 (−15% to −30%) and have given back the pre-IPO run-up; 30D losses of −20% to −32% on the most-directly-competing names.
• Dispersion-by-relationship persisted: diversified/picks adjacents held up far better — IRDM −3.0% 7D (still strong-up, vsSma20 −9.0%), ARKX +2.9% 7D (strong-up, only −1.3% vsSma20). The basket ETF barely flinched while the single-name speculatives bled.
• Verdict: CONFIRMED. The "halo, then air out" pattern played out exactly — the FOMO run-up was sold, the unwind did not reverse on the event, and it concentrated in pure competitors while picks/diversified names held. This was the dominant visible leg, as it was pre-IPO.

Rotation tells (mega-cap reaction)

• No clean rotation-out-of-mega-cap leg. TLT +0.8% 7D / +3.6% 30D (weak-up, RSI 59.5) and IWM +2.4% 7D / +6.1% 30D (strong-up, RSI 58.4, only −2.0% off 52wk high) — IWM firm and small-caps leading, mild bond bid, but neither is a sharp defensive rotation spike.
• Mega-caps are weak, but in a way that predates and is broader than the IPO: MSFT −2.3% 7D (strong-down, −29.1% off high), AMZN −0.5% 7D (−7.1% 30D), META −0.2% 7D, GOOG +0.4% 7D (the only firm one, strong-up). NVDA −6.6% 30D. This is a mega-cap drawdown that is not attributable to the IPO — it's the index-level softness the whole scan is reading. The IPO did not visibly accelerate it; AI-infra small/mid caps (NBIS/CRWV/CORZ/IREN/APLD) are the ones that are firm, which cuts against an IPO-driven drain on the crowded cohort.

Forward: T+50 staggered lockup watch (the real next sell-event)

• The next supply catalyst is NOT a vanilla T+30/T+180 cliff — it's the staggered, earnings-triggered unlock at Q2 2026 earnings (~T+50, ~early August): up to 20% of eligible insider shares, +10% bonus tranche if SPCX is ≥30% above offering at that print. Musk fully locked through all 180 days.
• SPCX at 6/16 close ($201.80) is +49.5% above the $135 assumed offering — well clear of the +30% ($175.50) threshold. If it holds anywhere near here into Q2 earnings, the bonus tranche fires and supply accelerates — the self-dampening-on-strength mechanic engages. Strength now → more insider supply at T+50, not less. This is the single number to track into early August; the +30% line is the trigger, and SPCX is comfortably through it after two sessions.
• T+30 (~2026-07-12) remains the prep/lockup-math-check date; the mechanical event is T+50.

Net read (T+2)

• C CONFIRMED, A NOT-FIRED, B INDETERMINATE (favorable lean). The tape says: the IPO behaved like a halo + hot-debut event, not a liquidity-drain event. The crowded AI-infra silicon/power cohort the book is overweight is firmer, not weaker, two sessions in; the only cohort that bled is the speculative space halo we explicitly do not own. The historical "biggest-IPO-ever tags a top" pattern has NOT shown up in the AI-infra cohort tape at T+2 (it has, arguably, in the broad mega-cap drawdown — but that is pre-existing and not IPO-attributable).
• Discipline caveat: two trading sessions is a thin window; the primary scoring window runs through ~2026-06-26 (T+14). A NOT-FIRED can still flip if silicon rolls below SMA20 later in the window. This is an interim score, not the resolution.

---