Ainvest Option Flow Digest - 2026-07-09: 🧩 The $50M SPY "Call Buy" That Isn't — Plus a Defensive Tilt: an $18M Marvell Cap, a $12M Micron Put, and a Gene-Therapy Put Roll Up & Out

📅 July 9, 2026 | 🔎 THE READ: SPY's $50M "call buy" is a deep-ITM financing print (NOT a bullish bet) + MRVL's $18.2M short-call overwrite + NVDA's 2028 LEAP cross + EEM's emerging-markets call cross | ⚠️ The biggest dollar figure of the day is the least directional — read the mechanism, not the headline

✅ Updated July 10, 2026 — this session's provisional flags are now resolved. The next-day OPRA open-interest snapshot settled all 9 legs. RARE inverted: its "put diagonal" was actually a put roll (the sold leg was a close, not a fresh short). NVDA was corrected (prior OI was 35,052, not 100) and resolved as a partial open. Every other name opened as read. See the resolution section.

---

🎯 Quick Read: $95.1M Tracked — Real Directional Flow ≈$45.3M, and It Leans Defensive

Today is a masterclass in why the dollar figure is not the signal. We tracked ≈$95.1M across 7 names — but $49.8M of it (SPY) isn't a directional bet at all. Strip that out and the genuine directional/positioning flow is ≈$45.3M — and ≈$36M of it is defensive: a Marvell upside cap, a Micron put, a Humana put, and a gene-therapy put roll. Only ≈$9M (NVDA + EEM crosses) leans bullish.

• The mirage: 🧩 SPY $49.8M — a "$500 call buy" that's really a deep-in-the-money call at parity (it traded at $249.06 vs. $248.35 of intrinsic value — ≈$0.70 of actual optionality). It printed as a floor combo with no readable direction and no stock hedge — a financing/box structure, not $50M of S&P upside. If you only read the headline, you'd get this exactly backwards.
• The real trade: 🧢 MRVL $18.2M — a desk sold two layers of Marvell calls ($230 + $250, Sep) for an $18.2M credit — a short-call overwrite capping upside, carried through the ≈Aug-20 earnings. Bearish-to-neutral.
• The near-term hedge: 🛡️ MU $11.8M — a big Jul-24 $825 put cross on Micron (≈+214% YTD near $1,000, but ≈16% off its June high). No earnings before expiry, so it reads as a downside hedge/valuation fade, not a catalyst bet — direction unprovable (a cross).
• The LEAP: 🤝 NVDA $6.3M — a 2028 $250 call cross (≈24% OTM), a multi-year Rubin-era positioning bet — but a cross, so direction is a lean, not proof. ✏️ Corrected July 10: prior OI was 35,052 (not 100), and next-day OI confirms only a partial open (+1,408) — ≈30% of the block was a transfer, not new risk.
• The catalyst play of the day: 🧬 RARE $4.95M long-side / $2.98M net debit — ✅ resolved July 10: a bearish put roll, not a diagonal. After an ≈18% rally against them, a trader closed 6,000 Sep-18 $22.5 puts (which would have expired one day before the UX111 FDA verdict) and opened 6,000 Oct-16 $30 puts spanning both gene-therapy PDUFAs (Aug 23 + Sep 19). Rolled up & out, ≈65% more bearish delta. No short leg, no theta harvest — pure conviction.
• The insurer hedge: 🛡️ HUM $3.2M — a Nov $360 put cross on Humana (≈+50% YTD) — bearish/hedge-lean over Q2 earnings + the ≈Oct Star-Ratings release, direction unprovable (a cross).
• The EM bull-lean: 🌍 EEM $2.8M — a big Aug $69 call cross on the emerging-markets basket — bullish-lean, direction unprovable.

The discipline of the day: a deep-ITM-at-parity print flagged "Block + Spread" is mechanical financing, not conviction. The number that looks biggest is the one that means least.

---

✅ RESOLVED (July 10) — This Session's Provisional Flags

The next-day OPRA open-interest snapshot for the 2026-07-09 session posted July 10 pre-market. We pulled all 9 legs across 7 names. Six names opened as read. One inverted, and one carried a data error we have corrected.

🔄 Inversion

• RARE — "bearish put diagonal" → BEARISH PUT ROLL (up & out). We flagged the Sep $22.5 "short" leg as provisional because size (6,000) exactly equalled prior OI (6,000). Next-day OI FELL 6,000 → 1,227 (−4,773). Open interest only declines when both sides are closing, so the seller must have been long and liquidating — this was a Sell-To-Close, not a fresh short. Paired with the Oct $30 put's clean open (OI 1 → 6,001, +6,000), the trade is a roll: close 6,000 September puts, open 6,000 October puts, in one order. There is no short leg, no theta harvest, and no assignment risk. The trader is long premium on both legs (theta was negative on each). After RARE rallied ≈18% against them — and with their old puts set to expire one day before the Sep 19 UX111 FDA verdict — they paid a $2.98M net debit to roll up in strike and out in time, lifting bearish delta ≈65% (−99,900 → −164,820). Capital genuinely at risk is the full $4.95M of long puts, not a $3.0M spread. Our original "clever theta harvest that steps aside before the binary" read had it exactly backwards: that September expiry was the flaw they were paying to fix.

✏️ Correction

• NVDA — prior OI was 35,052, not 100. Our scanner reported an open-interest baseline of 100 contracts on the Jan-2028 $250 LEAP. The real figure was 35,052. That error made a routine 2,000-lot cross look like a 21x-prior-OI whale; it is actually 0.06x. With size far below prior OI, the trade was never provable as an open from size alone. Next-day OI rose 35,052 → 36,460 (+1,408), so it was a partial net open — only ≈70% of the block created new contracts; ≈30% was a transfer between existing holders. Unusual Score revised MODERATE-HIGH → MODERATE. Direction remains unprovable (a cross).

✅ Confirmations

• MRVL — the provisional $250 leg OPENED (OI 4,103 → 6,197, +2,094), as did the $230 leg (1,245 → 3,221, +1,976). The short-call overwrite read holds; both legs are fresh shorts.
• MU $825 put cross OPENED (125 → 5,546, +5,421). HUM $360 put cross OPENED (13 → 1,012, +999). SPY parity floor combo OPENED exactly (2 → 2,002, +2,000) — which changes nothing: a clean open on a deep-ITM-at-parity print still is not a directional bet.
• EEM $69 call cross OPENED (3,073 → 18,179, +15,106) — more than our 12,000 block, because a companion 3,000-lot cross printed at the same price and strike. Direction still unprovable.

What changed: RARE inverted from a diagonal to a roll, which raises the capital at risk from ≈$3.0M to ≈$4.95M and increases the bearish read rather than softening it. NVDA's headline shrank once the real open interest was known. The standing lesson holds in both directions: when trade size ≤ prior open interest, no order-type label is provable from the tape — and when the open-interest number itself is wrong, every ratio built on it is wrong too. Only the next-day OI snapshot settles it.

---

🔁 OI Review — Last Session's Provisional Flags, Now Resolved (covers 2026-07-08)

The next-day OPRA open-interest snapshot for the 2026-07-08 session is in (posted 2026-07-09 pre-market). We pulled it for all 16 legs across 9 names and resolved every ⏳ provisional flag. The headline: 8 of 9 names opened as read — but 🤝 ARM's $17.4M "call cross" turned out to have created no position at all. That's two sessions running where a big call-sale headline resolved to a close.

🔄 Inversions

• ARM — "sold ITM call, open/close unprovable" → CLOSE on BOTH sides. The 3,134-lot $270 call block printed with size below prior OI (13,195). Next-day OI FELL 13,195 → 10,061 (−3,134) — a decline exactly equal to the trade size. Open interest only falls when both parties are closing, so this was a mutual unwind (the seller closed a long call, the buyer bought back a short) — NOT the covered-call/overwrite we flagged as one possibility. No new position; risk came off the board 3 weeks before ARM's Jul-29 earnings.

✅ Confirmations

• DELL $500 LEAP call OPENED (fresh lit long, bullish, confirmed). GLW deep-ITM premium-collection combo OPENED (both legs). GDX bullish gold risk-reversal OPENED (puts sold / call bought).
• AMD bullish diagonal roll — long Aug $500 leg OPENED; short Jul $470 leg was a close (roll confirmed). MDB roll confirmed — long Nov $350 OPENED (+1,762), short Aug $340 CLOSED (−1,553, a Sell-To-Close). SMH 4-leg ratio structure OPENED across all legs. PDD far-OTM 2027 $120 call cross OPENED (+13,683) — direction still unprovable.

What changed: ARM inverted to a both-sides close — the second straight session a big "call cross" headline turned out to remove risk rather than add it. The standing lesson: a call-sale headline is never fresh positioning until next-day OI proves it OPENED.

---

---

📋 Today's Flow at a Glance

The SPY "$49.8M" is ≈99.7% intrinsic value (a deep-ITM parity floor print) — real premium, but not directional flow. Genuine directional/positioning flow ≈$45.3M (MRVL + MU + NVDA + HUM + RARE + EEM) — ≈$36M of it defensive. RARE's premium is shown as net debit for daily-total consistency; its live capital at risk is the full $4.95M long-put side, since the roll leaves no short leg.

---

🔬 The Seven Positions — Decoded

1. 🧩 SPY — The $50M "Call Buy" That Isn't (the teaching trade of the week)

SEE WHY A $50M SPY "CALL BUY" IS ACTUALLY A FINANCING PRINT →

• Flow: ≈$49.8M — a $500 call (Jul 24) at $249.06, flagged Block + Spread. Spot $748.35 → intrinsic $248.35, so it traded at parity (≈$0.70 of time value). The "$50M" is ≈99.7% intrinsic — not $50M of conviction.
• Why it's not directional: it printed as a floor combo (a complex order worked in the pit against scattered single-leg interest) with no aggressor to read, no matching option partner (checked strikes 500–1000 + the 500 strike across 8 expiries), and no stock hedge block on the equity tape. That's the fingerprint of a financing/box structure — direction unknown, real risk tiny. Do NOT read it as a bullish S&P bet.

2. 🧢 MRVL — The $18.2M Short-Call Overwrite (the real directional trade)

SEE THE $18M MARVELL CALL OVERWRITE CAPPING UPSIDE INTO EARNINGS →

• Flow: ≈$18.2M credit — sold 1,945 Sep $230 calls ($51.35) + 1,945 Sep $250 calls ($42.35), one electronic combo. Spot $247.20 (pulled back from a ≈$310 June high).
• Read: a short-call overwrite (bearish-to-neutral) — capping MRVL's upside above ≈$230–$250 for an $18M credit, carried through the ≈Aug-20 Q2 earnings. Most likely covered against a long position. ✅ Both legs confirmed OPEN by the July 10 OI print ($230: 1,245 → 3,221; $250: 4,103 → 6,197) — the provisional $250 leg was a fresh short, not a close. Not proven bearish conviction — it's a "capped upside" income view.

3. 🛡️ MU — The $11.8M Micron Put Hedge (fade on the $1,000 memory rocket)

SEE THE $12M MICRON DOWNSIDE HEDGE →

• Flow: ≈$11.8M — bought 5,500 Jul-24 $825 puts (≈17% OTM, ≈2 weeks) at $21.50 as a block cross. Spot $998.56. OI 125, size 5,500 → fresh open.
• Read: a big near-term put on a name up ≈214% YTD (but ≈16% off its June all-time high). Critically, MU has no earnings before the Jul-24 expiry (next print ≈Sep 29) — so this is a downside hedge / valuation fade, not a dated-catalyst bet. Direction is a lean (it's a cross), but the timing says "protect/fade an extended stock," not "position for an event."

4. 🤝 NVDA — The $6.3M 2028 LEAP Call Cross (multi-year positioning)

UNPACK THE 2.5-YEAR NVIDIA LEAP CROSS ON THE RUBIN ERA →

• Flow: ≈$6.3M — 2,000 Jan 2028 $250 calls (≈24% OTM) crossed at $31.45 (near the ask — a weak buy-lean). Spot $201.47. ✏️ Prior OI was 35,052, not the 100 we originally printed — a scanner data error. Size (2,000) was therefore far below prior OI, making open vs. close unprovable from the tape.
• Read: a multi-year LEAP positioning bet spanning ≈6 earnings prints and the Rubin → Rubin-Ultra cycles (avg analyst PT ≈$309 already clears the strike). But it's a cross — direction is a lean, not proof. Next-day OI: 35,052 → 36,460 (+1,408) = a partial net open; ≈30% of the block was a transfer between existing holders, creating no new risk. This is a normal-sized block on a crowded LEAP strike, not a whale.

5. 🧬 RARE — The $4.95M Gene-Therapy Put Roll, Up & Out ❗ (inverted July 10)

SEE THE PUT ROLL THAT CHASED ULTRAGENYX HIGHER →

• Flow: STC 6,000 Sep 18 $22.5 puts ($3.28, +$1.97M recovered) → BTO 6,000 Oct 16 $30 puts ($8.25, −$4.95M), one multi-leg auction. Net debit ≈$2.98M. Spot $34.58 (Ultragenyx, ≈+46% YTD).
• The inversion: we originally called this a put diagonal with a fresh short leg. Next-day OI disproved it — the Sep $22.5 put's OI fell 6,000 → 1,227 (−4,773), and open interest only falls when both sides close. The seller was long and liquidating. This is a roll, not a spread: no short leg, no premium collected, no assignment risk.
• The real tell: RARE rallied ≈18% in eight sessions ($29.25 → $34.61), leaving those Sep $22.5 puts ≈35% OTM, 100% extrinsic — and expiring September 18, one day before the Sep 19 UX111 (Sanfilippo A) verdict they existed to capture. That expiry was a flaw, not a clever design. The trader paid $2.98M to fix it: rolling up ($22.5 → $30, ≈65% more bearish delta: −99,900 → −164,820) and out (Sep 18 → Oct 16, now spanning both the Aug 23 DTX401 and Sep 19 UX111 PDUFAs). Conviction increased after an adverse move — the opposite of a theta harvest. Both legs carried negative theta; they were paying decay, never collecting it.

6. 🛡️ HUM — The $3.2M Humana Put Cross (bearish/hedge-lean)

SEE THE $3.2M HUMANA DOWNSIDE HEDGE →

• Flow: ≈$3.2M — bought 1,000 Nov $360 puts (≈9% OTM) at $31.79 as a block cross. Spot $395.98 (HUM ≈+50% YTD). OI 13, size 1,000 → fresh open.
• Read: a long OTM put cross — bearish/hedge-lean over Q2 earnings (Jul 29) and the ≈October 2027 Star-Ratings release (the biggest event before the Nov expiry). But it's a cross — the bearish read is a lean, not proof.

7. 🌍 EEM — The $2.8M Emerging-Markets Call Cross (bull-lean)

SEE THE $2.8M BULL-LEAN CROSS ON EMERGING MARKETS →

• Flow: ≈$2.8M — 12,000 Aug $69 calls (OTM) crossed at $2.35. Spot $66.58. OI 3,100, size 12,000 → fresh open.
• Read: a bullish-lean call cross on the AI-semiconductor-heavy EM basket (TSMC/Samsung/SK Hynix ≈30%), sitting just after a dense Jul-16→Aug-5 catalyst cluster. Direction unprovable (cross).

---

👥 How Four Different Traders Should Read Today

🎲 YOLO Trader

The only lit, real optionality today is elsewhere — SPY's print is parity financing (nothing to chase), and NVDA/EEM are crosses. If you want convex exposure, express your own view on the catalysts (SPY post-FOMC, NVDA into Aug-26 earnings via a defined-risk spread) rather than copying prints. MRVL's fade is only for advanced traders and only defined-risk — never mirror the naked short-call side.

📈 Swing Trader

MRVL is the cleanest read — a desk capping upside at $230/$250 after a 20% pullback; the $250 leg has now been confirmed open, so trade the Aug-20 earnings reaction off the gamma levels. For SPY, trade the $750/$755 gamma range and let CPI/FOMC break it. Don't confuse SPY's financing print for a directional signal. And study RARE's roll: the trader's original puts expired one day before their catalyst — always check your expiration against the event date.

💵 Premium Collector

MRVL is your masterclass — an $18M covered-call overwrite you can mirror small and covered (never naked). SPY's $750/$755 gamma pin supports a defined-risk iron condor into the near OPEX (sized for the CPI/FOMC event risk). Selling premium is the theme — but always defined-risk into a heavy macro calendar.

🌱 Beginner

Today's #1 lesson: the biggest number is the least meaningful. SPY's "$50M call buy" is a financing structure with no direction — copying it would be copying nothing. NVDA and EEM are crosses (positioning confirmed, direction not). Only MRVL is a readable trade, and even that is "capped upside," not "short the stock." Watch, learn to read the mechanism, and never trade off a headline dollar figure.

---

⚠️ Risk Control & Patience — Read This Before You Copy Anything

• A deep-ITM call at parity is not a bullish bet. SPY's $50M print is ≈99.7% intrinsic value — a financing/box leg, not $50M of upside conviction. Reading it as bullish is the exact trap this digest exists to prevent.
• A sold call ≠ a short bet. MRVL's $18M is an overwrite capping upside, most likely covered — "bearish-to-neutral," not "the stock is going down."
• Crosses confirm positioning, not direction. NVDA and EEM opened, but a cross has a known counterparty and no aggressor — the bull-lean is a lean, not proof.
• Open vs. close still rules. Yesterday's ARM "call cross" resolved to a both-sides close — the second straight session a big call-sale headline removed risk instead of adding it. ✅ And today's RARE "diagonal" resolved to a roll: its "sold" put was a close, not a short. When trade size ≤ prior open interest, no order type is provable until the next-day OI print. A SELL is not always bearish, and it is not always income.
• Check the open-interest number itself. NVDA's "21x prior OI" whale evaporated once we found the real baseline was 35,052, not 100. Every Vol/OI ratio is only as good as its denominator.

Tenor tags: 🗓️ Near-term: SPY (Jul 24), MU (Jul 24) · 📅 Monthly: EEM (Aug), RARE (Sep/Oct), HUM (Nov) · 🍂 Quarterly: MRVL (Sep) · 🚀 LEAP: NVDA (Jan 2028)

---

Every classification above is read from the real OPRA tape (mechanism, size vs. open interest, and — where valid — aggressor), not from a screenshot label. Deep-ITM-at-parity floor prints are financing, not conviction; crosses confirm open, not direction. Educational flow analysis, not investment advice — trade your own plan and manage risk.