Ainvest Option Flow Digest - 2026-08-11 — The Board Where Almost Nothing Is Scheduled, and the Four Trades Where Something Is

Last updated 2026-08-11 — four late names (TGT, CLX, VSAT, BIRK) added after first publication. Board total revised from ≈$202.8M / 12 names to ≈$337.1M / 16 names.

≈$337.1M net premium across 16 names. The pattern today is not a stock or a sector — it is a calendar. On name after name, big premium was sold into expirations that contain no earnings report at all: Marriott, AMD, FedEx and Qiagen are all short premium across windows where nothing is scheduled to happen. And in FedEx's case that is not an accident of timing — the September earnings print most calendars still show does not exist.

Then four late trades arrived and inverted the theme — they are all about dates, but dates that DO exist. Target and Clorox each sold deep in-the-money calls at essentially pure intrinsic value on the last session before an August 12 ex-dividend; Birkenstock bought a ladder of puts two days before a confirmed August 13 earnings print; and Viasat put on the one structure all day that genuinely takes a directional view. Same lesson from the opposite direction: find the date first, then read the trade.

⚡ Quick Read

1. 📦 FDX — the best piece of homework on the board. A desk sold 10,000 September $330 calls for ≈$9.76M into a window that most earnings calendars say contains a FedEx print. It does not. FedEx moved its fiscal year end from May 31 to December 31, effective June 1, 2026 — there is a seven-month transition period, no June-August quarter, and the next report is October 28, company-confirmed on FedEx's own calendar, ≈40 days after this option expires. Several third-party calendars still project September 17-18. Anyone selling or buying this contract as an earnings play is trading a print that will not come.

2. 🏨 MAR — ≈$22.75M of calls sold into a six-week hole. Marriott reported on August 3, eight days before this trade, and the next report lands ≈October 29 to November 4 — 6-7 weeks after the September 18 expiration. Marriott has not even announced the Q3 date, and that announcement itself comes after this contract expires. The only company event inside the window is the August 20 ex-dividend. Note the strike: $330 against a $348.47 stock, so the seller is capping upside below the current price, and consensus sits ≈15% above that strike.

3. 🔴 AMD — ≈$21.3M collected, and both expirations expire before AMD next reports. The desk sold the September $470 call — essentially exactly at the money — and bought the August $500 call, a reverse diagonal that is short the expensive at-the-money time value. Q2 landed August 4; Q3 arrives ≈early November, ≈7 weeks past September 18. The one real volatility source inside the window is Nvidia's August 26 earnings — someone else's event.

4. ✈️ RTX — ≈$139.9M traded to move ≈$486,430. Same strike, same size, two expirations: sell the December $125 call, buy the February-2027 $125 call. With the stock at $224.42 the $125 strike carries $99.42 of intrinsic value, so the December leg's entire time value was $0.91 and February's $1.61. The desk bought two extra months of identical exposure for roughly seventy cents a share. The February leg had prior open interest of zero — the cleanest certain open on the board.

5. 🚨 QGEN — the trade we first picked up never stood. A $1.3M August call purchase surfaced on the flow screen at 10:33:24. The tape then shows a cancellation at 11:14:34 matching that print on both size and price. What actually survives is two 10,000-lot crosses at $1.00 — and both printed at or near the bid, not the ask. So the surviving flow looks like selling, not the bullish buying the screen implied. The quoted open interest was wrong too: the screen said 40, the tape says 45,759.

6. 🇰🇷 EWY — yesterday's Korea strangle came off, and today a new one goes on. Monday's 11,250-lot October $165/$130 package closed overnight — the call fell 25,142 → 13,839 and the put 30,785 → 19,246. Today a fresh 10,000-lot version of the same two strikes prints again, this time with a paired stock leg attached, which means the options alone no longer express a clean view. And the calendar bites: SK hynix reports October 27 and Samsung October 29 — both after the October 16 expiration. That is ≈44.6% of the fund's weight delivering its biggest scheduled catalyst after these contracts are gone.

7. 📉 TLT — a genuine directional bet, and the rare name whose window is full. Buying the $82 put and selling the $82 call at the same strike and expiry is a synthetic short — delta ≈ −1.00, almost no time-decay or volatility exposure. Because a long-bond ETF falls when yields rise, this is a bet that long-term yields keep going up, with the 30-year already at ≈5.24%. Unlike most of today's board, the September window is dense: two CPI prints, payrolls, and the September 15-16 Fed meeting with a fresh dot plot, two days before expiry.

8. 🛡️ IGV — 34,368 December puts bought against 2,742 of prior open interest. A proven, unambiguous new hedge — ≈$9.5M, struck ≈13.8% below spot. The insight is what the fund actually is now: Palo Alto, Microsoft, Palantir and CrowdStrike are ≈37% of it, versus ≈17.5% in classic subscription software. This is a momentum-unwind hedge wearing a software-ETF label, and the $90 strike only starts paying on a full round-trip of the summer rally.

9. 🧬 ABVX — the flow screen showed half the trade. It surfaced a $3.16M purchase of 2,000 December $140 calls. Scanning the full chain found the other leg at the same millisecond: 4,000 December $200 calls SOLD at the bid. That makes it a 1×2 call ratio spread for ≈$2.16M net — a position that wants Abivax to rise to $200 and not through it, since above ≈$249 the extra short call turns the loss open-ended. And the calendar fits today's theme: no confirmed binary event lands inside December — the only company-guided item is an NDA submission pencilled to "Q4 2026", which is a quarter, not a date, and a filing in the last two weeks of December would satisfy the guidance and miss the option entirely.

10. 💵 TGT and 🧴 CLX — two desks, one mechanic, and it is the best lesson on the board. Both sold deep in-the-money calls priced at essentially pure intrinsic value on the last cum-dividend session before an August 12 ex-date. Target's January $75 call printed eight cents BELOW intrinsic; Clorox's August $95 call carried ≈$0.20 of time value against a $1.25 dividend — a 6.3× ratio. When a dividend dwarfs the remaining time value, rational call holders exercise early and short holders get assigned. Clorox's twenty identical 1,020-lot clips in fourteen minutes is the textbook signature. Neither trade says anything about where the stock goes — and if assignment runs through the Clorox position tonight, its August 21 expiration never arrives. Tomorrow's open interest at the $95 strike is the arbiter, and we have published the falsifiable prediction.

11. 🩴 BIRK — ≈$14.6M of puts bought two days before a confirmed earnings print. 27,000 contracts across three August 21 strikes, in two floor clips. The company confirmed the August 13 pre-market report back on July 15, so these are earnings contracts, not a vague hedge. The cost is the story: ≈$3.9M of pure time value for ten days spanning one print, and ≈1.9M shares of short delta — ≈3.6% of a float that is only 53.75M shares, because L Catterton controls ≈60–65%. With short interest at ≈21% of that float, a good print squeezes as violently as a bad one falls. The options market is pricing ±9.24%; the last report took the stock to ≈$31.

12. 🛰️ VSAT — the only genuine directional structure all day. A bear call spread collecting ≈$13.4M on a $35-wide spread, so ≈$30.8M of risk against it. Breakeven ≈$90.62 — and the 52-week high of $93.03 sits just ≈$2.40 above it. This position needs a stock that is +145.8% year to date on 1.2% LOWER revenue to stay below a level it already reached this year. One caution we will not bury: both legs printed just under existing open interest (12,618 against 12,721 and 12,675), so if this is a position being closed, the entire bearish reading inverts.

The habit to take from today: check whether the expiration actually contains the event before you assume the trade is about it. Five separate desks put on positions in windows with nothing firmly on the calendar. That is a deliberate, repeatable choice — and the mirror image is the warning for anyone copying them: you are being paid for time, not for a view, and the payment stops the moment something unscheduled happens.

✅ RESOLVED 2026-08-12 — Tuesday's ⏳ Flags, Settled by This Morning's Open Interest

The ≈06:30 ET OPRA snapshot timestamped August 12 (reflecting the August 11 close) is in. Every ⏳ flag on this board is now resolved. Baseline is the August 11 snapshot, which reflects the August 10 close — before these prints.

Seven of sixteen names inverted. That is a high rate, and it is worth saying plainly: on a board this heavy with negotiated blocks printed at or near prior open interest, the trade-day tape genuinely could not tell opening from closing — and more often than not, it was closing.

🔄 Inversions — a published provisional label flipped

✅ Confirmations — the published read held

⚠️ Neither branch — resolved into something we did not publish

What this board actually teaches. Eleven of these sixteen names printed at or below prior open interest, which is why so many carried a ⏳ in the first place — and of those, the closing branch won far more often than the opening one. When a negotiated block lands on a strike that already has a comparable position sitting on it, the base rate favours position management over new conviction. The headline premium is real money changing hands either way; what it is not is proof that someone just took a view.

🔁 OI Review — Monday's Provisional Flags, Now Resolved

Monday's board has been through the next-day open-interest test, and seven of twenty names came back meaning something different than they looked. All 54 legs resolved — nothing expired, nothing failed to pull. 37 opened, 13 closed, 4 were transfers. The theme is blunt: a large BUY is not a position. Five separate legs printed as purchases while the strike's open interest went down.

🔄 Inversions

• 🔄 SNDK — a $436.6M day that contained ≈$15.6M of actual new position. We published the falsifiable branch — "falls back toward ≈660 means it closed." Open interest printed 660, exactly. It was a close, not a quarter-billion-dollar short. The $132.0M June-2027 call moved 5,911 → 5,910 — one contract, a pure transfer. Only the smallest and only lit trade added new exposure. $436.6M of headline premium, ≈3.6% of it new.
• 🔄 EWY — the "long strangle" was a strangle being closed. Both strikes landed within ≈60 contracts of our published closing-branch numbers. Order type on both legs corrects from BTO to BTC.
• 🔄 GLD — the "self-financing bullish" September spread was an unwind. The $400 and $425 calls fell −17,962 and −18,028 — a 66-contract difference across a 55,070-lot print, the signature of one two-legged position being retired. Corrected: the desk closed long September upside and opened short spreads on top.
• 🔄 SE — the December bull spread was taken off the day before earnings. The $97.5 call fell 2,635 → 360, exactly −2,275, the print size. Of ≈$22.17M, only the ≈$11.27M October package was new money.
• 🔄 SPY — this one corrected us, not the classifier. We called the $780 print a buy-to-close and predicted open interest would fall toward ≈85,531. It rose 2,275. "Mostly someone covering a short" is retracted as a stated fact and reframed as a transfer.
• 🔄 SPHR — the long leg never opened; the package is ≈4.6-to-1 short-to-long, more exposed than we published. 🔄 COIN — the August leg was not new bullish money; every framing of it as a bet on clearing $160 is withdrawn.

✅ Confirmations

• ⭐ MSFT — the roll confirmed on both ends, and then some. The $550 calls opened to 55,221 against a predicted ≈30,500-34,498 — 168% of the package's own size, meaning other desks were buying the same strike alongside it.
• ROCK resolved to the contract (0 → exactly 4,500); SPCX was one contract off; CXW, TSM, CVNA, DOCU, GFS, QQQ and AAPL all confirmed as written.

⚠️ Direction right, magnitude overstated

Five names confirmed their published direction while retiring far less than their print size — DASH retired only 1,371 of 10,000 (≈14%), GLW 4,071 of 10,000 (≈41%), and AAPL, SE and SKHY around half. "The position came off" and "1,371 contracts came off" are very different statements.

The standing lesson, taught five times over on Monday: a BUY with falling open interest is a close. The tape prints "BUY" either way. Only the next morning's open interest separates new conviction from an exit — and when it comes back flat, the honest answer is "it changed hands."

📊 At a Glance

⏳ = at least one leg where size sat below prior open interest, so open-versus-close could not be proven from the trade-day tape. These flags are now resolved — see the ✅ RESOLVED section above. The August 12 ≈06:30 ET snapshot settled every one of them: 7 of 16 names inverted, 6 confirmed, and 3 resolved into something neither branch described.

📅 Upcoming Catalysts — and Which Expiration Actually Contains Them

Read this carefully: a catalyst date and an option expiration date are different things, and today most of these trades expire before the event a reader would assume they are about.

Nine mismatches on one board, and two of the new ones are the most instructive. Viasat's December-quarter results are a February event because its fiscal year ends March 31 — a calendar-year reporter would have fit two prints inside that expiration; Viasat fits one. And Target's holiday quarter is fiscal Q4, which does not even end until ≈late January 2027 and reports ≈early March. Anyone treating a December or January Target option as "the holiday trade" has the calendar wrong by a full quarter.

In almost every case the option expires before the event a reader would naturally assume it was about. That is not sloppiness by the desks — for the sellers it is the entire point. Birkenstock is the exception that proves it: there, the desk deliberately bought the window that does contain the print.

👥 Four Ways to Read Today

🎲 The YOLO trader. The late board handed you a cleaner setup than QGEN: BIRK's puts expire eight days after a confirmed earnings date, which is at least an honest binary. But price it properly — ≈$3.9M of that ≈$14.6M is pure time value, the market is only pricing a ±9.24% move, and with short interest at ≈21% of a 53.75M-share float a good print can squeeze as hard as a bad one falls. Being right on direction and still losing is entirely possible here. The older temptation is QGEN — a cheap, near-dated call on a diagnostics name. Two problems: the print that made it look exciting was cancelled, and what survives printed at the bid, which looks like selling. On top of that the stock's beta is 0.62 and there is nothing scheduled before expiry. If you want convexity today, ASML's December-2027 call is the intellectually honest version — but it is 100% time value with a breakeven ≈30% above spot, so size it as money you can lose entirely.

📈 The swing trader. TLT is the only clean directional statement on the board, and it comes with a dense calendar to trade around: CPI on the 12th, payrolls on the 4th, and the Fed on the 15th-16th. Note what the structure tells you — a synthetic short has almost no volatility or decay exposure, so this desk wanted direction and nothing else. The counter-case is real: July payrolls came in at −23,000 with heavy downward revisions. Do not confuse a well-constructed trade with a correct one.

💰 The premium collector. Read TGT and CLX before anything else, because they teach the trap you are most likely to fall into: writing a covered call the day before an ex-dividend. Both desks sold deep in-the-money calls carrying almost no time value on the last cum-dividend session — Clorox collected ≈$0.20 of extrinsic against a $1.25 dividend it was about to forfeit on assignment. If you overwrite for income, check the ex-date first; a call with no time value left is not income, it is a share sale waiting to be exercised. And VSAT shows the other edge: a bear call spread collecting ≈$13.4M risks ≈$30.8M — ≈2.3× the credit — on a stock up ≈146% this year, with a breakeven only ≈$2.40 below its own 52-week high. Capped risk is not small risk. Beyond those, today is your board — MAR, AMD, FDX and QGEN are all sellers, and all four picked windows with no earnings. That is the professional version of what you do. But look at what they are being paid: MAR's $22.75M headline contains only ≈$4.6M of actual time value once you strip out the $18.47 of intrinsic. And the risk is not symmetric — MAR's consensus target sits ≈15% above the strike they sold, and AMD sold an at-the-money call on a stock up ≈109% this year. An empty calendar removes the scheduled risk, not the unscheduled kind.

🌱 The beginner. Learn one idea today: check whether the option expires before or after the event. FedEx is the perfect teacher — a September contract that most calendars claim contains an earnings report, on a company that moved its fiscal year so the report does not happen until October 28. Anyone who bought or sold that contract "for earnings" was wrong about the basic facts. Before you react to any flow headline, find the expiration date, find the earnings date, and put them in order. Also learn the ⏳ symbol in our table: it means we genuinely do not know yet, and saying so is the honest answer.

⚠️ Before You Trade Any of This

Every leg on today's board was negotiated — floor blocks, crosses, or a price-improvement auction. On the two dividend trades the point goes further: TGT and CLX both printed at parity, and a block struck at intrinsic value has no aggressor at all. Nobody lifted an offer or hit a bid, so the BUY/SELL labels on those two are reported, not proven. There is a known counterparty on the other side of each one, and that counterparty may be just as informed. This is not urgent buying or panic selling.

We correct ourselves in public. Today QGEN's headline trade turned out to have been cancelled, and yesterday's review retracted our SPY read outright. The screen's open interest was wrong by more than 500× on Palantir and by three orders of magnitude on Qiagen — we caught both against the tape, and it changed the verdict in three names. If a number here differs from what you see on a flow screen, that is usually why.

✅ Updated 2026-08-12 pre-market — the snapshot landed, and it settled every ⏳ flag on this board. Clorox's August $95 strike was the number we told you to check: it printed 27 contracts, down from 3,369 — assignment ran straight through the position overnight, exactly as described. Across the board 7 of 16 names inverted, 6 confirmed, and 3 landed on neither branch we published. See the ✅ RESOLVED section near the top for the full leg-by-leg table. The standing lesson holds: a big headline premium is not a position until the next morning's open interest says it survived. Position for being wrong, not for being right.

This is market analysis and education, not investment advice. Options carry substantial risk of loss, and uncovered short options carry theoretically unlimited risk.