📊 AInvest Option Flow Digest — 2026-09-10

$142.2M net premium across 11 names. Six of the eleven structures are credits. The Fed meets in five days with a hike as the base case — and the biggest trades on the board are being paid to take the other side of that fear. One desk ignored all of it and bought upside instead. And the largest number on the board today, $416M in SpaceX puts, is not a bet at all — we solved it, and the real risk is $20.5M.

⚡ The Quick Read

Today has a single organising idea, and it is not the one a flow feed shows you.

Desks were being paid, not paying. Six of the eleven structures are net credits. Together they collected roughly $88M more than they spent. That is not a market buying protection — it is a market selling it.

Why they would: the variance premium is roughly double. Nasdaq-100 realized volatility is running ≈12.6% annualised against VXN at 22.32. On the S&P the gap is wider still — a ≈2× premium. When options price a ±5% month and the tape delivers ±2.5%, selling that gap is a coherent, well-known trade.

Why it might not work: the man setting rates is watching that same calm. Federal Reserve chair Kevin Warsh said at Jackson Hole that financial conditions show "few signs of policy restraint," and on inflation, "otherwise, we have work to do." July's hold was 9-3, with three dissents in favour of a hike. This morning's producer-price print came in at +5.4% year over year, up from 4.8%, pushing hike odds for September 16 toward ≈70%. A central bank that reads low volatility as evidence policy is too loose is a genuine hazard for anyone short volatility.

The upside is being sold in exactly the names that ran hardest. MU is up ≈243% year to date and a desk sold its $1,100-$1,350 call band for a $31.6M credit. ARM is up ≈134% and someone collected $10.3M to cap it above $400 until 2028. CRWV had its calls sold against bought puts.

Two of those credits are riskier than they look, and this is the most useful thing in today's digest. Micron's short $1,100 strike sits 12.4% below its own 52-week high — the stock traded straight through that band in late June — and the average daily range is $53.77, so it is barely two days' movement away. The DRAM fund sold two uncovered $45 puts for every one it bought, and that fund closed at $44.85 on July 29, below the strike, after a 31.9% four-week fall. Selling insurance at a level the market broke six weeks ago is a different proposition from selling a remote one.

One desk went the other way entirely. Q — Qnity Electronics, the DuPont semiconductor-materials spinoff — saw a May-2027 $140/$170 call spread bought and funded almost entirely by selling a $90 put. Net cost $401,200 for ≈$14.6M of long-equivalent exposure. In a session where everyone else was capping upside, someone paid for it.

The biggest headline of the day is not a trade, and we can prove it. SPCX — SpaceX, public only since June — had 100,562 put contracts cross the exchange floor for $416.3M. It is the largest gross number we have published in weeks, and it is almost entirely an illusion. Each tranche splits into two halves of exactly equal size, and once split, the premium difference equals the difference in strike prices to the dollar, with zero error, in four of the five tranches. Equal sizes mean the share obligations cancel. What is left is the strike-price difference: $20.5M. The headline overstates the real risk 20.3 times. One of the tranches is 5,000 puts at $200 against 5,000 puts at $205, priced $52.30 against $57.30 — exactly $5.00 apart, the exact distance between the strikes. That is bookkeeping, not conviction.

And the feed was wrong again — fourth straight session. NVDA showed two long calls. An 890,000-share hedge printed 44 seconds later. Two longs would need 1,443,050 shares to hedge (a 62% match); a $245/$300 spread needs 882,350 (a 99.1% match). It is a capped spread with $24.7M at risk, not $35.2M of call buying.

🔁 OI Review — Yesterday's Predictions, Scored

The September 10 pre-market snapshot settles all twenty-nine option lines from yesterday's sixteen-name board. Twenty-one opened, two closed outright, six were partial closes. Three of four falsifiable tests passed. Two reads inverted — and one of them was our lead story. Untraded control strikes were pulled against every declining line; none collapsed, so every decline belongs to the line that traded.

🔄 Inversion 1 — MSTR: the reversal never happened

We led yesterday's digest with a 24-hour round trip. The open interest says there wasn't one.

The article published the test in advance, in bold: a genuine unwind needed the November $135 line to fall toward ≈418, and said that if it "instead stays near 13,982, or rises... the reversal narrative would be wrong."

It rose. 13,982 → 31,094, a gain of +17,112 — larger than the 14,400 printed, and 91.9% of everything that traded on that line all session. The sale opened fresh short calls. The September $100 leg did fall (33,397 → 27,468, −5,929, 41.2% of the print), so that side was a partial close. What actually printed was a partial September close plus a new November short — a position going on, not coming off. Corrected MSTR analysis

The uncomfortable part, stated plainly: that November leg carried a HIGH-confidence sell-to-close label, supported by a matching strike, expiration, timestamp and near-matching size. All of it was circumstantial, and next-day open interest overturned it. This is the case to remember the next time an open/close label looks obvious.

🧩 Inversion 2 — SKHY: the backspread came OFF, it didn't go on

Both legs closed at essentially full capture. The $155 call fell 4,315 → 320 (−3,995 of a possible 4,000) and the $240 call 10,178 → 2,185 (−7,993 of 8,000). Both 99.9%. Eight untraded December control strikes did not collapse; the $250 line actually rose.

And the structure inverts with the verdict. A sale that retires contracts closes a long; a purchase that retires contracts closes a short. So the position that existed beforehand was long 4,000 $155 calls against short 8,000 $240 calls — the mirror image of the backspread we described being established. The $8M credit was the proceeds of an exit. Corrected SKHY analysis

🔧 Two structural revisions

• ORCL was an adjustment, not a covered call written on the day. The $160 line fell 8,094 → 5,538 (−2,556) while the $200 line rose 13,446 → 18,419 (+4,973). At least 2,556 of the calls sold retired contracts — the shape of a call position rolled from $160 to $200. ORCL
• PLTR: package or roll? ROLL. The September $135 line fell 21,572 → 15,921 (−5,651, exactly 50.2% of the print) against a November $165 line that added 9,016. We published both readings as equally live and let the snapshot decide. PLTR

🎯 Four falsifiable tests, four results

Test	What we required	What happened	Result
AFRM   — one spread or two prints?	both lines move similarly	+2,233 and +2,469, a 10% gap	✅ PASSED
NXPI   — one 2:1 package?	lines grow in a 2:1 ratio	9,268 against 4,630 —   2.0017 to 1	✅ PASSED within 0.1%
QQQ   — one collar?	call and put legs grow matched	+12,394 and +12,490, a 96-contract gap	✅ PASSED
MSTR   — a genuine unwind?	Nov $135 falls ≈12,000-13,900	it   rose +17,112	❌ FAILED, narrative inverted

✅ Confirmations

• All three September rolls confirmed in direction, all partial rather than full. NVDA's $195 line fell 58.7%, AMD's $470 line 37.0%, COIN's $170 line 45.3%. Every deferred leg opened. NVDA · AMD · COIN
• TSLA landed within four contracts of both predictions. TSLA
• CRCL cleared the exact threshold we set — needed ≈2,500 or higher, landed at 3,734. CRCL
• LYFT opened at full capture, +20,078 against a 9,995-contract print. LYFT
• M, GXO and NXPI all opened at or above full size. M · GXO · NXPI
• IBIT opened on both legs with one honest wrinkle — the short leg captured 82.0% against the long leg's 99.2%, so ≈2,695 contracts were a transfer, exactly what the article named as the thing to watch. IBIT

📉 Where our bands were wrong

Three legs opened above the top of our published range — AMD's October $490, COIN's November $170 and NVDA's November $210 — all in the same direction. Each time other participants opened the same strike alongside the block, and our ceiling assumed the block was the whole story. No leg opened below its floor. The bias is one-sided: when a strike is nearly empty and a large package lands on it, the block is often not the only opener.

What changed in the dollars: nothing. Yesterday's $198.0M net still describes the premium that changed hands. What changed is what it did — $8.0M of SKHY credit retired a position rather than establishing one, and MSTR's November leg created exposure rather than removing it.

📋 At a Glance

Ticker	Net Premium	Expiry Class	Catalyst (and when)	The Option Play	What It Actually Means
MU	$31.6M credit	Monthly (Nov 20)	Earnings Sep 30, company-confirmed	Sells the $1,100-$1,350 call band	Caps upside for $31.6M against $148.4M of risk
SPCX	$20.5M net (solved)	Monthly (Sep 18)	None — this is not an event trade	Deep-in-the-money parity put ladder, 16 strikes	Mechanical bookkeeping; the $416.3M headline overstates it 20.3x
QQQ	$20.1M credit	Monthly + Quarterly	CPI Sep 11; Fed Sep 15-16	A financed put hedge AND a hedged put write	Two desks, opposite views, 26 minutes apart
SPY	$13.4M credit	Quarterly (Sep 18)	Fed Sep 15-16; expiry + rebalance Sep 18	Six put legs, net ≈3:1 SHORT	Selling downside into the Fed, not buying it
RCL	$12.4M credit	Quarterly (Dec 18)	Earnings Oct 27; fuel costs	$270/$250 put spread — likely a roll down	Banking a hedge's profit, not a bullish bet
ARM	$10.3M credit	LEAP   (Jan 2028)	Earnings ≈Nov 4 (estimated)	Sells the Jan-2028 $400 call	Paid to cap a +134% winner for 16 months
CRWV	$0.5M credit	Monthly (Jan 2027)	Earnings ≈Nov (estimated)	Short $135 call, long $70 put	Collar or synthetic short — no stock leg found
NVDA	$24.7M debit	Quarterly (Dec 18)	Earnings ≈Nov 18 (derived, not confirmed)	$245/$300 call spread + an 890,000-share hedge	The only clean bullish bet, and it caps itself
Q	$0.40M debit	Monthly (May 2027)	Earnings ≈early Nov (estimated); chip capex cycle	$140/$170 call spread funded by a short $90 put	The only outright-long bet, funded by downside risk
XLY	$7.9M debit	Quarterly (Dec 18)	CPI Sep 11; Canada tariff Sep 15; Fed Sep 16	At-the-money $112.50/$107.50 bear put spread	The only clean paid-for bearish bet
DRAM	$0.5M debit	Quarterly (Mar 2027)	Memory cycle; a large buyback ends Nov 19	2:1 ratio put spread + a 50,000-share hedge	A volatility bet with an uncovered tail

Totals: $142.2M net · $557.5M gross · 11 tickers · 36 legs. Net is the number that matters — it is the capital actually committed. Gross counts both legs of every spread and so counts the same dollars twice. SPCX is carried at its solved $20.5M rather than either its $167.1M of reported legs or the $416.3M the floor session actually printed, because we were able to prove what the package really risks.

💰 Theme 1: The Winners Are Having Their Upside Sold

Three of today's four largest credits are in the year's best performers. Notice what that is and is not: nobody is betting these companies fail. They are being paid to give up the top end.

🐋 MU — $31.6M collected, $148.4M at risk, and the strike has already been visited

7,200 contracts on each leg: sell the November $1,100 call at $68.40, buy the $1,350 at $24.46. The spread is $250 wide and $43.94 was collected, so maximum profit is the $31.6M credit below $1,100 and maximum loss is $148,363,200 above $1,350. The credit is not the exposure — that distinction is where retail traders following credit structures get hurt.

The easy story is "selling upside after a 243% run." The truer one is more uncomfortable. Micron is already −21.9% from its 52-week high of $1,255, has managed only ≈+9.9% in three months, and the short $1,100 strike sits 12.4% below that high — the stock traded through it in late June. With a 14-day average range of $53.77, $1,100 is about two days' movement away. And earnings land September 30, company-confirmed, inside the expiry.

What justifies it, with an honest caveat: a Chinese competitor went from 4% to 10% of DRAM share in a year, and an 86% guided gross margin is a peak-cycle signature rather than a steady state. On prices, though, the evidence is two-sided rather than one-way — leading-edge DDR5 softened on September 10, but mainstream DDR4 spot rose 1.78% the day before, stockpiles are reportedly under ten days, and one bank called for a further September rise. A polarising cycle is weaker justification for selling upside than a clean downtrend would be, which makes this credit riskier, not safer. → The credit, the risk, and the strike it already broke

🦾 ARM — $10.3M to cap a stock 56% above where it trades, until 2028

One leg, no spread, no stock attached: sell 2,000 of the January-2028 $400 calls at $51.51. That is $10.3M collected to give up everything above $400 on 200,000 shares' worth of Arm for 16 months.

The price tells you something. $51.51 for a strike 56% out of the money is what the market thinks Arm's volatility is worth over that horizon, and selling it is a bet that realised volatility comes in lower. Worth knowing: $400 sits above any level Arm has verifiably closed at. Also worth knowing: this position spans roughly six earnings reports, and the tape cannot tell us whether it is written against stock or naked. → Why the premium is that rich, and what it costs to sell it

🎭 CRWV — calls sold, puts bought, and no stock leg on the tape

23,000 contracts on each side of a January 2027 structure: short the $135 call, long the $70 put, for a near-costless $460,000 credit. If the desk owns the stock this is a collar. If it does not, it is a synthetic short.

Here is why it matters: three other packages today came with a stock block proving they were delta-hedged. This one did not. No paired equity print appears, which is not the same as proving there is none — but if it is unhedged, the package carries ≈−1,151,610 shares of delta, roughly $105M of short-equivalent exposure. We give both readings because the tape genuinely will not settle it. → Collar or bearish bet, and why the difference is $105M

⚖️ Theme 2: The Index Insurance Fight

Both major index funds saw large flow, and in both cases the interesting part is that participants disagreed with each other.

📉 SPY — six put legs, and the net is the opposite of what it looks like

Every leg expires September 18 — the quarterly expiration, two days after the Fed decision. Two clean bear put debit spreads sit inside the package: a $750/$730 for $9.7M at 09:44, and a $760/$745 for $14.9M at 13:12. But alongside them at 13:12 came outright sales of 66,431 of the $740 put and 60,500 of the $750.

Total it up: 65,710 contracts bought against 192,641 sold — roughly 3:1 net short, for a $13.4M credit on $91.3M gross. So a package made entirely of puts is a net seller of downside insurance through the Fed meeting. That is the 2× variance premium being harvested, and it expires into the same closing auction where index funds must trade the S&P 500 quarterly rebalance.

We cannot prove all six legs belong to one desk, and we say so in the article. → How an all-put structure ends up short downside

🔀 QQQ — two desks, opposite views, 26 minutes apart

At 10:36 a desk built a hedge: long the $680 and $650 puts, funded by selling a $720/$735 call spread, for a net ≈$1.1M. Cheap insurance through month-end.

At 11:02 a different desk sold 10,000 December $690 puts for $21.2M, and hedged it with a 350,000-share block — a 99.2% delta match. A delta-hedged put write: selling downside premium with the direction neutralised.

One paid for protection. The other was paid to provide it. Same fund, same morning. Anyone who saw only one of those prints would have drawn a confident, half-wrong conclusion — and that is the honest lesson of reading flow. → Both trades, side by side

🛒 Theme 3: The Consumer, Two Ways

🛍️ XLY — the only clean paid-for bearish bet on the board

44,499 contracts on each leg of an at-the-money December $112.50/$107.50 bear put spread, $7.9M net debit. Maximum payoff $14.4M if the fund is below $107.50 — only ≈4.1% away. It risks $7.9M to make $14.4M on a modest decline, which makes it a considered moderate-conviction trade rather than a lottery ticket.

The macro backing is unusually specific: Brent above $105 and up ≈62% year to date, retail sales −0.6% month-on-month in July, consumer sentiment at 51.7, an effective tariff rate of 7.2% costing ≈$820 a household, and the Fed's own minutes describing higher-income households supported by stock gains while lower-income households face "increasing strains."

The detail we liked most: the fund is 42.5% Amazon and Tesla combined, and while it sits only ≈10% off its August high, McDonald's, TJX and Lowe's are all on their 52-week lows. Two mega-caps are masking a broad breakdown. This trade is arguably a bet that the masking stops. → Why 4% is all it needs

🚢 RCL — a $34M put sale that is probably an exit, not a bet

12,750 contracts each: sell the December $270 put, buy the $250, for a $12.4M credit. As printed that is a bullish credit spread. The open interest says otherwise.

The $270 line held 12,809 contracts dead flat for weeks and 12,750 traded against it — a 57-contract margin. The $250 line held 234 and 12,750 traded. One leg matches a long-static position almost exactly; the other is unambiguously new. And the stock has fallen ≈20.9% from its August 5 close of $327.42, with the relative-strength index near 21.9, so anyone holding $270 puts from the $320s is deep in profit.

So the likely story is a hedge being rolled down: bank the gain on the $270 puts, re-establish cheaper protection at $250. Tomorrow's snapshot settles it — the $270 line should fall toward ≈57 if it is a roll, or rise if it was a new short.

Note also that Royal Caribbean is 1.88% of XLY. One desk bought sector downside while another let go of single-name downside. Same sector, same day, opposite conclusions. → The two numbers that decide it

🌱 Theme 4: The Four That Are Not Credits

🎯 NVDA — the feed said two long calls; the hedge says a capped spread

35,000 contracts of the December $245 call and 35,000 of the $300, both marked as purchases. Then an 890,000-share block printed 44 seconds later with the contingent-trade marker.

Do the arithmetic. Two long calls carry 1,443,050 shares of delta — the block would be a 62% match, clearly wrong. A $245/$300 debit spread carries 882,350 — the block is a 99.1% match. So the $300 leg is short, the real risk is the $24.7M net debit rather than $35.2M, and the upside is capped at $300 with a breakeven of $252.06.

The timing is worth knowing: NVIDIA has not confirmed its next earnings date, but its own filing cadence — quarter-end October 25 plus a verified 24-day lag, seen three times — points to November 18. If that holds, the December expiry contains the print with a month to spare. → The 62% versus 99% test, worked through

🧪 Q — the one trade that wanted upside, and a divergence worth studying

Qnity Electronics is the specialty-materials business spun out of DuPont, supplying chemicals to chipmakers. It separated on November 1 last year and joined the S&P 500 on its first regular-way trading day. Most readers will not know the ticker.

Three legs, all at 12:54:54, all May 2027: buy 3,400 of the $140 call at $18.98, sell 3,400 of the $170 at $10.37, and sell 3,400 of the $90 put at $7.43. The call spread costs $8.61 a share; the put sale brings in $7.43. Net cost $1.18 a share, or $401,200 — against a maximum profit of $9.8M above $170.

But $401,200 is not the risk, and that is the lesson. Below $90 the seller must buy 340,000 shares — $30.6M of stock at that strike, and $340,000 more for every dollar below it. That obligation is what buys the cheap upside. Package delta is +117,470 shares, roughly $14.6M of long-equivalent exposure, and no stock block printed alongside, so it is not shown to be hedged.

What makes it interesting is the setup. Full-year guidance has been raised three times this year — from $3.55-3.95 to $4.40-4.60 — with four straight beats, while the stock fell ≈30% from a $175.64 peak close in June. Estimates up, price down. And the strikes are not round numbers: $140 is roughly the August 12 close, $170 is the June high zone, and $90 is a round-trip to January. The company's own buyback has been spending at an average of $124.12 — essentially today's price.

The honest counterweight: both earnings events inside the option's life arguably work against a long call spread, and a new chief financial officer starts on October 1 before the print that frames next year's guidance. → Why $401,200 is not the risk

🧊 DRAM — a delta-neutral volatility trade with an uncovered tail

Two legs in the Roundhill memory fund: sell 20,000 of the March-2027 $45 put, buy 10,000 of the $55. An exact 2:1 ratio, hedged with a 50,000-share block against a computed +50,500 of delta — a 99.0% match. So it is a volatility and skew trade, not a directional view.

The risk is specific and it should not be buried. That fund closed at $44.85 on July 29 — below the short strike — having fallen 31.9% from $65.86 in four weeks, before rebounding 16.70% in a single session. This structure sells two uncovered puts, in double size, at a level the fund traded through six weeks ago. A delta hedge neutralises direction; it does nothing about a gap.

Worth knowing what you would be buying: the fund is actively managed, is not thinly traded, and is built from ≈49% Treasury bills and ≈37% total return swaps. Effective exposure is ≈73% in three names — Micron, Samsung and SK hynix. And a large buyback supporting one of them expires November 19, four months before this option does. → The 2:1 ratio, the hedge, and the July floor

🛠️ Theme 5: The $416 Million That Turns Out to Be $20 Million

🚀 SPCX — the biggest number on the board is bookkeeping, and here is the arithmetic

SpaceX has only been a public company since June 12, and today its options were the largest thing on the tape: 100,562 put contracts crossed the exchange floor across 16 strikes for $416.3M of premium, all expiring September 18. Nothing else came close. It is also, we think, the least interesting trade of the day — and working out why is genuinely useful.

Three things say immediately that this is not a bet. Every deep-in-the-money leg traded below its intrinsic value: with the stock near $148.67, the $200 put is worth $51.33 if you exercise it on the spot, and it traded at $51.20. Nobody pays less than the guaranteed value of a contract because they expect the stock to fall. Second, the ladder reaches a $300 strike, and the highest price SpaceX has ever traded in its entire public life is $225.64. And third, there are no call legs and no stock hedge anywhere in the program, because a package like this hedges itself.

Then the arithmetic settles it. Take each tranche and split the contracts into two halves of equal size. The split exists, and it is unique:

Time	Gross premium	The two equal halves	Real net risk
14:24:32	$18.6M	1,300 at $230 vs 700 at $210 + 600 at $215	$2.30M
14:49:20	$164.0M	15,000 at $200 + 504 at $230 vs 15,000 at $205 + 504 at $210	$6.49M
15:05:24	$114.6M	9,600 at $205 vs 5,000 at $200 + 700 at $210 + 600 at $215 + 3,300 at $230	$6.70M
15:24:42	$54.8M	5,000 at $200 vs 5,000 at $205	$2.50M
15:29:36	$54.0M	5,000 at $200 vs 5,000 at $205	$2.50M
Total	$416.0M		$20.49M

In four of those five tranches the premium difference between the two halves equals the strike-price difference to the dollar — zero error. That is not a coincidence and it is not a rounding artifact. Equal sizes mean the share obligations on the two halves cancel exactly, whatever SpaceX does. The only thing that does not cancel is the gap between the strike prices, and that gap is $20.5M. The other $395.8M is intrinsic value going out one door and coming straight back in the other.

The two late tranches make it plainest: 5,000 puts at $200 against 5,000 puts at $205, priced $52.30 against $57.30, then $51.54 against $56.54. Both pairs are exactly $5.00 apart, the exact distance between the two strikes. That is the price of a $5-wide spread traded at its floor value, done twice.

Two more things a reader should take away. The reported legs were badly incomplete — the feed showed 7 legs and 28,900 contracts where the floor traded 100,562 across 16 strikes, omitting the whole low-strike side including 23,000 contracts at the $70 strike. And we cannot tell you whether this opened or closed anything, and we are not going to pretend otherwise. The usual test — trade size larger than existing open interest proves new positions — does not work on this kind of flow, because these prints historically create no open interest at all. The evidence for that is sitting in the record: the $215 strike has held exactly 266 contracts of open interest for twelve straight sessions while 1,599 traded today, and the $230 strike has held exactly 1,633 for twelve straight sessions while 5,797 traded. We checked that this is the program and not a data failure: an untraded control strike nearby updated normally, and the tape carries no cancelled prints.

Why now, though? That part has a clean answer. A 319-million-share lock-up tranche freed up yesterday and 59.1 million more today, and short interest in SpaceX runs at 9.98% of the tradeable float but only 1.36% of the shares outstanding — the borrowable pool is tiny next to the company. Every unlock loosens borrow. Then the index rebalance effective September 21 forces index funds to trade the September 18 closing auction, the very session these contracts expire into, with reported estimates of $12.4B to $15.5B of buying against a 1.85-billion-share float. Freshly unlocked stock, tight borrow, and a forced closing auction is exactly the environment in which a financing package like this shows up. It is positioned around borrow and the calendar, not around a view.

And this is the part worth remembering. The same fingerprint has now appeared four times in four weeks in this one name — August 20, August 25, September 1 and today — and three of the four target this same September 18 expiry, clustering on the same strikes. The $210 and $230 lines show up in three of them. The September 1 version, at $100.9M reported, resolved to no net position at all. Four separate desks do not independently pick the same strikes and the same expiry four times in a month. That is one position being re-papered on a schedule. If you see a headline tomorrow describing $416M of SpaceX put buying, you now know what it was. → The full ladder, the balanced partition, and the open-interest evidence

📅 Upcoming Catalysts — and Which Expiry Actually Holds Them

Catalysts and option expirations are two different calendars. Confusing them is one of the most common ways retail traders lose money following flow, so they are kept separate here.

Date	Event	Which of today's expiries contains it?
Sep 9-10	SpaceX share unlocks — 319 million shares freed yesterday, 59.1 million more today   (as reported, not read from the prospectus)	The structural reason SPCX's floor package appears now
Sep 10 (today)	August producer prices +5.4%, up from 4.8%   — hike odds moved toward ≈70%	Already in the price of everything below
Sep 11, 08:30	August consumer prices   — the swing vote before the Fed	Inside SPY's Sep-18 and QQQ's Sep-30 legs
Sep 11 (after close)	Nasdaq-100 September rebalance announcement	Inside QQQ's Sep-30 leg
Sep 15	New 50% Canada tariff takes effect	Inside XLY's Dec-18 spread
Sep 15-16	Federal Reserve decision with a fresh dot plot   — ≈70% odds of a hike	Two days before SPY's Sep-18 expiry
Sep 16	August retail sales, released the morning of the decision	Inside XLY's Dec-18 spread
Sep 18	Quarterly expiration + the S&P 500 rebalance trade in the close	This is where every SPY leg expires — and every SPCX leg
Sep 21	S&P 500 and Nasdaq-100 rebalances effective at the open — reported estimates put $12.4B to $15.5B of forced buying into SpaceX's 1.85-billion-share float	After the SPY and SPCX legs have expired
Sep 30	Micron fiscal Q4 earnings, company-confirmed	Inside MU's Nov-20 spread
Sep 30	—	QQQ's hedge legs expire
Oct 21	Tesla earnings (≈17.9% of XLY)	Inside XLY's Dec-18 spread
Oct 27	Royal Caribbean earnings   (also 1.88% of XLY)	Inside RCL's Dec-18 spread
Oct 27-28	Federal Reserve decision	Inside the Dec-18 expiries
Oct 29	Amazon earnings (≈24.6% of XLY)	Inside XLY's Dec-18 spread
≈early Nov	Qnity third-quarter earnings (estimated, NOT confirmed)	Inside Q's May-2027 structure
≈Nov 4	Arm earnings (estimated, not confirmed)	Inside ARM's Jan-2028 LEAP — one of ≈six it spans
≈Nov 18	NVIDIA earnings — derived from filing cadence, not confirmed	Inside NVDA's Dec-18 spread
Nov 19	A large buyback supporting a top DRAM holding ends	Four months before   DRAM's Mar-2027 expiry
Dec 8-9	Federal Reserve decision	Inside the Dec-18 expiries
Dec 16	November retail sales — the only hard holiday read	Two days before XLY's expiry
Dec 18	Quarterly expiration	NVDA, RCL and XLY all expire here
Oct 1	Qnity's new chief financial officer starts	Before the print that frames next year's guidance
≈late Feb 2027	Qnity full-year results and initial 2027 guidance	Inside Q's May-2027 structure
Mar 2027 / Jan 2028	No scheduled events this far out	DRAM and ARM are cycle and volatility bets, not event bets

Note how many dates are estimated rather than company-confirmed. Only Micron's is confirmed. We flag that in each article rather than presenting a guess as a date.

One name on today's board has no company event at all, and its context is structural instead. SpaceX has nothing on its own investor calendar inside the September 18 expiry — no earnings, no filing deadline. What it does have is a sequence of share unlocks, the latest yesterday and today, into a market where short interest is 9.98% of the tradeable float but only 1.36% of the shares outstanding. The borrowable pool is tiny relative to the company, so each unlock loosens borrow. Layer on an index rebalance effective September 21, which forces index funds to trade the September 18 closing auction, and you have the environment in which a floor-crossed financing package of this size appears. It is not positioned for an event. It is positioned around borrow and the calendar — which is a different thing, and worth understanding.

🎯 For Four Kinds of Reader

🚀 The YOLO Trader

Be honest about what today offers you: very little that is urgent. Almost every large print was negotiated — floor blocks, auctions and crosses, all with a willing counterparty who agreed on price in advance. Nobody was rushing.

The trades with clean directional content are XLY's bear put spread, NVDA's call spread and Q's funded call spread, and notice that all three are defined-risk on the upside leg rather than naked long options. That is not an accident. It is what people do when they want exposure without betting the outcome on timing. If you cannot resist, size at 1-2% and copy the structure, not just the direction.

And whatever you do, do not chase the biggest number on the board. The $416M of SpaceX puts is the sort of headline that gets passed around as a whale bet. It is a $20.5M bookkeeping package with no directional content at all, and the strikes go up to $300 on a $147 stock.

And read yesterday's open-interest review above before you copy anything. We led that digest with a confident MSTR reversal that the next morning's data flatly refuted.

📈 The Swing Trader

The tradeable observation is the calendar, not the direction. Four macro events land in six days — inflation tomorrow, a tariff on the 15th, the Fed on the 16th with retail sales the same morning, then quarterly expiration and a rebalance on the 18th. SPY's entire structure expires two days after the Fed. Treat mid-September price action as event-driven rather than trend.

The most useful single fact for you: the index has been range-bound for three months and is only +0.48% since June 1 while the 10-year real yield rose 39 basis points and inflation breakevens fell. That is flat price on a compressing multiple — not a rally, and not a crash, but a market getting more expensive on the numbers that matter while going nowhere.

💰 The Premium Collector

Today was your day, and also your warning. Six of the eleven structures are credits, harvesting a variance premium running around 2×. That is a real edge and today's flow is professionals taking it.

Now the warning, in two parts. First, the credit is not the exposure. Micron's desk collected $31.6M against $148.4M of capped risk — a 1:4.7 ratio. Know that number before you admire the income.

Second, look at where the strikes are. Micron's short $1,100 sits 12.4% below its own 52-week high, about two average daily ranges away, with confirmed earnings inside the expiry. The memory fund sold two uncovered puts at $45 when it closed at $44.85 six weeks ago. Selling premium at levels the market has already broken is a different trade from selling remote ones, and the option chain prices them differently for a reason.

The structure worth studying is QQQ's financed hedge: a short call spread paying for two long puts, net ≈$1.1M. That is how you buy protection without spending cash, and it scales down to a retail position.

🌱 The Beginner

Five things from today are worth more than any trade idea.

Gross is not net. Today's flow was $557.5M gross but only $142.2M net. A spread has two legs; adding them counts the same dollars twice. Always ask which number you are being shown.

And sometimes even the net is not the risk. The SpaceX ladder printed $416.3M and risks about $20.5M — a 20-fold difference, provable from the strike prices alone. A very large number is not the same thing as a very large bet.

A credit is not a profit. When someone "collects $31.6M," that is the most they can make — and the risk behind it can be several times larger. Selling options means getting paid up front to accept a risk later.

What you see in a feed is a report, not the trade. NVDA was displayed as two call purchases. The 890,000-share hedge that printed 44 seconds later proves one leg was short, and the real risk was $24.7M rather than $35.2M. That is the fourth session running where the feed and the tape disagreed.

A confident label is still only a label. Read the open-interest review above. We published a HIGH-confidence read on MSTR yesterday, backed by a matching strike, expiration, timestamp and size — and it was wrong. Only the next morning's open interest settles whether a position went on or came off, and it never settles direction at all.

⚠️ Risk Control — Read This Part

Do not follow unusual options activity blindly. Today's board and yesterday's scorecard make the case better than we could.

• We were wrong yesterday, on our lead story. The MSTR reversal did not happen. Two of sixteen reads inverted on the next morning's data. If a source never shows you its misses, you cannot judge its hits.
• You cannot see the whole position. A desk selling calls may own the stock, may be hedged elsewhere, or may be unwinding something built months ago. Open interest tells us whether contracts were created or destroyed — never who bought.
• Negotiated flow expresses no urgency. Floor blocks, auctions and crosses all have a counterparty who agreed on price in advance. That is position management, not conviction.
• Six of the eleven trades today were selling insurance. That works until it doesn't, and the thing that breaks it is exactly the event everyone is watching. The Fed chair has said he reads low volatility as evidence policy is too loose.
• Size below existing open interest proves nothing about intent. On several legs today we genuinely cannot tell open from close, and we say so rather than guessing. Come back tomorrow around 06:30 ET.
• Patience beats participation. There is no requirement to trade any of this. The most useful thing today's flow tells you is a calendar fact — four macro events in six days, and desks positioning around the insurance rather than the direction. That is information you can use without putting on a single trade.

Never risk more than you can lose entirely. Options can and do expire worthless.

Every figure in this digest was verified against the OPRA tape, official open-interest snapshots and the equity tape. Where something could not be proven, we say so rather than filling the gap with a guess. Prices reflect an open trading session on September 10, 2026.

Not investment advice. Options involve substantial risk and are not suitable for every investor.