AInvest Option Flow Digest — 2026-08-26 · $217M Across the Tape

Eight names, nineteen legs, $217.2M of premium crossing the tape — six packages paying out $71.7M, two collecting $22.5M.

But the number that should shape how you read today is a different one: on two of the eight — and they are the second and third largest tickets on the board — we cannot tell you which side initiated, and we are not going to pretend otherwise. Both printed as negotiated blocks, where the reported buy/sell tag is a bookkeeping convention rather than evidence. On INTC that ambiguity is worth $20.7M and flips the trade from bullish to bearish depending on which way it runs.

That is not a gap in the data. It is what a negotiated block is, and any service that hands you a confident arrow on one is guessing.

But the dollars are not the interesting part. Today gives us a clean way to sort ambition from patience: compare each trade's target to the range the options market itself is pricing.

• Microsoft's rolled call spread breaks even at $555.24. The November implied-move upper band is $556.40. It needs Microsoft to finish at the very top of the market's own range just to get its money back.
• Honeywell Aerospace's ratio spread pays most at $210. The December band tops out at $203.18. Its best case sits above the band entirely.
• Apple, by contrast, is patient. It is the day's second-biggest ticket and it is unambiguously bullish — but both strikes sit comfortably inside Apple's own implied range. Same direction, far more conservative geometry.
• UPS sits at the other end of the scale entirely. Both of its strikes also sit inside its band — which is precisely what a collar is meant to look like. You sell the upside the market still believes in, and buy protection short of the worst case.

That spectrum — from reaching past the band, to sitting inside it, to selling it — is worth more than any of the dollar figures.

⚠️ One headline was wrong by a factor of ten, and it is a useful reminder: a Microsoft trade appeared as a $20.4M print. It was cancelled. What stands is $2.04M. Third correction of this kind in four sessions.

🔁 OI Review — Yesterday's Predictions, Scored

Yesterday we changed method after a bad Monday: three branches and a capture rate, never a binary fork. This is the first session scored under the new rule, and it mostly worked.

✅ Wide-margin opens were near-exact again

Leg	Predicted	Actual	Capture
SBET   Apr-27 $10C	≈15,039	15,042	99.98%
AMGN   Oct $460P	≈4,128	4,131	99.9%
SCHW   Dec $115C	≈10,365	10,390	99.8%
TQQQ   Jan-27 $90C	≈24,293	24,385	99.6%
BIDU   Mar-27 $80P	≈5,123	5,075	99.1%
EWZ   Nov $43C	≈556,000	572,202	102.9%
VOYA   Nov $110C	≈5,005	5,317	106.2%

✅ And the three-branch framing earned its keep on gold

GLD's Package B unwound, as the leading read said — the August-28 $425 call fell 62,900 → 30,627 and the $440 call 64,100 → 32,371, a genuine close at ≈58% capture. Package A's September $392 call collapsed 27,761 → 2,466 — a 93% capture, landing at the top of the range we published.

And for the third session running, two legs of one structure behaved completely differently: in the January-2027 short straddle the call fell 1,762 (a 25% partial close) while the put rose 6,998 — a clean 100% open. Same package, same millisecond, opposite outcomes.

❗ Where we were wrong — and what we changed

SPCX's four legs came back exactly flat: +0, +0, +1, +0 — against 6,400 contracts traded.

We had published that size exceeded open interest on all four legs and therefore "a minimum of 17 / 216 / 36 / 234 contracts must be new by arithmetic." That floor did not hold. The $180 put printed +1 against a stated minimum of +216. What actually happened was a pure transfer — the position changed hands and essentially nothing was created.

So we are tightening the rule again: an arithmetic floor is only trustworthy when size exceeds open interest by a wide margin. On marginal cases the open-interest figure itself carries too much slack. From today, marginal legs get called marginal, and the transfer branch goes first. You will see that applied to Carlyle below, whose margin is ≈8%.

📊 At a Glance

Ticker	Net Premium	Expiration Type	Catalyst (and its date)	The Option Play	What It Means
BX	$46.5M debit	Monthly (Nov-20)	Q3 earnings ≈Oct 22 ; two FOMC meetings	26,500 in-the-money puts, prior OI of 19	Direction genuinely unresolved   — bought it is protection, sold it is stock acquisition at $137.45
AAPL	$16.06M debit	LEAP (Dec-2028)	A ≈2.3-year horizon — no single catalyst	Bought 5,600 $340 calls, sold 7,500 $260 puts	A patient bullish bet   — upside funded exactly half by selling downside
INTC	$20.7M — sign UNPROVEN	LEAP (Dec-2028)	Trading below its own   $95 offering price   since Aug 11	$140 puts against $150 calls, 5,000 each	A synthetic long OR short   — a floor print cannot tell us, and $20.7M turns on it
HONA	$4.22M debit	Quarterly (Dec-18)	Spin-off still finding its level after an   Aug 6 gap	1×2 call ratio: long 9,994 $175s, short   19,988   $210s	A recovery trade with a ceiling   — best at $210, hurt above ≈$241
MSFT	$2.44M debit	LEAP + Monthly	Nov expiry lands in a stretch with no earnings	Two packages : a call spread rolled up and out, plus a deep tail put	Banked ≈$3.19M, rolled it forward   — and bought catastrophe insurance
CG	$1.32M debit	Quarterly (Sep-18)	FOMC Sep 15–16 is the ONLY dated catalyst	100,000 far-out puts + a $20K wash	Week three of a de-risking programme
UPS	$1.78M CREDIT	LEAP (Jun-2027)	≈10 months; the dividend is the thing to watch	Bought 4,500 $85 puts, sold 4,500 $115 calls	The day's only seller   — a collar, paid to cap ≈450,000 shares
FRVO	$1.20M debit	Monthly (Jan-2027)	GeoBlock 1 first power, Q4 2026	5,000 calls ≈28% above spot	A bounce bet on a broken IPO

Gross premium across all nineteen legs was $217.2M. We publish the net — the actual money at risk — because gross double-counts both sides of every spread.

🔍 The Eight, In Detail

1. 🏛️ BX — $46.5M, and we genuinely cannot tell you which way it points

→ Why the same trade is protection or accumulation depending on who initiated

26,500 November $155 puts at $17.55 — $46.5M, on a contract whose prior open interest was 19. So the position unquestionably opened. What the tape cannot tell us is who started it, and here that matters more than usual:

• If the puts were bought, this is $46.5M of in-the-money downside protection with breakeven only −4.42%.
• If they were sold to open, it is a stock-acquisition structure — the seller agreeing to own Blackstone at $137.45 net of premium. That is a bullish posture and the exact opposite trade.

Both are consistent with a negotiated block. We are not going to pick one.

What is striking either way is the timing. The block was struck within 0.2% of the highest close in three months, after Blackstone ran +26.9% off its June low. Consensus price targets sit 4.5–8.0% BELOW the $155 strike, with only 2 of 24 analysts above it.

And one number from Blackstone's own deck is worth sitting with: private credit returned 0.4% net in the second quarter, and 3.9% over the last twelve months — at or below the 3.50–3.75% policy rate. Realized performance revenue in the credit segment fell 87%.

2. 🍎 AAPL — $16.06M, bought upside funded exactly half by sold downside

→ The trade that looks aggressive and is actually the most patient on the board

One package, three prints, expiring December 2028 — ≈2.3 years out: bought 5,600 $340 calls for $32.1M, sold 7,500 $260 puts for $16.1M. Net $16.06M. Both legs are wide-margin proven opens.

That is a risk reversal: pay for participation above $340 (+8.7%), and fund it by accepting the obligation to buy ≈750,000 shares at $260 (−16.9%, ≈$195M) if Apple falls that far. Bullish, and openly so.

The sizing tells you how it was decided, and it is refreshingly human. The calls cost $32,138,400; the puts brought in $16,080,000 — 50.03%. The put size that funds exactly half the calls is 7,495. They did 7,500. Someone said "sell enough puts to cover half," and that is the whole story of the 7,500.

One detail we want to flag rather than dress up. Every leg printed as a stock-and-options cross, and a 500,000-share block followed 61 seconds later — almost exactly the package's 486,697 delta. It is tempting to read that hedge as proof the buyer has no view. It is not. On a negotiated package the stock leg is how the facilitator lays off risk, which is why a trade this size can print at one agreed level instead of walking the order book. It tells you the dealer is flat. It says nothing about the customer.

And here is why this is the most conservative trade on today's board — measured against what Apple has actually done, not a model. Apple's 52-week range is $224.69 to $344.57. Both strikes sit inside the range the stock has already traded in over the past twelve months. It has been above $340 and below $260 within the year. Compare that with Microsoft's roll, which needs to finish at the very top of its implied range, and Honeywell Aerospace's ratio, which pays best above the top of its band. Same bullish direction, far more margin for error.

The analyst map explains both strikes. Of the 13 individually identifiable published targets, 10 sit at or above the $340 call strike — and none sits at or below $260. The lowest identifiable is Jefferies at $263.66. The sell side clusters above the call and has essentially abandoned the ground beneath the put, which is precisely why that is the leg being sold in size.

⚠️ What the put seller is actually underwriting is regulatory, not cyclical. Three dated proceedings can resolve inside this trade's life: the Supreme Court took up Apple v. Epic on June 30, a judge refused to stay the App Store fee-setting phase on August 11, and both the DOJ and Google have appealed the ruling over the ≈$20B-a-year default-search payment. It is already in the numbers — Services posted $30.7B, its first sequential decline since 2022. That is the risk in exchange for the $16.1M collected.

3. 🔲 INTC — $20.7M, and the sign in front of it is genuinely unknown

→ Why a floor print can hide a $41M swing in what the trade means

Two legs, both 5,000 contracts, expiring December 2028: the $140 puts at $65.50 ($32.75M) against the $150 calls at $24.10 ($12.05M). Reported as sell-the-put, buy-the-call, that is a synthetic long — roughly 483,800 shares of delta and a $20.7M credit.

Reported. Not proven. This was a floor block — negotiated by hand rather than executed on the displayed market — and on those the buy/sell tag is a reporting convention. Reverse it and the identical two legs become a synthetic short and a $20.7M debit. The whole meaning inverts.

And the reverse reading has evidence behind it. The $150 call's open interest climbed steadily for two weeks — 3,421 → 4,572 — while the stock fell. Today's 5,000 lots against 4,572 already outstanding is almost exactly the size that would close that accumulation. Someone giving up and reversing is every bit as coherent as someone doubling down. Tomorrow's open-interest print will tell us whether contracts were created or destroyed, but open interest is side-blind — it will not settle direction. We would rather say that than invent an arrow.

What we can tell you is why these strikes. Intel priced a $20 billion stock offering at $95.00 on August 11 and has traded below its own deal price ever since, hitting $85.24 on August 24. That is the seven-session, ≈15.6% slide, and it is why implied volatility is running near 69% on a large-cap chipmaker. Meanwhile the stock is still +136.5% year-to-date and +255% over twelve months — but −38.7% from its June high of $142.35.

Which is exactly where the strikes come from: the $140 put sits 1.7% below that 52-week high, and the $150 call just above it. They bracket the peak.

⚠️ The sobering part. Of 17 analyst targets from the last three months, exactly one sits at or above $140 — and it was cut from $160 on August 12. None sits at or above $150. On the credit reading, breakeven at expiry is ≈$98.60: Intel must rally ≈12.9% for the trade merely to be worth nothing, and the maximum loss is ≈$49.3M.

4. ✈️ HONA — a 1×2 ratio aimed at the pre-gap price

→ The structure that wants a recovery, but not too much of one

Honeywell Aerospace was separated from Honeywell on June 29 and has only traded since — so there is no year-to-date figure to quote, and anyone who shows you one is splicing in a different security.

One facilitated auction put on a structure worth understanding properly: bought 9,994 December $175 calls at $13.62, sold 19,988 December $210 calls at $4.70. Note the ratio — the short leg is exactly twice the long leg. That is a ratio spread, not a vertical, and it behaves very differently:

• Below $175, the whole $4.22M debit is lost.
• Breakeven $179.22. Maximum profit ≈$30.8M, and it occurs exactly at $210.
• Above $210 the profit shrinks, and past ≈$240.78 the extra 9,994 short calls start losing — without limit, if the desk holds no stock. If it does hold shares from the spin, those calls are covered and this is an overwrite. The tape cannot see stock positions, so both readings stand.

Here is what decodes the strike choice. HONA gapped down 23.2% in a single session on August 6 — from a $203.64 close to a $156.47 close, on roughly five times normal volume — and has chopped between $156 and $170 ever since. The $210 max-profit point is essentially where the stock traded before that gap. So this is a bet on a round trip back to the pre-gap level by December — a recovery trade with a ceiling, not a moonshot.

⚠️ And the counterweight: the December implied-move band runs $124.46 to $203.18. The trade's best case sits above the top of that band.

5. 🪟 MSFT — two packages, pointing opposite ways, and both are real

→ A cancelled headline, a proven roll, and ≈$3.19M actually banked

Package one is the correction. The feed showed 56,000 January-2028 $250 puts — about $20.4M. That print was cancelled. What survives is 5,600 contracts at $3.65 — $2.04M, a strike ≈49% below spot with a delta of −0.03. That is catastrophe insurance, not a view.

Package two is the one that matters, and it is a roll we can prove rather than infer. A four-leg auction, every leg exactly 14,839 contracts, that closes an October-02 $505/$550 call spread and opens a November-20 $545/$625 one.

How do we know it is a roll? The open-interest history. Both October strikes ran from essentially zero to ≈15,000 in a single day, and the tape shows that position being opened on August 17 at $7.82 net — the day Microsoft closed at $480.35, the low of the stretch.

So the round trip is a realised profit: in at $7.82, out today at $9.97. That is +$2.15 per spread — ≈$3.19M, or +27.5% in seven trading days. Someone bought the dip in size and is taking the money forward. The net cost of the roll is only $401K, but that number badly understates it: the credit is recycled profit, and the new position is a $118.7M-wide spread bought for $15.2M, with max profit ≈$103.5M above $625.

⚠️ Now the honest part. Breakeven on the new spread is $555.24 — a 12.2% move. The November implied-move upper band is $556.40. This trade must finish at the very top of the range the options market prices just to break even. A $118M headline and a coin that has to land on its edge are both true at once.

6. 🏦 CG — week three of a programme that now decodes cleanly

→ The cap came off, the floor doubled, and only one thing is on the calendar

This is Carlyle — not Citigroup, which trades under C.

A collar was struck around August 4 at $50.64, a local top and the day before earnings, with the cap 4% up. Then everything went against the cap: Q2 beat, Goldman upgraded the whole alternative-manager group, price targets went to $70 and $73. On August 21 the cap was retired — call open interest collapsed 141,303 → 70,908 — while the floor tripled, put open interest rising 26,587 → 92,204. Today they bought 100,000 more of the same $40 puts.

Read it as: stop being short 141,000 calls into a hawkish September, but keep the crash protection.

Two details make it convincing. Nothing company-specific happens before September 18 — earnings are ≈November 6 and the dividend already went ex — which leaves the September 15–16 Fed meeting, two days before expiry, as the only dated catalyst in range. And the sizing says insurance, not conviction: $1.3M buys roughly $400M of notional, about 0.33% for 23 days. Someone who genuinely expected $40 — the year's low, with no analyst target below $45 — would buy the $45s.

⚠️ Today's package also shows a 20,000 buy and a 20,000 sell of the same $52.50 call, a cent apart. That nets to a $20,000 book transfer, not a bullish position.

7. 📦 UPS — the clearest collar on the board, and the one credit we can actually assign a side to

→ Paid $1.78M to cap a position it already owns

One cross, two legs, both 4,500 contracts, expiring June 2027: bought the $85 puts for $1.69M, sold the $115 calls for $3.47M. Net $1,777,500 collected. Both legs are wide-margin proven opens. Intel's package also reads as a credit, but only if its reported side is right — UPS is the one credit on the board whose shape is unambiguous, because a long put against a short call at equal size is a collar whichever way you read the tag.

Equal size, long put and short call, is a collar. Against stock it covers ≈450,000 shares, about $47.8M of UPS: everything above $115 is given up, everything below $85 is protected, and they are paid to make that swap. Counting the credit, the effective floor is $88.95 and the effective cap $118.95.

Look at the asymmetry, because that is where the craft is. The cap sits 8.3% above the market. The floor sits 20% below it. They sold something with a real chance of being reached and bought protection against something much further away — and the gap between those two probabilities is the $1.78M.

Both strikes sit inside the market's own June-2027 range ($77.45–$134.75). That is exactly what a collar should look like, and it is the opposite of the reaching we see elsewhere today.

⭐ And here is the arithmetic that makes the collar reading convincing. Three dividend ex-dates fall before this option expires — around November, February and May — totalling $4.92 per share, or ≈$2.21M on 450,000 shares. That is MORE than the $1.78M credit they collected. A holder who collars keeps that stream; someone short the stock would have to pay it. The dividend is the bigger prize, and it points at a long position being protected.

⚠️ Except the dividend being protected may itself be under strain — and this is the part worth your attention. UPS yields ≈6.17%, but the trailing payout ratio is 121.95%, and second-quarter free cash flow was $166M against $1,356M of dividends paid — 12% cover. The buyback has been zero for five straight quarters, so the shock absorber is spent. Debt has gone $25.6B → $28.7B while cash fell $6.8B → $4.7B. And the dated tell: February 2026 was the first February in at least three years with no raise — seven flat quarters now. The company's own pledge is that the dividend has been "maintained or increased" since 1999, which is a floor, not a promise of growth. The ≈February 2027 declaration falls inside this option's life, and it is the single event that resolves the thesis.

⚠️ The caveat we cannot resolve. This is a negotiated cross, so we cannot prove who initiated, and the tape cannot see whether anyone holds the underlying shares. Against stock this is a conservative, income-minded collar. Without stock, the same two legs are a bearish bet — a synthetic short with a distant floor. A collar is the more natural reading for a long-held income name, but that is our judgement, not something the print proves.

8. 🌋 FRVO — a bounce bet on a broken IPO

→ What ≈94% implied volatility does to a call 28% out of the money

5,000 January-2027 $20 calls at $2.39 — $1.20M — on a contract with 84 contracts of prior open interest.

Fervo Energy is a geothermal developer that came public on May 13 at $27. It now trades at $15.42 — down 42.6% from the IPO price and roughly 64% from its high, at all-time lows, after a second quarter where earnings missed by $0.29 and revenue was $113,000.

So why own a call? Because there is a real, dated binary inside the option's life: GeoBlock 1 is scheduled to begin generating test power in the fourth quarter of 2026 — the first commercial-scale enhanced-geothermal power. And cash is not the constraint: $2.1B on hand covers the capex plan without a forced raise.

The mechanism to understand is volatility. At ≈94% implied volatility, a strike 28% above the market still carries a delta near 0.46 — the market is pricing an enormous range, which is what makes this a coin-flip rather than a lottery ticket.

⚠️ The honest counterweight: short interest is small and falling, so there is no squeeze to lean on — this looks like long liquidation. And elevated put volume was flagged in the same name today, meaning other flow leans the other way.

⏳ Tomorrow Morning Is the Scorecard

Test	Full capture (≈100%)	Partial (40–90%)	Flat = transfer
BX   Nov $155P	≈26,519	—	(prior OI 19 — wide margin, floor reliable)
AAPL   Dec-28 $340C / $260P	≈8,048 / ≈9,473	≥+3,152 / ≥+5,527 floors	—
HONA   Dec $175C / $210C	≈10,145 / ≈20,185	—	(prior OI 151 / 197 — wide margin)
MSFT   Nov $545C / $625C	≈16,091 / ≈15,029	≥+13,587 / ≥+14,649 floors	—
MSFT   Oct $505C / $550C	—	a fall of ≈14,839 confirms the roll's close	flat would refute it
MSFT   Jan-28 $250P	≈8,242	≥+2,958 floor	—
UPS   Jun-27 $85P / $115C	≈4,751 / ≈5,135	≥+4,249 / ≥+3,865 floors	—
INTC   Dec-28 $140P	≈5,466	≥+4,534 floor	—
INTC   Dec-28 $150C	≈9,572	40–90% capture	≈4,572 unchanged — listed first, margin only ≈8.6%
CG   Sep $40P	≈192,210	40–90% capture	≈92,210 unchanged — listed first, margin only ≈8%
FRVO   Jan-27 $20C	≈5,084	—	(prior OI 84 — wide margin)

Note what changed in that table. Carlyle's margin is narrow, so we lead with the transfer branch rather than asserting a floor — precisely the mistake that cost us on SPCX yesterday. And Microsoft's October legs are the one place where a decline is the confirming outcome, because they close a position rather than open one.

👥 How to Read Today, By Who You Are

🎲 The YOLO trader. FRVO is the only thing here built for you, and it is a genuine binary: a company at all-time lows with a dated technical milestone in the fourth quarter and enough cash to reach it. But be clear-eyed — ≈94% implied volatility means the option is expensive precisely because the outcome is uncertain, short interest is falling rather than building, and same-day put flow leaned the other way.

📈 The swing trader. Two dated items sit above everything else. Intel's ≈October 22 report is the first real test of whether a company that beat by $1.7B and still fell 7.9% can change its own narrative — and it trades below its $95 offering price going in. UPS's ≈February 2027 dividend declaration is the cleanest binary on the board: seven flat quarters, 12% cash cover, and a collar already positioned around it. Then the ordinary calendar: the September 15–16 Fed meeting sits two days before Carlyle's expiry and is the only dated catalyst in that window, then Blackstone's Q3 around October 22 inside its November expiry. One genuine oddity to diarise: the Nasdaq-100 annual reconstitution takes effect at the December 21 open, which means index funds trade the December 18 close — the exact day Honeywell Aerospace's options expire. Note also what is absent — Apple's and Microsoft's positions expire into stretches with no scheduled catalyst at all, which is the point of them.

💵 The premium collector. Six of eight desks were paying today, which usually means the cost of optionality is being bid — generally a better moment to sell than to buy. And the seventh desk is you. UPS is the cleanest worked example of your trade on the board: a collar that gets paid $1.78M to cap ≈450,000 shares at $115 while buying a floor at $85. Note the asymmetry, because it is the craft — the cap is only 8% away, the floor is 20% away. They sold something likely to be tested to buy something that probably will not be, and pocketed the difference. Three other things should interest you. Apple's package is half your trade already: someone sold 7,500 December-2028 $260 puts to fund calls — a cash-secured put sale at a 17% discount, on a 2.3-year horizon, with the size chosen purely to cover half the call cost. Before you copy it, price the thing being underwritten: a 2.3-year put is a bet that the App Store and default-search cases do not re-rate the company, and Services has already posted its first sequential decline since 2022. If the Blackstone block was a put sale, someone very large is doing exactly what you do — getting paid to agree to own a stock at $137.45. And HONA's ratio is the trade you already know: the second short call is only "unlimited risk" if it is uncovered. Against stock, it is an overwrite with a $210 target.

🌱 The beginner. Start with INTC, because it teaches the most important limit of this whole exercise: sometimes the honest answer is "we don't know." A $20.7M trade printed today whose meaning flips completely — bullish to bearish — depending on which side started it, and the tape simply does not record that. Anyone showing you a confident arrow on a negotiated block is guessing. Then read AAPL for a habit worth building: check the target against the market's own expected range before you judge a trade. Three desks today were bullish. Two aimed past the top of what the options market prices; Apple aimed inside it. Same direction, very different odds — and you can run that check on any trade, including your own, in under a minute. Then read MSFT, which teaches two things in one name. First, a headline number can be wrong by ten times — a 56,000-contract print appeared and was cancelled; the real trade was 5,600. Second, and more useful: the roll shows what a professional actually does with a winner. They did not simply take the profit and leave. They closed a spread up 27.5% in seven days and immediately committed the proceeds to a bigger, longer, higher-strike version. Then read BX, which teaches the hardest lesson — the same trade can be protection or accumulation depending on which side started it, and often nobody outside the deal can tell.

⚠️ Risk, Plainly

A large trade is not a recommendation, and it is not a prediction. Every desk here has information, hedges and mandates you cannot see. The tape does not carry the broker, the customer, or any stock held away from the print. Everything we call intent is an inference from structure, size versus open interest and strike geometry — never a disclosure.

Four habits worth more than any flow alert:

Check whether the print survived. Cancels are common, and today one of them cut a headline by 90%.

On a negotiated block or a facilitated auction, nobody can prove who initiated. Where our tables say "reported side," that is a convention, not evidence — and Blackstone shows how completely that can change the meaning.

Watch the sizing, not the notional. Carlyle's $1.3M covers $400M of stock. That ratio tells you it is insurance. A conviction bet looks different.

Compare the target to the market's own implied range. Two structures today are aimed past the top of their bands, one comfortably inside. That does not make any of them wrong — it does tell you where the market disagrees, and it is the single fastest sanity check you can run on any trade, including your own.

On a negotiated block — a cross, an auction, or a floor trade — nobody outside the deal can prove who initiated. Today that ambiguity covers $67M across Blackstone and Intel. Where our tables say "reported side", that is bookkeeping. Treat any service that gives you a confident direction on one of these as guessing.

A hedge printing alongside a block does not mean nobody has a view. On a negotiated package the stock leg is usually how the facilitator neutralises itself, not evidence that the buyer is indifferent to direction. We nearly read Apple's 500,000-share leg the wrong way round, and it would have inverted the whole conclusion.

Nothing here is investment advice. Options carry the risk of total loss of premium, and short option positions carry risk far beyond the premium collected.