📊 Ainvest Option Flow Digest — 2026-08-05

Thirteen names, ≈$219M net — a hedging board, two clean rolls, and not one lit sweep all day

✅ Updated 2026-08-06 pre-market: every leg on this board has now been resolved by next-day open interest. Five names inverted — TSM, XLP, WMT, SPY and GDX. NVDA's $65.3M ambiguity resolved as an add. See the ✅ RESOLVED section near the foot of this issue for the full per-leg table.

---

🎯 Quick Read

Not a single trade today was a lit, liquidity-taking sweep. Every package on this page was negotiated — a block cross, a facilitated auction, or worked on the exchange floor. That alone tells you what kind of session it was.

• ≈$219M net across 13 names, and the tilt is defensive. Credit puts, sector puts, gold-miner put spreads, and an index collar with a ceiling less than 1% away.
• ⭐ The thread that ties it together isn't in any one trade — it's in the two prices. HYG, the high-yield credit ETF, sits ≈1.1% off its 52-week LOW. SPY sits ≈0.6% off its HIGH. VIX is 15.59. Credit is not confirming equity, and today's flow looks like people acting on that.
• HYG bought 75,000 January $65 puts for fifteen cents each. That is ≈$596M of credit notional hedged for $1.125M — about 0.19% of what it protects. The strike is 18.2% below spot, so it pays only in a genuine credit event. Cheap insurance that usually expires worthless, and it should be described that way.
• SPY crossed a $32.2M collar — which turned out to create no new position at all. Buy the October $707 put, sell the $783 call, 20,500 each; floor 8.9% below, ceiling barely above spot. Next-day open interest was flat on both legs, so an existing collar (built July 31) simply changed hands. Separately — and unrelated — somebody bought $53.0M of deep in-the-money August $600 calls at a 0.995 delta. Those behave like 298,620 shares, with under a dollar of time value. That is financing, not a directional bet.
• ⭐ TSM is the day's biggest commitment at $40.3M, and the timing is deliberate: TSMC's July monthly revenue has not been released yet — the company's own page shows January through June posted and July blank. It is due roughly August 10–13, inside the August 21 expiry the risk reversal was placed in. June was +67.9% year over year, so this print is the first real test of an above-consensus third-quarter guide.
• WMT sold 19,877 deep in-the-money $119 puts against a prior open interest of TWO. As clean a new position as this tape produces — confirmed next day at 2 → 20,950. (The paired $105 put buy, however, resolved as a close, so the "floor" in this package does not exist.) Selling a deep in-the-money put is a synthetic long — an effective purchase price near $111.73 — so despite being four put trades, the package is bullish, with delta of +1,059,688 shares. Earnings are estimated for August 20, one day before expiry.
• INTC sold $17.9M of deep in-the-money November $70 calls — off a position built in late July. Open interest at that strike went 8,216 → 73,263 in a single session on July 28, days after Intel's July 23 beat. Today's 5,000 lots are ≈6% of it. That reads as a trim of a winner on a stock up 510% this year, not a fresh short — and the next-day open interest confirmed exactly that, falling 83,141 → 78,127 (−5,014) against the 5,000 print.
• ⭐ NVDA took $65.3M of deep in-the-money October $180 calls — onto a position built in a single July session. Open interest at that strike went 2,126 → 90,997 on July 23, then sat still for two weeks. Today's 14,200 lots are ≈15% of it. At a 0.87 delta with under $5 of time value, these behave like ≈1.24 million shares, not a speculative bet. Resolved: it adds. Open interest rose 91,722 → 105,697 (+13,975), ≈98% of the print — $65M of genuinely new exposure, not an obligation being retired. Worth noting: NVIDIA is +4.73% today on the SpaceX exclusive-Nvidia announcement, and SpaceX is also on this board with a synthetic short. Two connected names, opposite positioning, same session.
• ⭐ WYNN is the cleanest roll we have seen in weeks — and it is now proven, September open interest collapsing 16,627 → 235. Sell 16,500 September $90 calls, buy 18,000 December $90s, same second. The September strike had held ≈16,600 contracts since May — and they sold 16,500 of them, 99.2% of everything open there. Package delta moves by just −20,730 shares. The view didn't change; the clock did. They paid $9.2M for three more months.
• HUBS paid 8.1% of the share price for 16 days of protection — hours before tonight's earnings. $19.90 for a $240 put with HubSpot at $246, and the company reports after the close today (confirmed). That reframes the price: it is not an anomaly, it is what protection costs on the afternoon of a print, in a stock down 52% from its high with a target range running $180 to $660. Breakeven sits ≈10.5% below spot — and implied volatility collapses tomorrow either way.
• CC bought a bull call spread into a 17.6% collapse. Chemours reported Tuesday, beat on EBITDA at $247M, and fell anyway. Both legs are proven new — the $14 strike had zero prior open interest.
• SPCX is SpaceX — the stock, listed June 12 at $135, not an ETF. Somebody put on a synthetic short at $115 covering ≈750,000 shares. It could be a holder hedging, or a bearish bet; the tape cannot tell you which.

The honest frame: hedges are not predictions. Every defensive position on this page is somebody paying to be wrong most of the time. That is what insurance is, and it is a very different thing from a forecast.

---

🔁 OI Review — Last Session's Provisional Flags, Now Resolved

Yesterday's board has been through the next-day open-interest test, and it moved a lot of meanings. Four names turned out to be position exits wearing headline-flow clothing.

🔄 Inversions

🔄 SPY — $43.1M of "call buying" was a short being covered. Open interest on the August-21 $775 leg fell 82,982 → 56,680. Contracts are destroyed only when both sides close, so that leg was a buy-to-close. Corrected: only the $785 leg ($28.1M, 4,062 → 79,505) is genuinely new upside. Our "≈$3.06B of new directional exposure" overstated it.

🔄 TSM — the "new bear call spread" was an existing BULL call spread being closed. Open interest fell on both legs — $425C 11,575 → 6,553, $475C 12,289 → 6,271 — with barely 300 contracts of non-package volume all day. Corrected: the $44.4M tail describes risk retired, not risk taken on. The September $400 call also confirmed a close. The whole TSM session was an unwind.

🔄 META — the $8.0M Jan-2028 $1,000 call was a buy-to-CLOSE. Open interest fell 15,440 → 13,651 against a 2,200 print. Corrected: a desk covering a short call, not paying for a 72%-move lottery ticket.

⚠️ One prediction we got wrong, said plainly

⚠️ GOOGL — we predicted the $375 open interest would fall to ≈12,800. It rose to 66,176. We also said the check "should be conclusive." It was not. Only 11,730 contracts of non-package volume traded there, which bounds the package's own contribution between −939 and +22,521 — it created ≈zero net open interest, the signature of one side opening while the other closes. Open interest cannot say which side was which. The roll is not refuted — the July 29 accumulation tape (55,544 contracts at a weighted $7.274, 87–93% across the offer, no companion leg) still favours it. But the ≈$76.7M realized gain now rests on that pre-trade evidence alone, and we have corrected "effectively ruled out" to "disfavoured, not ruled out."

✅ Confirmations

• MCHP — the roll down and out is proven, and so is the ≈$102.6M loss. September $75C 108,624 → 56,577; December $65C 11,235 → 66,239. Both published predictions landed. Note the widely quoted prior open interest of 7,000 would have produced a confident, exactly backwards conclusion.
• QQQ — sell-to-close confirmed; the de-risking roll holds and the net-short alternative is dead.
• SLV — the bust is independently confirmed: open interest rose only +2,643, not the ≈30,000 a live 30,009-lot would have created.
• EEM, RKT, PLTR, XYZ, MSFT, MMM, MRNA — all confirmed open as published.
• ⚠️ SNDK opened, but only about a third of it — 2,797 → 3,814 against a 2,750 print, so net-new short-put inventory is ≈1,017 contracts, not 2,750.

What changed

The dollar totals are untouched — open interest never moves the cash. What moved is what those dollars mean. SPY's $43.1M and META's $8.0M were obligations being retired, not exposure being bought. TSM's $30.2M credit was a desk being paid to leave positions it already held.

The standing lesson, sharper than usual: "bought" can mean opened a bet or closed an obligation, and a big credit can mean risk sold or risk abandoned. Only the next morning's open interest separates them — and when it can't, the honest answer is to say the check came back inconclusive rather than claim the win.

---

📋 At a Glance

Net, not gross. Every figure is net — premium paid minus premium collected. SPY's two trades gross far more than the $20.8M shown; WMT's four prints gross $16.5M and net to a $12.5M credit.

---

🔍 Three Worth Your Time

🛡️ HYG — $596M of credit protected for $1.125M

At 14:05:40 somebody bought 75,000 January-2027 $65 puts at fifteen cents each, negotiated on the exchange floor. Prior open interest was 1,842, so this is overwhelmingly a new position — and the next-day snapshot confirmed the full size, 1,842 → 77,359.

Run the arithmetic, because it is the whole point. 75,000 contracts is 7,500,000 shares of notional ≈ $596M of high-yield credit exposure. The cost was $1,125,000 — about 0.19% of what it protects.

The strike sits 18.2% below the $79.48 price. High-yield ETFs do not fall 18% because rates drift; they fall 18% when credit spreads blow out — a genuine risk event. This is not a rate view. That distinction matters, because retail readers routinely read a bond-fund put as a bet on the Fed.

Be clear-eyed about the odds: this expires worthless in almost every scenario, and the buyer knows that. It is a line item, not a forecast. What makes it worth your attention is the company it keeps — HYG is ≈1.1% off its 52-week low while the S&P sits ≈0.6% off its high. Credit is the market's early-warning system, and somebody is paying, cheaply, for the possibility that it is right.

One honest gap: we could not source current high-yield spread levels, so we cannot tell you whether HYG's softness is a credit story or a rate story. We would rather say that than guess.

🛒 WMT — four put trades that add up to a bullish bet

The tape shows Walmart puts being sold in three clips and bought in one, all in the same second. That looks bearish. It is the opposite.

The three sales are 19,877 contracts of the August $119 put — against a prior open interest of two. Essentially the entire position is brand new, which is rare and worth noting on its own.

Those puts are deep in the money against a $113.04 stock. Selling a deep in-the-money put is a synthetic long: you take on the obligation to buy at $119, but you are paid $7.27 for it, so your effective purchase price is about $111.73 — slightly below today's price. The $105 puts bought against them cap the downside. Net: paid $12.46M to take on Walmart exposure with a floor. Package delta is +1,059,688 shares.

The timing carries real risk. Walmart's earnings date is estimated at August 20 — one day before this expires, and it is not company-confirmed. Selling premium into an event inside your own contract is a deliberate choice, and the seller is being paid for exactly that.

⭐ TSM — the catalyst hasn't happened yet, and that's the point

Taiwan Semiconductor drew $40.3M today: a September $400 call bought outright, plus an August-21 risk reversal — sell the $380 put, buy the $400 call, 10,000 each.

Here is the detail that makes it interesting. TSMC's July monthly revenue has not been published yet. The company's investor page shows January through June posted and July blank. It is due in the usual 8th-to-13th window, roughly August 10–13 — inside the August 21 expiry.

That matters because June printed NT$442,680M, +67.9% year over year, the strongest month of the year, and management guided third-quarter revenue to $44.6–45.8B against a $42.8B consensus — about 5.6% above the Street. The July number is the first hard check on whether that guide is real.

The caveat: all three legs traded below their existing open interest, so we cannot prove any of them opened. And after yesterday — when TSM's entire session turned out to be an unwind once open interest came in — that caveat deserves more weight than usual here.

---

📅 Upcoming Catalysts — and Which Expiry Each One Lands In

⚠️ A catalyst only matters to a contract still alive when it happens. Two names here expire before their own company reports.

On rates, because it cuts against the reflex: the Fed held at 3.50–3.75% on July 29 on a 9–3 vote, and the statement names Hammack, Kashkari and Logan as preferring a quarter-point HIKE. That is the confirmed record. Anyone building a thesis on cuts this month is arguing against the Fed's own minutes.

One transparency note: the official CPI and PPI schedule pages were unreachable when we checked, so those dates follow the usual pattern rather than a confirmed calendar. Do not choose an expiry off them without verifying.

---

👥 Four Ways to Read This Board

🎲 The YOLO Trader

Honestly? There is very little here for you today, and that itself is information. The cheapest-looking thing on the page — HYG's fifteen-cent puts — needs an 18% collapse in high-yield credit to pay. That is not a lottery ticket with good odds; it is insurance somebody is happy to lose money on.

The closest thing to a directional trade is CC's bull call spread, which needs +13.9% for the full payout on a company carrying $4.06B of debt against $671M of cash. Read the balance sheet before the chart.

📈 The Swing Trader

TSM is the one with a dated, checkable catalyst — July revenue lands inside the August 21 expiry, and June's +67.9% sets a high bar. The caution was warranted: the open interest came back split. The $19.75M September call was a short being covered — the fourth straight session of unwinding at that strike — while the August risk reversal genuinely opened. Read the conviction from the risk reversal alone, not the headline $40M.

INTC is the more interesting study. A stock up 510% this year, now below its 50-day average, with two fresh downgrades and a consensus target only ≈6% above spot — and somebody trimming a position they built in late July. That is what taking profits looks like from the inside.

💰 The Premium Collector

GLD is the cautionary tale. A $13.6M credit sounds like income until you price the tail: $46.0M of maximum loss if gold runs above $425 by September, with an FOMC meeting two days before expiry where a hike is live. Credit received is not profit earned.

WMT is the professional version of the cash-secured put you probably already trade — with a bought floor most retail sellers skip, and an earnings print inside the contract.

🌱 The Beginner

Today is a good day to learn one idea: a hedge is not a prediction.

Four of the ten names here are people buying protection — HYG, XLP, GDX, and half of SPY. None of them is forecasting a crash. They are paying a small, known cost so that a large, unknown one cannot hurt them. Most of that money will be lost, on purpose.

And two of today's trades look like the opposite of what they are: WMT's four put trades are bullish, and SPY's $53M of call buying is financing, not a bet on direction. If you take one habit from this issue: before reacting to "puts" or "calls", ask what the position's delta is and whether open interest says anything opened at all.

---

✅ RESOLVED — the Next-Day Open Interest Is In (updated 2026-08-06 pre-market)

The ≈06:30 ET OPRA snapshot reflecting the August 5 close has published, and all 26 legs on this board are settled. Score: 15 opens, 5 closes, 1 partial close, 2 partial opens, 3 pure transfers. Five names moved meaning.

🔄 Inversions

• 🔄 TSM — the $19.75M September $400 call buy was a short being COVERED, not a bullish open. Open interest fell 11,647 → 7,711 (−3,936) against a 5,000 print. It is the fourth consecutive session that strike has been unwound (21,661 → 16,725 → 11,647 → 7,711). Corrected: only the $20.5M risk reversal is a new bullish position — the "$40M bullish combo" framing overstated the commitment by roughly half.
• 🔄 XLP — calendar vs double-long is settled, and it was neither: it was a ROLL. We published the exact test — September open interest falling to ≈5,526 would prove the near leg closed. It printed 5,522, four contracts off. September was wiped out (−80,107) and December opened (+85,032). Corrected: a put roll-out at a true net cost of ≈$6.57M, not a $10.1M new bearish bet. The trader kept the same $78 strike and bought three more months.
• 🔄 WMT — the "$105 floor" was not a floor. That strike lost 15,930 contracts against a 23,702-lot buy, so it was predominantly a buy-to-CLOSE. The likely structure is a short-put roll UP from $105 to $119, which means there is no protective floor beneath the position. The $119 sale confirmed opening (2 → 20,950). Direction still bullish; the risk profile is materially worse than published.
• 🔄 SPY — the $32.2M collar created ZERO new open interest. Both legs printed flat ($707 put −14, $783 call +105) against a 20,500-lot size. It was a pure transfer — the collar itself dates from the July 31 close and merely changed hands. Corrected: no new hedging demand here. The $53.0M August $600 call buy did confirm opening (26 → 3,009).
• 🔄 GDX — the two $67 put sales never opened a short. November fell −17,278 and December −1,193; we said a decline would mean "an unwind/roll on the $67 side layered under two fresh $75 longs, not a clean new spread." Corrected: a roll of protection UP from $67 to $75, with an uncapped payoff rather than a defined-risk $8-wide debit spread. Both $75 put buys confirmed opening (+42,339 and +44,410).

✅ Confirmations

• ⭐ NVDA — the $65.3M question is answered: it was an ADD. Open interest rose 91,722 → 105,697 (+13,975) on a 14,200 print, ≈98% new contracts. The buy-to-cover alternative is dead. This is genuinely new deep-in-the-money exposure, staged onto the July build.
• ⭐ WYNN — the roll is proven on both sides. We predicted September open interest of ≈100–150; it printed 235 (16,627 → 235, −16,392). December confirmed at 630 → 18,043. The provisional STC marker is retired.
• INTC — a trim, exactly as the open-interest history suggested. OI fell 83,141 → 78,127 (−5,014) against a 5,000 print, a ≈100.3% match. Not a new short.
• GLD — a genuine brand-new bear call spread. Both legs rose ($400 call +29,826, $425 call +26,961), ruling out the "bullish trader cashing out a winner" alternative.
• SPCX — both legs opened (call +13,044, put +10,600); the provisional put resolved as a new long, confirming a clean synthetic short. HYG — the full 75,000 lots became new open interest (1,842 → 77,359), no transfer leakage. HUBS — confirmed at 171 → 3,248, slightly above prediction.
• ⚠️ CC opened, but only ≈81% of it — the $14 strike landed at 13,540 against a predicted ≈16,700 and the $17 strike at 13,687 against ≈16,900. Roughly one contract in five was matched against a closing counterparty. Direction and structure unchanged; the net new position is ≈13,500 spreads, not 16,666.

What changed

The dollar totals are untouched — open interest never moves the cash. What moved is what those dollars mean. TSM's $19.75M and WMT's $2.0M were obligations being retired. SPY's $32.2M established nothing at all. GDX's and XLP's packages were protection being moved, not protection being bought.

The sharpest lesson on this board: a flat open-interest print is a real answer, not a failed test. SPY's collar was neither opened nor closed — we offered readers only those two branches, and reality took a third. A 20,500-lot cross can be economically enormous and still leave the market's aggregate position exactly where it started.

---

⚠️ Risk & Reality Check

A large trade means somebody with capital took a position. It does not mean they are right, and it is never a recommendation.

What today should make plain: this was a board of hedges and unwinds, not conviction. Not one trade took liquidity from the open market. The institutions behind these prints are buying insurance they expect to lose money on, financing stock positions through options, and trimming winners — none of which translates into a trade you can copy.

Three specific cautions:

• Cheap options are cheap for a reason. HYG's puts cost fifteen cents because an 18% credit collapse is unlikely. That is the deal, not a bargain.
• A credit is not a profit. GLD collected $13.6M and accepted $46.0M of tail risk to do it.
• Estimated dates are not confirmed dates. WMT's earnings, TSM's Q3, INTC's Q3 and the CPI schedule are all estimates in this issue, clearly marked. Trading an expiry off an unconfirmed date is an avoidable mistake.

Position sizes you can hold through a bad week, and the patience to skip what you do not understand, will do more for your results than any flow alert. Nothing here is investment advice.

---

Last updated: 2026-08-06 (pre-market) — the next-day OPRA open-interest snapshot resolved all 26 legs from the August 5 session. The provisional callout was replaced with a ✅ RESOLVED section, the At-a-Glance status column now carries the resolved verdicts, and the Quick Read was corrected on INTC, NVDA, SPY, WMT, WYNN and HYG. Five names inverted: TSM, XLP, WMT, SPY, GDX.