Biggest Red Day of the Run — I Went Shopping for Two Notes (Plus: Intel Earnings Land)

Today the market took its biggest single-day hit of this stretch. QQQ closed at 691.96, down 1.90%; my portfolio's time-weighted return (TWR, deposits excluded) fell 2.26%, trailing by 0.36 points. Unlike the last two no-trade days, I placed three orders today and filled two.

First, why the tape fell — three things stacked up:

1. Alphabet's capex shock. The quarter itself was strong — $119.8B revenue, Cloud +82% — but management raised 2026 capex guidance to $195–205B and said 2027 would "increase significantly." Capex doubled to $45B this quarter and free cash flow went negative for the first time in Alphabet's public history. The stock fell 7%, its worst day in 14 months. The market isn't afraid Google is broken; it's afraid of who pays the AI infrastructure bill.
2. Tesla missed big — expenses growing faster than revenue, down 14%. The two together erased hundreds of billions in market value and dragged big tech down with them.
3. Oil broke $100. Houthi attacks on Saudi tankers in the Red Sea pushed Brent from $94 through $100 in two days, reigniting inflation worries.

Among my names: Netflix closed green at +0.53% against the tape; Intel fell 2.33% during the day, then crushed earnings after the close (section four).

1. The day's P&L, decomposed: the cushion's biggest day yet

• Long side −4.90%: legacy QQQ shares −3.20%, the stock-replacement LEAPs −1.51%, Intel shares −0.22%. A leveraged long book eats the full red day.
• Short side cushion +2.64%: 51 QQQ short calls gave back +2.52% — the most work the cushion has done in a single day since this book opened — plus +0.12% from the Intel, Netflix, and Google calls.

Net: −2.26%, trailing by 0.36pp. Regular readers can recite the reason: net long > 1x means losing more on down days and winning more on up days, with call premium sanding down the volatility.

Third day testing the straight-line formula: short-call leg ≈ +0.125 − 1.297 × index move. Plug in −1.90% → predicted +2.59%, actual +2.52%, residual 0.07pp. Three days, three regimes (big rally / small dip / big drop) — one first-order line caught them all.

2. Today's real work: shopping in the panic

On the red day I bought two "notes" — both deep in-the-money buy-writes (buy stock + sell an ITM call):

• QQQ note: bought 200 shares @690.79 + sold 2× 650 calls (Aug 21) @50.82. Breakeven 639.97 — a 7.4% cushion up front; as long as QQQ doesn't fall below 650 (−5.9%) by expiry, it pays 1.57% in 29 days, ~19.7% annualized.
• GOOG note: bought 100 shares @319.61 + sold 1× 300 call @25.15. Breakeven 294.46, 7.9% cushion; above 300 at expiry (−6.1%) it pays 1.88%, ~23.7% annualized.

Why today? Because panic inflates IV — the same strike distance buys a thicker cushion. Fear is the raw material these notes are made of. The GOOG one goes further: it's catching a knife down 7.3% post-earnings, where the selling was about capex fear compressing the multiple, not a broken business (Cloud +82% says so). I don't know when the knife lands, so I don't catch it bare-handed — I catch it inside a note that doesn't lose until −7.9%.

The honest entry in the ledger: these weren't funded by expiring positions rolling off — they were bought with new margin. Yesterday's post ended with "new notes only get rolled from freed-up capital, no new financing." Today I broke that rule. Margin debt crossed 1x of net worth for the first time, the financing rate is now 4.53% annualized (above the ~3.6% Treasury), and the zero-order distance to a margin call narrowed from 32.5% to 28.3%. I think what I bought was right; how I bought it crossed my own line. Both sentences belong in the same day's review.

3. Scoreboard

Indexed to July 9's close = 100: me −3.24% vs QQQ −4.33%, +1.09pp of excess. The biggest red day of the run hammered both lines; mine fell more, and the lead narrowed from +1.48pp to +1.09pp.

The pattern is now unmistakable: the lead is earned on down days (the cushion) and partly refunded on big down days (the leverage). A car with softer suspension still feels the big potholes.

4. Intel lands — just below the cap

Yesterday's ±15% implied-move cone got its answer tonight. Intel Q2: revenue $16.1B, +25% — fastest growth in nearly 15 years, a 12% beat; EPS $0.38 vs $0.21 expected; Q3 guidance raised across the board. The stock spiked above 110 after hours, then settled around ~104 (+4%).

For my position this is nearly the most comfortable page in the script: 10 calls at the 110 strike expiring tomorrow, and the after-hours price sits just under it — close below 110 tomorrow and those 10 calls expire worthless, full premium kept, +0.37% of net worth, 1,200 shares intact; punch through 110 and 1,000 shares get called at 110 (still a gain against the 108.95 basis) plus automatic deleveraging. Both endings acceptable — that's the point of walking into earnings covered: you trade "guessing the direction" for "two scripts you can live with."

But a review grades the process, not the outcome: during the day I posted a roll of the 110Cs out to August (asking 5.50 credit), it didn't fill, I didn't chase, and I cancelled before the close — which means I carried 10 one-day calls into an earnings night. Settling at 104 was luck; at 115 I'd have been pinned on a gamma cliff with zero time-value protection. A fine outcome doesn't excuse a broken process. Points deducted, in writing.

Tomorrow: the 110C settlement, and whether the two 90Cs go deeper ITM (assignment there = $18k of automatic debt paydown). One more number worth savoring: this Intel program, rolled since May, now has an effective cost of −$8.48 per share — the premium collected exceeds what the shares cost. If the 110Cs and month-end 90Cs all get assigned and all 1,200 shares are called away, the program closes out at literally negative cost, coupons collected the whole way.

---

A formula that held for three straight days, two panic-discount notes, and an earnings print that stopped just under the cap — a full page in this book. And one new line in red on the discipline list: the pace of buying notes must not outrun the pace of capital being freed. The next note waits for the Aug 21 batch to expire.