SanDisk to rise 42%, betting only 4 days on optical module decline

Sandisk at 11:44 and 11:55 swept in two blocks of $2350 calls expiring on 11/20, totaling $12.91mn—the largest and most expensive trade of the day. At $184.46 per contract, the breakeven is 2534.46, a 41.84% premium over the 8/17 close of 1786.850. Later that day at 13:21, someone bought $970 puts expiring on 8/21 (only four days left) for $1.32mn. The close was 968.900, pushing those contracts immediately in-the-money. One side bets on a 95-day horizon while the other plays a four-day expiry; both trades led their respective directional flows for the day. Total volume: 136 orders, long-to-short ratio 3.3:1 ($60.76mn longs vs. $18.14mn shorts), with 36 pre-filters applied.

SNDK Sandisk

Direction: 🟢 Bullish (Deep OTM far-month call, cost $12.91mn)
Expiry: 2026-11-20
Strike: $2,350
Notional Size: ~$12.91mn (cost)
Volume: 700 contracts (swept in two blocks at 11:44 and 11:55)
Structure: Single-leg same-strike same-expiry · Deep OTM Buy Call

Data Highlights: 700 contracts, premium $12.91mn, unit price $184.46, breakeven 2534.46, max loss per contract $18,446, notional exposure $164.5mn. Entered at 31.1% OTM with 95 days to expiry (DTE). This capital landed on an already steep rally: Sandisk closed at 1212.210 on 8/7, rising to 1786.850 by 8/17—a cumulative 47.41% gain over six trading days. Notably, on 8/13 it gapped up 13.68% (1344.290 → 1528.110, vol 22.09mn shares). On 8/17, it opened at 1700.745, hit a high of 1827.990, low of 1698.000, and closed at 1786.850 (+8.88%), with volume of 18.04mn shares and turnover of $3.20bn. The catalyst was clear on the surface: NAND flash supply tightness, easing concerns over AI capex, the Philadelphia Semiconductor Index returning to bull market territory, and Micron and Sandisk leading the charge. The company's 10-K summary also dropped that day—FY2026 revenue of $20.25bn, EPS of $73.76. On the flip side, worth noting: after surging over 6x then halving, hedge fund holdings show major divergence; funds like Bridgewater and Tiger Global have exited.

Bull Case Comparison: Consensus target from 24 analysts is 2107.70, 17.96% above the close. 16 Strong Buys vs. 4 Buys, 3 Holds, 0 Sells, 1 Underweight, with a range of 1000 to 3600. The breakeven of 2534.46 is 20.25% higher than consensus—meaning these 700 contracts need the stock to rise 20% above the average analyst view to profit. However, there is still 29.60% room to the ceiling of 3600, placing this trade in the "above consensus, below extreme" zone. The reference frame is wide, which itself signals divergence.

My View: No follow. A 41.84% move to break even in 95 days requires another 8/13-style gap surge from the storage chain, yet we've already seen a 47.41% gain in six days—the momentum itself is becoming resistance; each subsequent leg gets more expensive. If you believe NAND supply tightness is a capacity issue rather than sentiment, 2534.46 is reachable, with the high target at 3600—I get that stance. But capacity cycles are measured in quarters, and this 11/20 contract only has three months. The direction might be right, but time is tight. Sandisk appears for the 4th time on this account. Last appearance was 7/22—that day saw six near-the-money call blocks totaling $10.66mn, the largest long position, with expiry precisely covering the 8/5 earnings report, betting on a specific event. This time, DTE95, 31.1% OTM, no calendar event attached. The shift from "betting on an event" to "betting on a capacity cycle" is the key change. Invalidating conditions are clear: if a second wave of adds occurs at the same strike before 11/20, or if strikes roll up to the $2600 level, indicating this isn't a one-off odds play, I will recalculate. Watch levels: 1827.990 (8/17 intraday high) and 1698.000 (8/17 intraday low).

LITE Lumentum

Direction: 🔴 Bearish (Four-strike put ladder expiring on the same date, plus a short put to reduce cost, net cost $3.22mn)
Expiry: 2026-08-21
Strikes: $970 · $950 · $930 · $920 (Buy) · $910 (Sell)
Notional Size: ~$3.22mn (net cost)
Volume: 1374 contracts (1316 bought + 58 sold)
Structure: Five-leg same-expiry · Ladder Buy Put 叠 Put Spread

Data Highlights: At 09:46, sold 58 $910 puts for $130k (unit price $22.41, seller breakeven 887.59) to lower costs. Then densely bought four strikes bottom-up: 09:51 & 09:56 $930 (226 contracts, $621k, unit $27.48, BE 902.52); 10:00 $920 (249 contracts, $573k, unit $23.01, BE 896.99); 13:21 $970 (441 contracts, $1.32mn, unit $29.93, BE 940.07); 13:32 $950 (400 contracts, $840k, unit $21.00, BE 929.00). Total bought: 1316 contracts, $3.35mn, notional exposure $124.7mn. Spanning nearly four hours, per our account's 口径 not treated as one institution, but all four legs are pinned to the single 8/21 expiry, with a duration of just 4 days. This money was placed during a bullish candle: Lumentum opened 948.410, high 999.675 (just 0.33% from 1000), low 927.000, closed 968.900 (+4.61%), vol 5.39mn shares, turnover $524.7mn. The entire optical comms chain rose intraday; LITE surged >7%, Coherent ~11%, Marvell >6%. Catalyst: Nvidia Spectrum X co-packaged optics switch entering mass production, GFS forecasting 100k units shipped by 2027, combined with China supply tightening driving up AI data center material prices. Same day, company 10-K summary showed FY2026 revenue $3.01bn, EPS loss $92.96. At close, the 441 $970 puts were ITM by $1.10.

Bull Case Comparison: Consensus target from 24 analysts is 1148.30, 18.52% above close. 15 Strong Buys vs. 5 Buys, 4 Holds, 0 Sells, range 820 to 1400. All four leg breakevens fall within this range—the shallowest 940.07 is 18.13% below consensus, the deepest 896.99 is still 9.39% above the lowest target 820. Implication: These puts don't need any analyst to be proven wrong; they just need the stock to return to the lower bound of the covered range.

My View: Follow, but only the 441 $970 contracts. Already ITM at close; a further 2.98% drop hits 940.07, with only 4 days left to 8/21. The intraday high wick at 999.675 shows exhaustion shape. I won't touch the $920 and $930 strikes: 896.99 and 902.52 require drops of 7.42% and 6.85% in four days. On a chain just ignited by CPO mass production news, such magnitude needs sector-wide negative news, which didn't happen on 8/17. If you think deeper strikes offer cheaper unit prices and better odds, I understand the math—but four days aren't enough for an exhaustion cycle to complete; cheap paper goes to zero against time anyway. Lumentum appears for the 3rd time. Last was 8/12: optical comms rallied hard, LITE +13.63%, heavy money all on puts. Five trading days later, the sector rises again, heaviest short remains here, but now shifted from cross-strike/cross-expiry to four legs squeezed into one expiry, indicating higher urgency. Invalidating condition: closing above 999.675—if it reclaims that, time value across all four legs will be drained in three days. Watch levels: 999.675 (8/17 intraday high) and 927.000 (8/17 intraday low, breach implies 940.07 enters range).

NVDA NVIDIA

Direction: 🟢 Bullish (Short put financing stacked with three-strike ladder calls, net cost $7.07mn)
Expiries: 2026-08-24 · 2026-08-28 · 2026-08-31
Strikes: $230 · $240 · $242.50 (Buy) · $225 (Sell)
Notional Size: ~$7.07mn (net cost)
Volume: 29,061 contracts
Structure: Five-leg multi-expiry · Risk Reversal 叠 Ladder Buy Call

Data Highlights: At 09:30, first minute sold 492 $225/08-31 puts for $321k (unit $6.52, seller BE 218.48, needs -2.90% to start losing, assignment obligation $11.07mn). Then calls spread across three strikes: 10:12 $230/08-24 (3689 contracts, $7.75mn, unit $2.10, BE 232.10); 10:29 $242.50/08-28 (3006 contracts, $6.7mn, unit $2.23, BE 244.73); 10:55 & 10:59 $240/08-28 (21874 contracts, $59.5mn, unit $2.72, BE 242.72). Call-side notional exposure $682.7mn. The three expiries weren't random: Nvidia reports Q2 earnings after market close on 8/26, management previously guided ~$91bn revenue. The $230 leg expires 8/24, settling two days before earnings, completely excluding earnings risk. $240 and $242.50 expire 8/28, two days post-earnings. The sold $225 put extends to 8/31. Same day, same stock, buyer splits "pre-earnings jitter", "the earnings itself", and "post-earnings" into three separate pricing segments. Nvidia 8/17 open 225.980, high 227.920, low 224.860, close 225.010 (-0.06%), vol 93.68mn shares, turnover $21.17bn, seven-day range narrow between 217.500 and 225.300. Biggest news that day was partnership-related: Nvidia agreed to provide up to ~$105bn support for a massive data center campus in Ohio; OpenAI will lease up to 8GW compute, first 800MW expected online before 2028, plus another $1.5bn investment in SB Energy; Jensen Huang denied this was "circular financing" and cited a $600bn compute market opportunity.

Bull Case Comparison: Consensus target from 63 analysts is 302.83, 34.58% above close. 48 Strong Buys vs. 10 Buys, 2 Holds, 1 Sell, 2 No Opinion, range 180 to 500, updated 8/17. All three call breakevens are far below consensus—232.10 is 23.35% below, 242.72 is 19.85% below, 244.73 is 19.19% below. This isn't a position challenging seller expectations; it just wants a small segment of the path to consensus realization.

My View: Follow the 3689 $230 contracts, skip the 21874 $240 contracts on 8/28. 232.10 requires only a 3.15% rise, essentially back to slightly above 8/13's 227.230. The 8/24 expiry keeps earnings completely out—betting on pre-earnings sentiment accumulation, not the result. This leg has the cleanest risk exposure. I won't take the $595mn on $240: 242.72 requires a 7.87% rise within two days post-earnings. Management already laid out the $91bn guidance, compressing upside surprise space. The 21874 contracts bet on "surprise beyond guidance," which isn't an odds-favorable position. I hear the counter-argument—the $105bn OpenAI commitment plus $600bn market size narrative could turn earnings night into a storytelling night rather than numbers night. But narrative-driven gaps usually price in far-month contracts first, not clearing in two days. Nvidia is a regular here. On 8/11, it was the longest-duration holding in the "four techs pressing 2027" trade, buying 493 days. Six trading days later, money on the same stock shrinks to 7-14 days, all queuing around one earnings date. Invalidating condition: Post-earnings gap directly above 242.72, indicating I underestimated incremental upside beyond guidance. Watch levels: 227.920 (8/17 intraday high, also seven-day high) and 224.860 (8/17 intraday low).

DELL Dell

Direction: 🟢 Bullish (Near-month call buy + far-month put sell, both legs aligned, net income $2.33mn)
Expiries: 2026-09-04 (Call) / 2026-10-16 (Put)
Strikes: $500 · $470
Notional Size: ~$2.33mn (income)
Volume: 764 contracts (200 + 564)
Structure: Two-leg multi-expiry · Risk Reversal (3h 35m interval, not treated as one institution)

Data Highlights: 10:53 Sold 564 $470/10-16 puts for $2.93mn (unit $51.95, seller BE 418.05, needs -12.87% to start losing, assignment obligation $26.51mn). 14:28 Bought 200 $500/09-04 calls paying $596k (unit $29.80, BE 529.80, needs +10.42%). Both legs aligned directionally, positioned one above, one below. Money received from short put is nearly five times the cost of the long call. Expiries also align with calendar: Dell reports Q2 FY2027 after close on 9/3. $500 call expires 9/4, exactly the day after earnings. $470 put extends to 10/16, spanning the entire earnings window. Dell 8/17 opened high, closed low: open 496.650, high 496.860, low 469.050, close 479.810 (-2.24%). Close was 3.43% below intraday high, intraday amplitude 5.60%, vol 5.27mn shares. Previous week saw a steep slope: 8/11 close 440.970, 8/12 gapped up 9.87% to 484.500, 8/13 close 494.510 (seven-day high), then two days of declines. News flow was split—Melius Research raised PT to 650 pre-market citing AI cloud capex growth; but GPU/CPU 板块 diverged intraday, Dell fell ~3% against Marvell's rally; two days prior, Mubadala Investment Company liquidated its Dell stake on 8/14.

Bull Case Comparison: Consensus target from 28 analysts is 506.61, 5.59% above close. 14 Strong Buys vs. 5 Buys, 8 Holds, 0 Sells, 1 No Opinion, range 360 to 700—8 Holds represents the highest divergence list in this piece. Call BE 529.80 is 4.58% above consensus, meaning it must beat average seller view to win. Put BE 418.05 is 16.13% above the lowest target 360, placing the assignment point above the lower bound of the coverage range. One leg of this stock must outperform consensus; the other just needs consensus not to collapse completely.

My View: Follow the 564 short put contracts, skip the $500 call. 418.05 requires a 12.87% drop to start losing, 60-day duration spans 9/3 earnings, assignment point is 16.13% above the most pessimistic view among 28 analysts—this is reasonable pricing for exchanging time for cash flow; the compensation ratio of $2.93mn income to $26.51mn exposure makes sense. I won't take the 200 $500/09-04 contracts: 529.80 requires a 10.42% rise the day after earnings, yet this stock just played a high-open-low-close drama on 8/17, crashing from 496.860 to 469.050 close at the bottom. Short-term momentum is downward. Dell appears for the 4th time. Last was 8/06—that day it was the largest long position at $9.39mn, near-end also $500 strike expiring 8/21, I judged "near-end no follow"; today close 479.810, only three trading days left to 8/21, that leg's BE 513.75 wasn't reached, judgment holds. This time, buyer pushed the same $500 expiry from 8/21 to 9/4, moving it exactly to the day after earnings—same price, second reschedule, itself admitting the first buy was early. Correction: The 8/06 piece noted observation point as "8/27 earnings", verified actual date this round is 9/3 after close. Watch levels: 496.860 (8/17 intraday high, reclaim invalidates high-open-low-close thesis) and 469.050 (8/17 intraday low).

YPF YPF Argentina

Direction: 🟢 Bullish (Single-leg OTM medium-term call, cost $2.20mn)
Expiry: 2026-10-16
Strike: $55
Notional Size: ~$2.20mn (cost)
Volume: 10,000 contracts (round lot single block)
Structure: Single-leg · OTM Buy Call

Data Highlights: 10:35 Single block bought 10,000 $55/10-16 calls, premium $2.20mn, unit $2.20, BE 57.20, max loss $220/contract, notional exposure $55mn, entered 6.0% OTM, DTE60. 10,000 round lots indicate a single institution's one-time bet, no splits, no second tier. Price action cooperated that day: YPF open 51.120, high 52.850, low 50.880, close 52.700 (+5.29%), highest close in seven days, vol 1.93mn shares (2.10x previous day's 916.1k), turnover $100.5mn. 8/17 catalyst was rating upgrade—HSBC upgraded YPF to Buy after strong Q2 results. Preceding string of positive news: 8/10 company reported Q2 net profit $1.21bn, accepted Edenor's $780mn offer to acquire MetroGAS and MetroENERGÍA shares; 8/11 CEO raised 2026 EBITDA guidance from $6bn to $8bn; 8/12 stated starting Jan, daily crude export of 180k barrels via VMOS project.

Bull Case Comparison: Consensus target from 13 analysts is 59.48, 12.87% above close. 6 Strong Buys vs. 4 Buys, 3 Holds, 0 Sells, range 48 to 78. BE 57.20 is 3.83% below consensus—the only one among the five stocks where BE falls below consensus; others either exceed consensus (Sandisk +20.25%, Dell Call +4.58%) or bet on consensus being disproven (Lumentum Put). In other words, these 10,000 contracts don't need seller expectations to be exceeded, just half-plus realization.

My View: Follow. 57.20 sits 3.83% below consensus 59.48, 8.54% break-even distance paired with 60-day duration, the smoothest odds in the entire piece—direction, fundamentals, and seller views all aligned, no need for anyone to be slapped in the face. Fundamentals side isn't a single-point news: net profit, EBITDA guidance, export capacity, asset divestment linked into a string over eight days, HSBC upgrade provided a rating footnote to this sequence. Risk lies in structure, not direction: 10,000 round lots is a one-time bet, no splits, no add-on traces, lacking confirmation, so position weight must be lighter than recurring fingerprints like Sandisk/Dell. YPF Argentina debuts on this account, no historical positions to compare, following new ticker rules for step one. Invalidating condition: Drop back to 48.900 (8/12 seven-day low)—that means the positive news string since 8/10 is fully digested, 60-day time value must be recalculated. Watch levels: 52.850 (8/17 intraday high) and 50.880 (8/17 intraday low).

The first milestone is 8/21, settlement of Lumentum's 1316 puts. Whether they can 压 968.900 below 940.07 in four days is the first falsification test of this optical module surge. Three days later, 8/24, Nvidia's 3689 $230 contracts expire, deliberately avoiding the earnings report two days later on 8/26—that quarterly report with $91bn guidance is the real payoff date for the 21874 $240 calls on 8/28, separated by only two trading days. After that, 9/3 after close, Dell delivers Q2 FY2027; next day 9/4, the 200 $500 calls settle, giving exactly one day. Further out, 10/16 has two hanging positions: YPF's 10,000 calls guarding 57.20, Dell's 564 short puts guarding 418.05, one betting on realization, one on non-collapse. The furthest out is 11/20, Sandisk's 700 contracts targeting 2534.46—still 41.84% away from 8/17 close, with no calendar events to leverage along the way.