Nasdaq-100: One Clean Beat, Then Two With Fine Print

For six weeks the bear case on this index reduced to one sentence: all that artificial-intelligence spending is not turning into revenue. On Thursday afternoon one company answered it. Consensus had cloud revenue growing near 39.3 percent. It grew 43, excluding currency, the quickest quarterly expansion this company has managed in four years. The stock added better than 15 percent to finish at 451, its biggest day since October 2008, and roughly 500 billion dollars of value appeared.

Everything followed. The semiconductor group added around 9 percent, memory and storage names leading it with gains of 26 and 18, then 15.37, 13.00 and 11.41. The Nasdaq contract settled 3.28 percent higher while the broad index managed 1.66 and blue chips 1.19. When the wider market follows this one rather than leading it, you know the move started in a single place.

Then, after the bell, two more of the largest constituents reported. Neither was clean. Wednesday night the problem was that every barrier sat above spot while the index was oversold. Thursday cleared several of them in one go.

One clean beat, then two with fine print

The hardware name. Estimates called for 1.89 dollars. The print was 2.02, though 0.11 of it came from tariff refunds, which makes the real beat considerably thinner than it reads. Revenue cleared its bar, 109.42 billion against 108.85 expected, and products did too at 78.68 versus 77.25. Services is where it breaks down. Expected 31.36 billion, delivered 30.74. That segment carries the higher margins and the higher multiple, and it came up short. Tidy headline, softer interior.

The commerce and cloud name. Against an estimate of 1.99 dollars it printed 5.75, a gap so wide it almost certainly carries non-operating items. Sales landed at 200.6 billion where 197.01 was expected, and at 27.46 billion operating income beat comfortably. The guide is the problem. For the third quarter the company sees sales landing between 197.0 and 202.0 billion. Consensus sits at 203.93. Every point of that guided range falls short of it. On operating income the guide spans 26.5 billion down to 22.5, which does at least straddle a 25.07 expectation. Strong quarter, soft outlook.

Neither result is bad. Neither is the unambiguous thing the market received on Thursday afternoon either. Friday's open is the first chance anyone has to price that difference, and the difference is the whole question.

The squeeze had fuel and someone lit it

Timing explains the violence better than the earnings do. Wednesday printed 27,201.50. That is both a one-month and a 13-week low simultaneously, and it landed the same afternoon the policy decision did. Roughly a day later the catalyst arrived on top of it.

Look at who was positioned. The July 21 report has fast money carrying 46,344 contracts long against 121,034 short, leaving them net short somewhere near 74,690, and across that week they piled on 9,294 more shorts while cutting 1,233 longs. Asset managers held the other side at roughly 72,625 net long. So a heavily short cohort walked into a 13-week low eight days later, and then good news turned up.

Covering a book like that produces precisely what Thursday looked like. It also runs out, because it ends the instant those positions reach flat. What decides Friday is whether real money picks up where the squeeze stopped.

Dealer mechanics added to it. The market traded and closed on the negative side of the gamma line, where hedging amplifies rather than dampens, and the desk note attributed part of that mega-cap run from around 390 up to 450 to the amplification alone. Stripping event volatility out, first the policy meeting and then the inflation print, released more hedging-related buying on top. There is also unconfirmed talk doing the rounds, and it deserves labelling as talk rather than fact, that the best-performing names in the complex were concentrated in one fund whose book got sold this week. Were that accurate, a slice of the advance was positions changing hands rather than new conviction arriving.

One day did not repair anything structural

This is where the enthusiasm meets arithmetic. Settlement came in under two averages: the 20-day, up at 29,001.16, by 763 points, and a 50-day of 29,656.80 by 1,419. The month is still 6.88 percent lower. The index sits 8.60 percent under the record it set on June 3, and yesterday's drawdown reached about 10.

The direction readings are blunter still. Trend strength on the 9-day window reads 43.66, which is high, and its components say downward without hedging: negative movement 26.71 against positive 10.37. The 14-day and 20-day windows agree. One session, however violent, does not shift measurements like those.

Momentum tells the same story from the other side. Strength readings all sit fractionally under neutral, and price needs roughly 29,021 before momentum so much as reaches it. Over five sessions the net gain is 141.75 points, half a percent, which measures the preceding damage better than anything else here does.

What did improve is real, though. The 9-day average has been reclaimed, price finished at the extreme of its own range, the overnight extended instead of fading, and price has taken back the volatility inflection threshold. Sitting just under both the overnight low and the settlement is a 100-day average of 28,077.24, which makes that pocket the shelf worth defending.

Still short gamma, in both directions

Closing at 685.91, the fund proxy gained 3.67 percent and finished above its high volatility point of 681, which is a modest positive for stabilisation. Gamma on the call side reads negative 552.71 million, on the put side negative 1.37 billion, so even after the reclaim this surface is still net short gamma. Whichever direction Friday picks first ought to run further than looks sensible.

Flow was genuinely bullish and part of it was durable. Roughly 7 billion dollars net went through the broad index, split about 3 billion of same-day put selling against 4 billion of call buying further out. Single names added another 3.5 billion of the latter, with 2.3 billion of it sitting in the seven biggest technology companies. Longer-dated call buying is the more meaningful half, because it keeps dealers short gamma and forces them to hedge into strength over time rather than for a single afternoon.

Two cautions on the options surface. Implied rank reads 80.95 percent, skew rank 78.26, both high, which makes options dear and downside protection richly bid. Better to use defined-risk structures than to pay up for long premium. Separately, only 8.19 percent of total gamma sits at the next expiry, with the largest out in December, leaving near-dated positioning thin and this surface less pinned than it usually is.

The overnight risk is a currency decision

Friday has no domestic releases at all. Everything heavy came out Thursday morning, so direction falls to how the earnings get read, to month-end mechanics, and to whatever positioning is left over.

The one scheduled item that matters arrives before any of that. A central bank decision lands at 23:30 Eastern with its policy rate expected to stay at 1 percent. Two things sharpen it. Possible intervention in that currency was flagged in a Thursday headline, and reporting on July 22 suggested the bank would consider moving faster than its customary half-yearly rhythm. A hawkish surprise or a disorderly currency move pressures carry-funded positions, and no domestic benchmark has historically been more sensitive than this one to that channel. It voids the setup below outright.

Month-end works the other way. This index finished July down 6.88 percent, and funds targeting fixed allocations generally have to buy the underperformer to rebalance. That leaves a mild structural tailwind running into the close, partly offsetting the urge to square up before a weekend closing out a week holding one policy meeting, one inflation release and three of this index's largest members reporting.

The trade

Buy a pullback, not the breakout. Entry runs 28,280 to 28,340, which covers Thursday's settlement, the overnight low and an average-convergence stall, and it sits right on top of the base at 28,060 to 28,080, where a 100-day average, a 61.8 percent retracement and a stochastic threshold all land inside 20 points. Stop 28,180. That is about 130 points of risk from the midpoint, 2,600 dollars a contract. Targets are the overnight high at 28,570, then 28,700 at the top of that four-method convergence, then 28,877. Roughly 2.0, 3.0 and 4.4 to one.

Nothing gets chased above 28,500. Invalidation is acceptance under 28,130, which means consecutive closes below that base and not a single wick through it.

A counter-trend short exists only if 28,630 through 28,700 rejects clearly, ideally while the semiconductor complex declines to confirm. Enter on a 30-minute rejection candle, stop 28,790, aiming 28,380, then 28,240, then 28,080. Smaller size. Selling into a market that has just put on 3.28 percent while dealers sit short gamma is a hazardous way to spend an afternoon.

Weighting it: constructive consolidation, holding over 28,240 and working a band from 28,300 up to 28,700, takes 45 percent. Failure and retracement back to the support base takes 30, and closing below it would bring 27,950 into play, then the pocket between 27,816 and 27,775, badly weakening the recovery case. Clean continuation through 28,700 toward 28,833 and 28,950 takes the remaining 25.

Two caveats belong in the open rather than a footnote. The dealer-positioning levels quoted here come from a publication referenced to cash at 27,192, a price from before the rally happened, so the futures translations resting on a 129-point basis are approximate and they reset whenever the surface refreshes. And the liquidation attribution is talk circulating in the market, labelled as talk, not established fact. It appears here because it bears directly on how durable this move should be judged, which is precisely why the label matters.

The honest read on Friday: one company answered the question that had been driving this index lower for six weeks, and the answer was genuinely good. Two more then answered it with qualifications. Underneath, every trend measure still points down, the month is still deeply negative, and a large part of Thursday looks like short covering meeting a low. Trade the shelf and let the open tell you whether real money showed up.

The complete data picture

Every number behind Friday’s plan, charted first; the full numeric reference follows underneath.