NQ / Nasdaq 100: Both Beat, Both Sold, August 5, 2026

Two artificial-intelligence bellwethers reported after Tuesday's close. Both beat. Both got sold. That happened at the end of the best four-session run semiconductors have managed in six years, and it is the reason Wednesday is not a simple continuation trade.

The Nasdaq had just put on 3.22 per cent, its most forceful day in months, settling the September contract at 29,863.50. Then the session ended, the numbers came out, and nobody wanted them.

What lifted it

Shipping headlines. All morning the reports built: Tehran and Washington closing on terms that would restore commercial traffic through Hormuz, and at 07:43 Eastern the Treasury Secretary told an interviewer a deal might land "tomorrow." Crude sold off hard, Treasuries firmed, and the inflation premium buried in equity prices came out in a rush.

Chips did the carrying, up 7 per cent as a sector. Earnings supplied the rest: an analytics name added better than 29 per cent, beating on revenue and lifting its outlook, a scanning-hardware business gained over 28, a heavy-equipment maker better than 5. Breadth was real, not a handful of tickers, with the broad index up 1.79 per cent and the industrial average up 1.71, both at records.

Then the close, and the problem

The chip designer earned 1.66 dollars where 1.62 was expected. Revenue of 11.54 billion against 11.31 billion. Operating income ahead. Third-quarter sales guidance bracketing a midpoint above consensus. Shares fell about 7 per cent.

What the market disliked was not the top line but what it cost to get there. Capital spending came in at 808 million dollars where 298.6 million was penciled in. Research ran 2.53 billion against 2.45. Operating margin improved by a tenth of a point. Desks read the guidance as light on sales and noted that investors now want demonstrated returns from data-centre spending, not just more of it.

The space and connectivity business told a similar story with better numbers. Loss of 0.09 dollars where 0.24 was forecast. Revenue 7.8 billion against 6.81. Adjusted earnings before interest and tax of 3.5 billion where 2.0 was expected. Its AI segment lost 1.26 billion against a feared 2.39. Shares fell roughly 5 per cent, comfortably inside the 14 per cent options had priced.

Two companies cleared every printed hurdle and both traded down. Read that as positioning rather than as fundamentals. Four days of buying had already discounted good news, and good news arrived on schedule with nothing left to pay for it.

The index that led is the index still behind

Here is a fact worth sitting with. The Nasdaq gained more than either the broad index or the industrials on Tuesday, and it is the only one of the three that did not set a record. It closed 3.98 per cent under 31,100.00, the 52-week high it set for itself back on 3 June. That is 1,236 points of unrecovered ground.

Which produces two valid readings of the same chart. Measured against the past month, the contract sits high in its quartile and looks stretched. Measured against the past quarter, it sits mid-range and looks unfinished. Neither one's wrong. Wednesday is where the argument gets settled.

Above every average, and the stack still upside down

Price trades above all five major averages, which is the strongest single point in favour of continuation. The one that counts is a 50-day of 29,643.80, sitting just 220 points beneath spot. It was above price for most of last week and got reclaimed during Tuesday's run. Losing it again would say plainly that Tuesday was one day rather than a turn.

The stack itself hasn't finished repairing. At 28,547.85 the 5-day still sits under a 20-day of 28,971.31, left over from the July decline, and that takes several more sessions to unwind even if price simply holds.

Momentum tells a split story too. Nine-day directional strength reads 37.66 with positive movement clearly on top, a genuine short-term uptrend. Stretch to fourteen days and the two lines almost touch, 22.02 against 21.39. Go to twenty and fifty days and negative movement is still the bigger number. Four sessions have repaired the damage without reversing the intermediate picture, and the multi-system composite agrees, reading 56 per cent buy with the trend signal itself on hold.

The volatility complex would not confirm

On a day the broad index rose 1.79 per cent to a record, the volatility gauge rose 4 per cent to 16.49 and its second-order measure rose 2 per cent to 92.57. Fixed-strike readings picked up between two and ten points across the surface. Second session running of that pattern.

Durable advances usually drain option prices, because participants get comfortable selling upside and trimming protection. Rising volatility beside rising price says the opposite: this is being bought through options rather than accumulated through shares, which forces intermediaries to chase in both directions. Fast up, fast down.

Skew agrees. On the proxy it ranks in the 91st percentile, so downside insurance is dear while overall volatility is not. People are participating in the rally and paying up to be protected from it at the same time.

One more measurement worth carrying into Wednesday: this contract moves more than its options charge for. Historic volatility over nine days on the futures reads 31.05 per cent; September options imply 22.05. One-month realised on the proxy is 24.07 against 21.83 implied. So treat that implied band as a middle expectation rather than an edge.

No structure left overhead

On the proxy, dealer gamma is negative both sides of the book, put gamma reading negative 1.22 billion and call gamma negative 468.46 million. Short-gamma intermediaries must buy strength and sell weakness to stay hedged, which is the mechanical reason Tuesday accelerated the way it did.

More importantly, price has run clean through every piece of gamma structure above it. The heaviest call-side concentration equates to roughly 28,550 on the index. The primary concentration strike is near 29,000. The volatility inflection sits at 28,530 and the flip level at 28,274. Every one of them is now underneath the market.

Two things follow. Nothing pins price between here and the 30,269 area, with the one-month high at 30,094 on the way, which is exactly why the advance ran on without meeting resistance. And the flip level is roughly 1,400 points below, so the mechanical bid dealers supply while price rises would take a long time to become mechanical selling while it falls. All of these were calculated off a stale reference price. Treat them as shape, never as precise numbers.

Who was positioned for this

The positioning snapshot dates to 28 July, the day before the low, which limits what it can tell you about now but says a great deal about what the rally started with.

In one week dealers dumped 16,787 longs and put on 5,094 shorts, the largest swing in the dataset, arriving at the low net short 27,581. Fast-money accounts added 14,889 longs while still carrying a net short of 58,298, which is covering into weakness rather than conviction. Asset managers stayed structurally long at 76,759 net and trimmed shorts, treating July as an accumulation window.

So the market entered this move with intermediaries short and fast money short. That is mechanically supportive and explains the speed of the retracement. Whether any of that fuel is left is precisely what the next update will show.

What Wednesday turns on

Ten in the morning. The services survey, and inside it the prices-paid component, carrying a 65 forecast after a prior month at 67.7.

Services inflation is the piece policymakers have had the least success moving, and only a week ago three officials broke ranks to vote for a hike, splitting the committee 9 to 3. A print above 67 puts September back on the table, lifts the long end, and pressures a long-duration index directly through valuation. At or below 65 and the disinflation story that cheaper energy started gets a second leg.

The rest is supporting cast. Private payrolls at 08:15 carry a 65,000 forecast after 98,000. The refunding announcement at 08:30 is a rates event that reaches technology through duration. Inventories at 10:30 matter mostly through the Hormuz narrative. No large technology name reports after the bell, so one habitual overnight hazard is absent.

How to trade it

Don't chase the gap. Let the opening range work off that earnings weight first, then buy a pullback that gets defended, 29,560 to 29,630 where the 13-week retracement at 29,610.77, the session pivot at 29,550.58 and the reclaimed 50-day all sit inside 80 points.

Defended is the operative word. Take a rejection candle on the 15-minute frame, or stabilisation in the chip complex, or a break that fails and takes 29,630 back inside two bars. First touch alone isn't evidence. Stop beneath 29,395, under the stochastic shelf at 29,405.50 and under a 40-day crossing of 29,429.41. From 29,600 that is roughly 205 points of risk, paying about 1.3 to 29,860, 1.7 to 29,956, and 2.4 to the one-month high at 30,094.

The first fifteen minutes carry one question above all others: how do semiconductors behave. Open lower on the chip reaction and stabilise inside ten minutes, and the four-session advance holds, with pullbacks worth buying. Let them accelerate lower and pull the index under 29,741, and the move is unwinding, in which case stand the setup down until 29,550 to 29,630 is reached and actually held.

The conditional short only exists on a failed push into the overhead band. Price trades into 29,956 to 30,094 and closes a 15-minute bar back under 29,940, ideally alongside a hot prices-paid figure. Sell 29,900 to 29,950, stop above 30,105, targets 29,741, then 29,610, then 29,550. Note that the first objective falls short of a one-to-one hurdle. This structure only justifies itself held for the second.

Stand aside if the open lands mid-range, somewhere around 29,700 to 29,850, and price is still sitting there at 10:30 having shrugged off the services numbers entirely. When a session ignores its own main event, participants are simply waiting on Friday payrolls, and days like that do not pay directional structures.

The unsigned part

Every bit of this four-session, 2,755-point recovery rests on an agreement nobody has signed.

Two weeks ago the administration was threatening infrastructure targets. Strikes were announced on 29 July, and 27,201.50 was where this contract bottomed that same day. By 31 July there were reports of preparations to hit energy targets. Then it turned: Qatar confirmed drafted language, the Treasury Secretary said "tomorrow" on television, and European governments were reported willing to fund part of a reopening.

Tomorrow is Wednesday. A confirmed deal is already largely in the price. A collapse in the talks is not priced at all. Late Tuesday already brought friction, with a Saudi official denying at 16:58 that mediated talks with Houthi representatives were happening.

That asymmetry is the whole argument for patience at these levels. Buy the dip that gets defended. Don't pay up for the high.

The broad index closed at a record on the same headlines, and it has the same problem: ES / S&P 500: a record on loan.The complete data picture

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