ES / S&P 500: A Record at 10am, Gone by the Bell, August 6, 2026

The E-mini printed an all-time high of 7,820.25 on Wednesday morning and finished the day 70 points beneath it, settling 7,749.50. That same session, the Dow finished at a record of its own. The Nasdaq lost 0.83 per cent. Three indices, three completely different afternoons.

That spread is the story, and it decides how much weight Wednesday's rejection actually deserves.

The number that decided it

Yesterday the entire constructive case rested on one line: the services prices-paid figure, forecast at 65 after 67.7.

It printed 70.3, five and a half points hot.

Alongside it, private payrolls printed 44,000 against a 65,000 forecast and 98,000 had come before. The services employment sub-index fell to 47.4, into outright contraction, against 51.2 expected. The headline services index, though, stayed in expansion at 54.1, and both composite and services purchasing surveys beat.

Growth holding, hiring failing, input costs climbing again. Markets have no comfortable way to price that combination.

Then at 16:05 a sitting governor put inflation threats above job-market concerns, and that room to postpone disinflation is running out. The same remarks were explicitly two-sided, allowing that a rate increase might still prove unnecessary, that policy has not cost significant jobs, that the labour market is holding up. Quote only the hawkish half and you overstate it. What genuinely changed is that further tightening stopped being unthinkable, which follows a July vote already carrying three dissents for a 25 basis-point increase.

Rotation, not retreat

Here is why the bearish read has a ceiling. The Dow gained 0.49 per cent to a record. The broad benchmark lost 0.17. The Nasdaq lost 0.83. Spread those apart and you get 1.32 percentage points in a single session, which is money moving rather than money leaving.

Capital walked out of long-duration technology and got redeployed elsewhere. A chip designer fell roughly 7 per cent after third-quarter sales guidance came in under what the AI capital-expenditure crowd had assumed. Semiconductors followed it down. An aerospace and satellite name lost more than 13 per cent on projected AI spending.

Meanwhile a travel-booking platform gained over 6 per cent on gross bookings, and an e-commerce marketplace beat on every line, 1.60 against 1.51 on earnings, revenue of 3.13 billion where 3.02 billion was expected, with guidance above consensus.

In a real risk-off session, the Dow doesn't set a record. So the E-mini, carrying far less technology than the Nasdaq and far more than the Dow, sat between them: damaged, then cushioned by where the money went. Extending the downside from here requires the selling to spread past semiconductors. If technology steadies on Thursday while the Dow holds, this index reverts higher almost mechanically.

A poor close is not a broken structure

Wednesday settled in the bottom 5 per cent of its range, three and three-quarter points off the low. Supply did not just appear at the highs, it stayed all afternoon.

One qualification changes the reading materially. That 7,745.75 low came in a full 116.75 points clear of where Tuesday bottomed. So Wednesday delivered a higher high, a low above the last one, and an ugly finish, not an outside day and not an engulfing reversal. The swing sequence running from the 29 July low at 7,324.00 is undamaged.

For that to change you would need 7,745.75 broken decisively, and then acceptance under 7,706.70, the point at which dealer hedging goes mildly pro-cyclical. Until both of those happen, Wednesday reads as one distribution bar inside an uptrend.

The run into it was remarkable, incidentally: 496.25 points, 6.78 per cent, in five sessions off the July low, and not one down day arrived until Wednesday.

Stretched, and no longer accelerating

Price sits 195.16 points above the 20-day, about 2.58 per cent, call it two average daily ranges apart. Stretches like that resolve by pausing sideways, or by reverting. They don't often add another leg without resting.

Every lookback shows stochastics deeply overbought, and on the 9-day the %K has crossed under %D at 89.69 versus 91.27, the run's first negative cross. Nine-day relative strength at 70.04 sits right on the conventional threshold and fell 1.32 points on the session.

None of that's a sell signal by itself. Overbought stays overbought inside a strong trend. It does say the momentum tailwind has stopped strengthening.

And the trend measure argues for restraint on the short side. Nine-day directional strength reads 34.02, positive direction 31.19 against negative 12.82. Selling into a reading above 30 makes the trade counter-trend by definition. Size it that way, and manage it that way.

The volatility configuration

This is the most informative thing in tonight's data. Implied volatility rank sits at 20.05 per cent, near the bottom of its annual range. Skew rank sits at 100.00, the very top.

September options carry 13.94 per cent implied at 43 days. One-month implied on cash reads 13.45 where realised over the same window is 14.47, so protection costs less than the index has been delivering. Call skew in the major proxies sits at the 98th percentile, a product of the recent price-up-and-volatility-up stretch.

Traders bidding up convexity on the upside at a record high, while holding outright volatility on the cheap, is a setup that has historically pointed toward reversion rather than acceleration.

One further detail. The volatility index fell about 4 per cent to near 15.8 on a session where the underlying reversed off its record. Falling volatility into that kind of session is not a warning, it confirms that Wednesday got treated as rotation and not as the opening of a drawdown.

What sits above and below

The band that matters overhead runs 7,798 to 7,826.70. Inside it sit an options magnet at 7,801.70 carrying 99.41 probability, a first computed resistance of 7,797.92, Wednesday's record and rejection at 7,820.25, the single highest-probability magnet on the entire board at 7,824.70, and the call-side dealer wall at 7,826.70.

That wall is the number to watch. Above it, dealer hedging stops suppressing rallies and starts chasing them.

Beneath, that first shelf is a tight one. An options magnet at 7,755.70, the settle at 7,749.50, the modelled move's lower bound sitting on 7,747, and 7,745.75 for Wednesday's low. Four references inside ten points. Wednesday's low already defended that modelled boundary once, holding it to within two points, so the modelling has stayed properly calibrated right through this move. A second successful defence would favour continuation strongly.

Deeper, 7,721 to 7,726.70 packs four independent methods into six points and is the natural first objective for any short. Then 7,706.70, the structural line, where a corrective pullback would get reclassified as something more serious.

How to trade it

Fade strength, don't short weakness. Scale into 7,798 to 7,826 rather than committing everything the first time 7,798 trades. The best fill comes from a push up into 7,820 or 7,826 that fails and cannot stay above the wall.

Stop 7,849, sitting clear of the magnet at 7,848.70 and of 7,846.33 where a second computed resistance sits, so invalidation lands beyond the next real supply layer rather than at some arbitrary distance. Average in at 7,812 and you are risking 37 points. Take a third at 7,749.50, a third at 7,723.42 inside the confluence, and and run the last third at 7,697.33, valid only on a morning that trends instead of chopping. Roughly 1.7, 2.4 and 3.1 to one.

The breakout trade is the mirror. Acceptance above 7,826.70 inverts the hedging to pro-cyclical, and the long entry is a retest of 7,827 to 7,842 with a stop at 7,796, targeting 7,846.33, then 7,872.42, then 7,881.48.

Two things to respect. As of 28 July, fast-money accounts carried 297,476 contracts net short and had already cut it by 16,259 that week, so covering was under way before the advance even accelerated. A short that size under a market making records is fuel. And a signed Hormuz agreement would invalidate a short position instantly regardless of what the chart says, just as and a visible collapse in those talks pulls the foundation out from under the whole five-session advance.

Why Thursday is quiet, and why that matters

There isn't a first-order American release. Claims at 08:30 carry more weight than usual after Wednesday's payroll miss and the contraction in services employment, because a third consecutive labour disappointment would harden the cooling narrative directly into Friday.

The under-appreciated line is unit labour costs, forecast 2.1 per cent against 1.8. After Wednesday's prices-paid acceleration, an upside surprise there extends precisely the rate repricing that pushed money out of duration. A soft print removes it.

Beyond that, Thursday is about positioning into Friday's payrolls, forecast 80,000 after 57,000. Sessions like that compress ranges and drag price into heavy options concentration instead of trending. The pinning strike at 7,778.70 is where that gravity points.

A record given back in one afternoon looks decisive on a chart. Wednesday still made a higher high, and its low held above Tuesday's. Trade the band, not the narrative.

Yesterday this index closed at a record and the whole case rested on one number. That number came in hot: 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.