📊 Friday's close, and what's moved since.

🔀 The rotation Friday's index hid. Semis down hard, retail and software up.

📉 Marvell. Beat the quarter, raised the outlook, fell 10%.

🛢️ Oil and German inflation. Both moving this morning.

✅ One thing worth doing before the open.

Read the top of that first table and Friday looks like a nothing day. S&P down a quarter of a percent, Dow flat.

Now look underneath it.

🔻 Electronic Technology, the biggest sector on the market at $24.8 trillion, closed down 1.81%. That's semiconductors and hardware.

🔺 Retail Trade closed up 2.12%. Technology Services, which is software and platforms, up 0.92%.

📐 That's a spread of nearly four percentage points between two halves of the same market, in a single session.

The names make it plainer. Marvell down 10.28%. Nvidia down 4.57% on half again its usual volume. Intel down 2.85%, AMD down 2.33%. Against that, Amazon up 3.97%, Microsoft up 1.68%, Apple up 1.63%, Salesforce up 1.57%.

Money didn't leave on Friday. It moved. Out of the companies that build the hardware and into the companies that rent it out.

And the VIX rose 5.41% into the weekend on a flat index, which is the tell. Nothing looked like it happened, and the price of insurance went up anyway.

🎯 The Desk Take: An index is an average, and an average is designed to hide exactly this. If you glanced at the S&P on Friday afternoon you'd have closed the laptop thinking it was a quiet end to the week. If you were long semis it was one of the worst days of the month. The index is not the market, it's a summary of the market, and summaries are where the trade goes missing.

Worth sitting with, because this comes round every quarter and catches people every time.

Marvell reported a beat on the quarter, driven by data centre strength, and raised its outlook. Headlines sitting on the same page as the price: "MRVL Q2 Earnings Beat on Data Center Strength, Outlook Raised."

The stock fell 10.28%.

Not because the news was bad. Because the guidance wasn't as far ahead of what was already priced as people wanted. By the time a company reports, thousands of people have positioned on what they think the numbers will be, and every bit of that guesswork is inside the price before a word is published.

So the market isn't asking "was that good?" It's asking "was that better than what we already paid for?"

🎯 The Desk Take: Good news and a rising share price are related. They're not the same event. The distance between them is where most earnings-day losses live, and it's why a results date is a sizing decision rather than a conviction one.

Disclosure, since it's on the public record page anyway: Nvidia is one of the desk's open positions. Levels are members-only, the fact that it's open isn't.

🛢️ WTI is up 3.40% at $86.29 this morning, which makes crude the biggest mover on the board right now. Energy equities closed Friday almost unchanged, up 0.23% as a sector, so the crude move hasn't reached the equity side yet. Worth watching which way that resolves when the US opens.

🇩🇪 German inflation ticked up in every single state this morning. The preliminary state readings all came in above their prior year-on-year figures: Brandenburg 3.1% against 2.7%, Hesse 3.0% against 2.7%, North Rhine Westphalia 2.9% against 2.7%, Bavaria 2.9% against 2.8%, Saxony 2.9% against 2.8%, Baden-Württemberg 2.6% against 2.5%. The national preliminary is forecast at 3.0% against 2.8% prior.

🎯 The Desk Take: Six regional prints all leaning the same way is harder to dismiss than one national number. Not because it's dramatic, but because it removes the "one bad reading" excuse.

The board as it sits before the US open. This describes what's there. It isn't a forecast.

Gate 1, trend. Intact at the index level, and that's the problem with reading it at the index level. Underneath, semis and software finished Friday pointing in opposite directions and you cannot be trend-aligned to both.

Gate 2, momentum. Splitting rather than fading. Retail and software have it, hardware lost it on Friday. Gold is the same thing at a different speed, down 4.71% over five sessions and still up 8.25% on the month.

Gate 3, risk. Tighter. A VIX that rose 5.41% into a weekend means the same position size carries more risk this morning than it did Friday lunchtime, and you didn't do anything to cause that.

Verdict: smaller, and pick a side. Gate 3 says size down. Gate 1 says the index has stopped telling you which way you're pointing, so the sector matters more than usual this week.

Take your open positions and group them by sector rather than by name.

Most people think they hold eight different things. Group them properly and it's often two bets held four ways each, which is the same trade in a costume. After a Friday where two halves of the market finished four points apart, that distinction is the difference between a diversified book and a concentrated one you didn't know you had.

If it takes more than five minutes, that's the finding.

Friday carries the month's biggest scheduled print and the calendar is heavy from Wednesday. If you're holding into it, the sizing decision is this week's work rather than Thursday night's.

Educational analysis, not financial advice. US equity levels are Friday 28 August's close; commodity, currency, bond and crypto levels are Monday 31 August morning. All of them will have moved. Named companies are market commentary, not recommendations. Past performance is not indicative of future results.