There are twelve, all born in the week of 20 July, and what matters more than any single name is where they are not: none of the twelve belongs to the sector that led yesterday's advance. It is a defensive list — energy, defense, data, telecoms and utilities.

Microsoft stock gained 15.51% with its cloud division up 43% and annual revenue past 100 billion: roughly 450 billion in market value added in one session, an all-time record. On the other side MetaShort lost 7.95% on third-quarter guidance below consensus and worries about spending that produces no return; QualcommShort shed 7% with net income down a quarter. After the bell AmazonShort added around 8% with cloud growth of 37%, its fastest in eighteen quarters, while Apple stock fell about 4% despite iPhone sales up 22%. This morning's pre-market widens the gap: Amazon near plus 9%, Apple near minus 6%. It is the same dividing line as Wednesday night — cloud that delivers against hardware that no longer suffices — and today opens from there.

The uncomfortable part belongs here, because it is the part that actually helps a reader. Our model was short MicrosoftLong from 8 June, and yesterday's close ran through both stops we had published: anyone following that level was out, anyone who wasn't watched the position go to minus 15.45%. The weekly signal is flipping to a buy on the candle still forming, and tonight's close settles it. The reverse holds for MetaShort: there the buy died under the results, and the fresh signal this week is a sell. Two names stay solid — FortinetLong, up 78.8% since the signal with results above expectations, and StarbucksLong, which gained 8.6% on a fourth straight quarter of growth.

Twelve stocks moved together between 18% and 31%, and they come from very different families: FuelCell EnergyShort up 30.6%, IRENShort up 30.5% on 2.8 billion of new cloud contracts, Cipher MiningShort up 28.1%, NebiusShort up 27.1%, Bloom EnergyShort up 26.5%, CoreWeaveShort up 21.5%, CleanSparkShort and Applied DigitalShort above 20%, Astera LabsShort at 20%, MicronShort up 18.4%. Crypto miners turned data-centre operators, fuel-cell makers, memory and connectivity: what binds them is the theme, not the business.

When a group this mixed moves in unison it is absorbing a flow rather than answering a piece of news, and flows that arrive in a day leave just as fast. Our model is short all twelve and took the hit yesterday while staying in profit across most of the line: CoreWeaveShort up 26.5% since the signal, NebiusShort up 14.2%, SanDiskShort up 5.5%, NvidiaShort up 5%, with MicronShort down 3%. Applying our protocol, a stock rising 20% against a weekly sell signal does not become a buy: it becomes a case to watch until the week closes. TSMCShort and IntelShort — the latter up 12.5% on pure sympathy, with no news of its own — tell the same story.

Europe's session was an avalanche of results pointing in opposite directions. Rolls-RoyceLong gained 5.5% with half-year operating profit up 46% and guidance raised, the best performer in London; Schneider ElectricLong rose 10.83% on record half-year revenue of 21.2 billion and full-year organic growth lifted to 10-13% on data-centre demand; CapgeminiLong added 5% and also raised its outlook; Lloyds Banking GroupLong gained 4% with pre-tax profit up 23%, the dividend up 30% and a one-billion-pound buyback; BAE SystemsLong rose 2% with an order backlog of 84 billion.

On the other side, adidasShort lost 11.5% in the worst session in its history: record quarterly revenue, but close to a billion in World Cup marketing costs and the departure of the finance chief did the rest. SanofiShort fell 8.9% despite beating on results and raising guidance, because it discontinued three research programmes and wrote down 952 million. Our model is long Rolls-Royce, Schneider Electric, Capgemini, Lloyds, BAE Systems and ShellLong, and short Sanofi since May and adidas since the current week: on the two that collapsed yesterday the signal was already on the right side, and that deserves saying with the same candour we use when it isn't.

The Milan market followed the bounce through the sector it does best. Seven of the eight Italian names moving with their own sector are banks and insurers: AzimutShort up 3.9%, UniCreditLong up 2.4%, Banco BPMLong up 2.4%, BPER BancaLong up 2.3%, Banca MediolanumLong up 2%, UnipolLong up 1.7%, FinecoBankLong up 1.3%. None of them led the sector, all of them followed it, and on a day of vertical moves following is a virtue. On results, Ferrari stock rose 2% with guidance lifted to about 7.6 billion in revenue, CampariLong gained 4.5% and was the best performer at the open, EnelLong posted half-year operating profit of 11.8 billion to a muted reaction.

The other side has two names. StellantisShort lost 6.2% on the Milan line: quarterly revenue up 13%, but net income of just 0.3 billion and an operating margin of 1.8%, below consensus, with 1.0-1.2 billion of tariffs to absorb. PrysmianShort fell 1.9% even after raising its estimates, the classic case of a market selling good news. Our model is short both, since early June and early July, and on Stellantis stock the monthly signal confirms the long-term read. Among the banks we are long every name except Azimut, where the weekly signal turned to a sell on 20 July.

ValeShort posted its best second quarter since 2018, on solid output and favourable prices, and was the most traded stock of the session even while gaining only 1.4%. At the other extreme UsiminasShort fell 9.25% and was the worst on the index: quarterly profit up 236% and operating margin up 86%, yet weak indications for the second half on costs and margins. It is the same pattern seen in Europe with Prysmian, and in Brazil it spread through the steel sector, with CSN alone down 4.8% in sympathy.

The banks held the market up: Itaú UnibancoLong rose 2.2% and led the advance in the global bounce, BradescoLong was flat while announcing a capital raise of up to 10 billion reais, AmbevLong reported quarterly profit up 24.5%. PetrobrasLong gained 2% as oil found a floor. Our model is long Itaú, Bradesco, Ambev and Petrobras, and short Vale since early May and Usiminas since mid-June: on Brazilian steel the position has been on the right side for weeks.

Looking at the session through the baskets removes any ambiguity. US TechnologyShort gained 5.5% and was by far the best, followed by its two near-twins Vanguard TechnologyShort and Fidelity Technology, both at 5%; at the other end US CommunicationsLong lost 2.68% and US Consumer StaplesLong 2.16%. In Europe the picture is tidier: Europe Basic ResourcesShort up 2.64%, European BanksLong up 2.54%, industrials up 1.87%, all three also higher on the week and on the month.

The difference between the two continents sits right there. The American jump is a bounce inside a negative month — technology remains 7.8% lower over thirty days — while the European advance continues a trend running for weeks. Our model is short US technology, with the signal having turned to a sell in the current week, and long European banks since early April. The two positions do not contradict each other: they describe two markets doing different things right now.

The four European indices we follow all carry a weekly buy signal fired at the start of April, and all four sit between 2.9% and 3.6% above their own weekly reversal point: none of them is close to losing the signal. The returns, though, diverge widely — FTSE MIBLong up 12.2% since the signal, CAC 40Long up 4.8%, DAXLong up 3.3%, FTSE 100Long up 2.1%.

The number that matters is a different one, and it concerns the distance from the record. DAX, CAC 40 and FTSE MIB all sit within 2% of their all-time high; the FTSE 100 is the only one 7.6% below. London carries the same signal as the other three, the same distance from its own reversal level, and a third of their return: it is the basket that joins the European advance without ever reaching the top. Anyone looking for continental exposure has, in that number, the reason to prefer Milan, Paris or Frankfurt.

We close with the comparison that explains the rotation better than any commentary. SPYLong, the S&P 500 ETF, sits 2.46% below its record and still carries the weekly buy signal fired on 6 April, with a 9.16% return; QQQShort, the Nasdaq 100 one, sits 8.7% below its own and has been on a sell signal since 20 July. Yesterday the first gained 1.68% and the second 3.3% — double, because technology weighs 50.5% in the second basket against 32.9% in the first, and financials are worth 0.24% against 12.6%.

Technically the two are in opposite states, for reasons that are easy to follow. The first reclaimed yesterday the weekly reversal point it had lost on Wednesday, but today it completes the seventeenth week of the trade and with it the last profit-taking window: from Monday there is nothing ahead of it, only a gain to defend. The second handed back in one session everything the sell had accumulated over two weeks, and finds itself 0.1% from entry with the daily signal a quarter of a point from flipping.