Weekly action plan.

H P E leads our watchlist. Here is the business case, the chart, and the opening setup.

First, the business. Hewlett Packard Enterprise sells servers, networking and hybrid cloud infrastructure.

On September second, it reported quarterly revenue of twelve point two billion dollars, up thirty-four percent. GAAP operating margin reached eleven point four percent.

That margin matters: revenue growth needs to become profit. Management also raised its outlook. Strong results earn our attention; the entry still needs a structure.

Look at the actual Finviz chart, through September eleventh. Friday’s green candle pushes through the descending purple trend line. The rising cyan support shows the compression underneath.

The close was sixty-two point zero nine. Our planned pivot is sixty-two fifteen. Then come the obstacles: sixty-three forty-four, and sixty-four twenty-five.

We need buyers to hold the breakout and build a fresh base. Monday’s opening will decide.

The US market opens at thirteen thirty GMT, fifteen thirty Paris. Our earliest entry is fifteen minutes later.

Wait for two completed five-minute candles, with overlapping ranges and both closes above the pivot. Mark the base high, B, and low, L. The stop goes two ticks below L.

A later candle must close above B. The formula caps the purchase price and preserves at least one-point-five R to the first target.

Then one limit IOC attempt within five minutes. Fresh ask above B, at or below the ceiling; last price at least one tick above B. A return to B cancels the signal.

H P E has both a continuation path and a deep-pullback backup. The backup begins only after price touches the fifty-eight fifty to fifty-eight ninety zone. A later completed fifteen-minute candle must close above fifty-nine.

The purchase is capped at fifty-nine fifteen, with an initial stop at fifty-seven eighty. The fixed targets are sixty-two ten and sixty-four twenty. At the maximum purchase price, the gross reward-to-risk ratio to the first target is about two point one nine.

A pullback becomes a new thesis only when the decline is not explained by adverse company information. The recovery confirms demand; it does not erase delivery, margin, or valuation risk.

Dell is the second intraday priority. Its September first release reported forty-seven billion dollars of fiscal second-quarter revenue, up fifty-eight percent year over year.

Infrastructure Solutions Group revenue was thirty-one point eight billion, and AI-optimized server revenue was sixteen point four billion, up one hundred percent. These are reported revenue figures, distinct from orders, backlog, or full-year guidance.

The additional S and P one hundred inclusion creates a separate calendar catalyst. Together, these facts explain why we prepare the name early. AI server growth is not the same thing as durable free cash flow.

Supply, customer concentration, shipment timing, and competitive pricing can change the revenue-to-profit conversion.

For Dell, the reference is five hundred sixty-seven dollars and seventy-five cents. A continuation needs a base above that pivot. In the historical window previously inspected, no higher obstacle had been identified.

That statement is deliberately limited: it does not certify an all-time high, and a fresh Monday review can reveal new structure. Without a known resistance, targets must derive from the actual fill and the original stop.

The Dell backup is a much deeper setup. Price must first touch the five hundred thirty-five to five hundred forty zone, then a later fifteen-minute candle must close above five hundred forty dollars and fifty cents.

The maximum purchase is five hundred forty-one, the initial stop is five hundred thirty-two, and the fixed targets are five hundred sixty-two and five hundred sixty-seven fifty. The gross first-target ratio at the ceiling is about two point three three.

Semtech is the aggressive alternative. Its reported fiscal second-quarter sales were three hundred forty-one point nine million dollars, up thirty-three percent year over year. The September eighth product announcement concerns two hundred twenty-four gigabit optical connectivity components for advanced data-center architectures.

That is a product catalyst, not a disclosed revenue contribution or proof of broad customer adoption. The company also serves other markets, so we cannot attribute all historical growth to the new optical products.

Semtech’s reference is one hundred seventy dollars and forty-three cents. The nearby resistance cluster runs from one hundred seventy seventy to one hundred seventy-one thirty-seven. That cluster has to enter the ceiling calculation.

We cannot use the absence-of-resistance target rule while ignoring known highs that are already on the map. A wide base or a stop that sits too far below the entry can leave no acceptable purchase price above the base high.

If that happens, the correct classification is ceiling too low.

Semtech’s deep-pullback backup uses the one hundred fifty-nine to one hundred sixty-one zone. After a touch, a later fifteen-minute candle must close above one hundred sixty-one ninety.

The maximum purchase is one hundred sixty-two, the initial stop is one hundred fifty-eight fifty, and the targets are one hundred seventy thirty and one hundred seventy-five. At the maximum entry, the gross first-target ratio is about two point three seven.

Again, those numbers do not establish a probability of success.

Here is the relative performance that frames the comparison. Between September fourth and eleventh, the energy ETF X L E gained one point six nine percent. The semiconductor ETF S O X X gained one point three nine percent.

GRID gained one percent, while Q Q Q lost zero point five seven percent. That makes energy, semiconductors, and electrical infrastructure useful places to compare strength. But one week is a short observation. Relative strength is useful only while it persists.

Monday’s gap and first bases decide whether it remains investable.

The next chart puts the comparison on a common starting point. Rebasing each series to one hundred makes the path comparable without mixing dollar prices and ETF share prices. Watch how sustained leadership differs from a single strong session.

Also remember what this picture leaves out: it is a historical price comparison, not a total return portfolio and not a forecast. We do not infer tomorrow’s leader from the endpoint alone.

This matrix measures how daily returns moved together over the stated common window. A value near plus one means stronger co-movement in that sample; a value near zero means little linear relationship there.

It does not mean that two businesses are independent, or that their next loss cannot happen together. Historical correlation can change quickly around a macro event.

That matters because buying H P E, Dell, and several semiconductor names can concentrate the same broad AI spending exposure.

The AI label connects several businesses, but the economics are different. Taiwan Semiconductor sits upstream in manufacturing. Soitec supplies specialized semiconductor materials, while Semtech has an optical connectivity catalyst. H P E, Dell, and NetApp give us systems and data infrastructure exposure.

Fortum brings a separate electricity contract with a much longer economic timeline. These arrows describe a value chain, not measured cash flows between these companies. They also do not prove that all the shares must rise together.

Wednesday carries the densest event risk. August retail sales arrive at 12:30 GMT, 14:30 in Paris. The Fed decision follows at 18:00 GMT, 20:00 Paris, and the press conference begins at 18:30 GMT, 20:30 Paris.

Trip.com reports after the US close on Tuesday. Lennar reports after the US close on Wednesday; its call is Thursday at 15:00 GMT, 17:00 Paris. Thursday housing construction is 12:30 GMT, 14:30 Paris. Friday industrial production is 13:15 GMT, 15:15 Paris.

These are scheduled releases, not forecasts. Their purpose is to identify when uncertainty rises and when trade instructions must already be explicit.

Every position that could cross the Fed needs a recorded YES or NO before entry. With NO, Wednesday US submissions stop at 17:50 GMT, 19:50 Paris, and the position is out by 17:55 GMT, 19:55 Paris.

European intraday positions retain their 15:20 GMT, 17:20 Paris exit. With YES, the planned holding horizon remains, subject to exposure and combined-risk limits. If the choice is absent, no new position may cross the event.

Missing data does not create an automatic sale of an already-held position. That distinction keeps an incomplete trade sheet from silently changing the strategy during a headline-driven move.

The index changes become effective before the opening on Monday, September twenty-first. Bloom Energy, Everpure under ticker P, and Illumina join the S and P five hundred. Dell, Palo Alto Networks, Arista, and SanDisk join the S and P one hundred.

Engie and Nokia enter the Euro Stoxx fifty. This makes Friday’s closing activity worth watching, but it does not establish a guaranteed net buying imbalance.

An index announcement begins a chain of portfolio adjustments. Funds following the affected benchmarks may have to buy or sell, but the final market impact depends on weights, assets, existing positioning, and how execution is distributed through time.

Other traders can anticipate those flows. Opposing index changes can offset part of them. That is why we do not translate a confirmed addition into a guaranteed closing rally.

Our intraday plan still requires an exit at nine fifty in the evening Paris time.

After price crosses the pivot, wait for two completed, consecutive five-minute candles. Their ranges must overlap and both closes must be above the pivot. Freeze the base high as B, the base low as L, and width W as B minus L.

The original stop is L minus two actual exchange ticks. A strictly later five-minute candle must close above B. The two candles that define the base cannot also be the confirmation candle.

Freezing the structure prevents the entry rule from moving with a rising quote. Monday’s tick size must come from the instrument, not an assumption that every share trades in one-cent increments.

The first purchase ceiling is B plus one quarter of W, rounded down to the actual tick. When known resistance exists, first target T1 is one tick below it. Let S be the original stop.

The reward constraint gives a second ceiling: T1 plus one-point-five times S, divided by two-point-five. The permitted cap is the lower of these two ceilings. It must still be at least B plus one tick.

If that fails, the base cannot support a purchase. Confirmation can occur above the ceiling; it is the limit price, not the confirmation close, that stays capped.

After confirmation, the execution window is one five-minute interval. The ask must be strictly above B and at or below the ceiling. The last traded price must be at least one tick above B. Then make one limit IOC attempt only.

Abort before purchase if either ask or last returns to B or below, if the structural stop is touched, if the window expires, or if fixed T1 is reached before the buy. An unfilled IOC is not retried.

Any restart requires a new, complete base. The protocol records an unavailable setup rather than turning a late quote into a chase.

This table collects the backup numbers in one place for the presentation. The trigger is not simply a print at the recovery level. A fifteen-minute candle must touch the specified lower zone, and a strictly later fifteen-minute candle must close above C.

Only then does the five-minute execution window begin. The ask must be above C and within the maximum purchase price, while the last price must be at least one tick above C.

Position management starts from the average price actually paid. R is that average entry minus the original stop. Protect the quantity actually filled immediately. At the first target, sell half, rounded down to a whole share with a minimum of one.

Only after that entire tranche is effectively sold do we move the stop on the remainder to the average entry. Sell the balance at the second target.

Time is part of the setup. US entries may begin at 13:45 GMT, 15:45 Paris; Europe may begin at 07:15 GMT, 09:15 Paris, once the required structure exists.

Normal entry cutoffs are 18:00 GMT, 20:00 Paris for the US and 14:00 GMT, 16:00 Paris for Europe.

Intraday positions close by 19:50 GMT, 21:50 Paris in the US, and 15:20 GMT, 17:20 Paris in Europe, subject to the earlier Fed NO deadline.

After sixty minutes, exit a position that has never reached plus one-half R and remains at or below entry. A losing intraday position is never relabelled a swing merely to defer a decision.

We have defined the trading logic, but quantity still depends on the account. Sizing must satisfy the tightest constraint among risk to the stop, capital committed, the per-position ceiling, and combined exposure to correlated names.

That requires a current view of positions, working orders, cash, currencies, and the loss budget. We do not have a validated complete account state in this preparation, so the presentation does not propose a share quantity.

We now change horizon. The five swing candidates are NetApp, Soitec, Fortum, Taiwan Semiconductor, and Neste. Their catalysts and reference levels help us prepare, but the complete trade sheets do not yet exist.

Each one still needs a maximum entry, a structural stop, objectives, an intended holding period, and a Fed maintenance choice. A level in this section is therefore an alert or an obstacle, not an executable purchase instruction.

NetApp leads the swing preparation. Its September second release reported fiscal first-quarter revenue of two point zero two five billion dollars, up thirty percent year over year. All-flash revenue reached one point three billion, up forty-seven percent.

Public cloud revenue and billings also grew, while management raised guidance. The business mechanism is demand for data infrastructure across enterprise and cloud environments. That does not make every revenue dollar an AI dollar.

NetApp’s surveillance sequence is one hundred ninety-nine dollars and sixty-two cents, then two hundred two thirty-two, then two hundred nine zero-six. We do not automatically rename these three numbers entry, stop, and target.

They are reference levels to compare with the next structure. A continuation may become preferable if a clean base forms with sufficient room; a pullback may become preferable if it creates a tighter, defensible invalidation.

The choice depends on actual price behavior and new information.

Soitec’s catalyst is an outlook revision, so the tense matters. On September second, the company raised its expected fiscal second-quarter revenue growth to around fifty percent year over year at constant scope and exchange rates, from more than thirty percent previously indicated.

The acceleration is concentrated in Photonics S O I. Management did not announce that every business line had suddenly accelerated at the same rate.

Soitec’s reference sequence is one hundred forty-three euros and fifteen cents, one hundred forty-five thirty, and one hundred forty-eight seventy-five. Keep the currency visible. Comparing this chart directly with US dollar share prices would be meaningless without a return transformation.

The swing question is whether the price can consolidate the catalyst in a structure that defines a clear maximum entry and a structural stop. We do not have those completed fields yet.

Fortum offers a different kind of catalyst. The September ninth announcement describes a twenty-two-year nuclear power purchase agreement with Google. Delivery starts at reduced capacity in twenty twenty-eight and can reach fifty percent of Loviisa’s capacity for twenty thirty through twenty forty-nine.

That supports future economic visibility, not an immediate twenty twenty-six earnings uplift. The company’s estimate of an eventual improvement in return on net assets is conditional and forward-looking. There is also investment work associated with the plant’s extended life.

Fortum’s references are twenty-four euros and ten cents, then twenty-four seventy-four and twenty-five zero-two. These levels organize the swing review, but they are not a complete bracket.

We still need a maximum entry, a structural invalidation, targets tied to the chosen structure, a holding period, and the Fed choice. The company belongs in the wider electricity-demand discussion, yet it is not interchangeable with an AI server manufacturer.

Its economics depend on power markets, contractual terms, generation, and investment execution.

This timeline is the key to presenting Fortum accurately. First comes the announcement in twenty twenty-six. Reduced-capacity delivery begins in twenty twenty-eight. The larger contracted share applies from twenty thirty through twenty forty-nine.

Fortum estimates an eventual improvement of around one point four percentage points in comparable group return on net assets once fifty percent of capacity is contracted. That is an issuer estimate about the future.

The separate memorandum on possible new assets is exploratory.

Taiwan Semiconductor adds an upstream semiconductor perspective. The September tenth monthly report showed August consolidated revenue of about five hundred fourteen point eight one billion New Taiwan dollars.

That was up fifty-three point three percent year over year and ten point one percent from July. The unit is New Taiwan dollars, not US dollars, and the monthly consolidated figure is unaudited.

It is a realized revenue observation, but it cannot isolate AI demand, an individual customer, or one manufacturing node.

For the US-listed Taiwan Semiconductor shares, the surveillance levels are four hundred thirty-six dollars and four cents and four hundred forty-four twenty-nine. The daily chart helps us identify location, but the swing entry and stop remain unvalidated.

The stock can share a broad semiconductor factor with other positions while retaining its own currency, geographic, manufacturing, and customer-demand risks. Do not confuse an August revenue release with a forecast for the next quarter or a guaranteed response to the Fed.

Neste is our fifth swing preparation. Its second-quarter report showed comparable EBITDA of one billion two hundred three million euros, and a comparable Renewable Products sales margin of one thousand two hundred twenty-three dollars per tonne.

Those are different metrics in different units, so they should not be combined or treated as interchangeable. Neste described the second-quarter market environment as exceptional. We should therefore avoid annualizing the quarter or assuming the same margin persists.

Neste’s upper references are thirty-three euros and thirty-three cents and thirty-three seventy-nine. The lower zone, thirty euros and seventy-one to thirty eighty-four, must be revalidated before use. It is not a standing order and not an approved swing stop.

Because the thesis depends on margins, a fall caused by deteriorating economics deserves a different response from a routine market pullback. The chart alone cannot tell us which one has occurred.

Fortum and Neste both belong in an energy-related conversation, but they are not a single trade. Fortum’s catalyst is a long-term electricity contract. Neste’s near-term evidence concerns renewable-product margins and operating performance, with future capacity execution still ahead.

A change in power prices, a change in feedstock prices, and a change in regulation can affect these businesses in very different ways. The useful comparison is therefore mechanism by mechanism.

The reserve list keeps the process competitive. Matador can move forward if it offers a defensible local recovery and relative strength in energy, but its sheet needs rebuilding.

Iren, Neo, and I E S C require a better-placed aggressive setup and a check of their specific risks. Prysmian and R W E can become interesting if a clean intraday structure develops in electrical equipment or utilities.

P A C S and Natera need a consolidation breakout with identifiable invalidation.

Group One Automotive and P V H remain conditional on insider buying being accompanied by improving price behavior. An insider filing alone does not define a trade. S A I C requires a better entry while retaining the concern about order intake.

More broadly, the seventy-three-dossier research radar stays active. Fresh analyst revisions, insider transactions, and earnings should feed the same comparison at each update. We are not promising that every file receives an equally deep review every minute.

Monday begins with information that did not exist when this deck was prepared. Check company news, opening gaps, current prices, and the actual trading increment. Then observe the first completed structures and compare relative strength across the candidates.

For the swings, finish the missing fields instead of borrowing the intraday stop rules. Update the account before assigning quantity. This order of operations is deliberate: old reference levels can guide attention, but they cannot substitute for current evidence.

Here is the weekly sequence in practical terms. Monday converts the research into current structures. Tuesday repeats the comparison and prepares for Trip dot com’s after-close release.

Wednesday combines retail sales, the Fed decision, the press conference, and Lennar’s later results, so event instructions need to be settled early. Thursday separates the market’s digestion of the Fed from housing data and Lennar-specific information.

Friday brings industrial production and the index-rebalancing watch.

The interactive deck includes the source documents, so you can move from a claim to the original release. Use company investor-relations pages for the reported numbers and guidance. Use the Census and Federal Reserve calendars for scheduled macro events.

Use the index providers for composition changes, and the issuers for earnings timing. A corporate release can support what the company reported or expects; it does not prove a future share-price reaction.

The linked source hub also records the publication dates and units.

The protocol’s net expectancy is still unknown. To learn anything useful, the journal must distinguish an executed purchase, a ceiling that was too low, a cancelled signal, an unfilled I O C attempt, and the eventual net result.

A chart that later rises is not evidence that our capped order would have filled. A target touched before entry can invalidate the setup rather than count as a win. We also need actual fees, slippage, share rounding, and timestamps.

That is our attack plan for the week. H P E and Dell get first attention for intraday structures. NetApp, Soitec, Fortum, Taiwan Semiconductor, and Neste remain swing preparations until their sheets are complete.

Semtech stays available when its tight overhead structure still leaves enough room. The wider radar can overtake any of them. Keep the Fed choice explicit, keep the purchase capped, and protect only the quantity actually executed.