THE WEEK IN REVIEW The week began quietly and dishonestly. Monday's S&P slipped 0.16% while the Nasdaq 100 eked out a gain, but decliners beat advancers 338 to 164 and equal-weight lagged — a placid index sitting on a deteriorating median stock. The only real information in the session was structural: dealers were short gamma across all three major indices with the S&P roughly 3% below its flip level, and the options market was pricing a ±1.3% week into a calendar containing Alphabet, Tesla, Intel and ServiceNow. Volatility was, in effect, pre-paying nothing for the biggest earnings week of the quarter. Tuesday detonated. A sell-side call for at least 25% memory price increases into Q3 turned a two-day bounce into a melt-up — SanDisk +14.3%, Western Digital +12.5%, Micron +12.2%, Seagate +11.1% — dragging Technology up 2.9% and the Nasdaq 1.85%. The same trade sold its mirror image: information services and vertical software were liquidated (MSCI -10.1% on an expense guide, Equifax -3.9%, Tyler -5.7%, S&P Global -3.8%) on the day Adobe was cut to Underweight with a target slashed 34%. Long AI hardware, short AI-exposed software became one crowded position. The tell was that the index gained on negative breadth (229:271) while the VIX was crushed 8.6% to 17.05. Underneath it, crude rose 2.7% and gold 2.0% together on U.S.–Iran escalation — an unusual pairing the equity market chose to ignore. Wednesday looked flat and was not. The S&P moved 0.12% while utilities rose 2.25%, materials 1.44% and energy 1.20%, and enterprise software was liquidated for a second consecutive session (ServiceNow -6.5%, PTC -6.3%, Workday -6.2%, Palantir -6.1%). Super Micro's preliminary margin disclosure — gross margin guided to nearly double the prior range plus more than $60B of new orders — sent it up 19.8% and Dell 9.3%. Then the prints landed. Alphabet delivered the best top-line quarter in its history, Cloud accelerating to +82%, and fell after hours anyway on a 2026 capex guide raised to $195–205B. Tesla missed EPS by 35% with operating margin at 1.4% and free cash flow negative. ServiceNow beat cleanly and bounced. The reaction function had visibly changed: the market began charging for AI capital intensity rather than rewarding ambition.Thursday was the break, and it came from two directions at once. Brent punched through $100 (+7%) on Houthi attacks against two Saudi tankers and a U.S. threat to strike Iranian infrastructure, while Alphabet fell 7.1% and Tesla 14.5% in the cash session. The S&P lost 1.23%, the Nasdaq 1.90%, and the Mag 7 basket 4.95%. But the money did not leave — it moved into hard-asset earnings power: defense (Lockheed +10.5%, RTX +7.3%, both on raised guidance), rails (CSX +5.8%), life-science tools (Thermo Fisher +8.7%), equipment (United Rentals +10.1%). Equal-weight beat cap-weight by 86bp on a heavily negative index day. The macro consequence was immediate: fed funds futures moved from under 12% to weekly-2026-07-24 a 38% probability of a hike at the July 28–29 meeting, and the 10-year printed its high for the year. Intel then reported its fastest revenue growth since 2011 after the close. Friday inverted everything and settled nothing. The S&P closed up 0.10% and the Dow up 0.48% while the Nasdaq fell 1.12%, and the internals were the strongest of the week: advancers beat decliners 363:140, equal-weight outperformed by 68bp, 52 names made new 52-week highs against 8 new lows. The entire index drag was one complex — SK Hynix's plan to raise roughly $29B through a U.S. listing detonated the memory book (SanDisk -10.8%, Micron -7.0%, Western Digital -6.9%), and Intel closed down 7.9% after beating EPS by 91%. MaxLinear guided next-quarter revenue about 22% above consensus and fell 21.5%. The capital rotated into what the AI narrative had left for dead — ServiceNow +7.4%, Adobe +6.1%, Accenture +6.0%, Intuit +5.3% — alongside data-center REITs (Digital Realty +11.0% on a 34% core FFO beat), packaging (International Paper +11.2%) and energy services (SLB +11.0%). Crude retraced 3.1% on reported U.S.–Iran mediation, which is the only reason equities held.Now the part the daily tape obscures. Netted across the full week, the memory complex the market "crashed" on Friday finished higher — Micron +8.5%, Western Digital +8.9%, Seagate +8.1%, SanDisk +6.0%, Teradyne +8.6% — because Friday only clawed back part of Tuesday's melt-up. Meanwhile the software complex that "squeezed" on Friday finished down: the software group lost 5.2% on the week against a semiconductor group up 1.0%, with ServiceNow -4.3%, Adobe -5.1% and Workday -6.5%. Friday was a partial round trip, not a regime flip, and anyone who read only Friday's tape has the week backwards. The same correction applies to breadth: over a week in which the S&P fell 0.59%, the share of index members above their 20-day average rose from 53.5% to 58.8%, and the share above the 200-day was unchanged at roughly 67%. The damage was cap-weighted concentration, not market-wide deterioration. The market therefore ends the week having spent five sessions demonstrating that at the single-name level, results do not set prices — positioning does. It walks into a week in which a divided Fed and four mega-cap prints are supposed to do exactly that, with dealers short gamma and spot 3.1% below the level where hedging turns stabilising.THE WEEK'S DOMINANT THEMES The market stopped paying for results and started paying for positioning. The dispersion this week was not explained by what companies printed. Intel beat EPS by 91% with its fastest revenue growth in fifteen years and closed the week down 2.9%, falling 7.9% on the day after the print. MaxLinear guided next-quarter revenue roughly 22% above consensus and fell 21.5% in a session. SAP missed EPS by 9.1% and rose 9.3%. Verizon missed revenue and rose 5.8%. T-Mobile beat EPS by 15% and fell 10.8%. Danaher beat and raised and fell 11%. The single variable that explains all six is what was owned into the print: Intel entered July up 163% year-to-date and is now down 28% month-to-date; SAP and the software complex entered the week already de-rated on the AI-disruption thesis and had no marginal seller left. This matters more than a curiosity. It means fundamental surprise is temporarily not a tradable catalyst in the crowded end of the market, and it means the reaction function into next week's prints is a positioning question, not an estimates question. The quantitative frame is stark: the 25 best-performing Russell 1000 names of the first half, averaging +235%, have surrendered roughly a quarter of their value in under three weeks, while the 17 worst first-half names have bounced about 10.6%. That is a mechanical factor unwind, and mechanical unwinds end when the forced seller is done, not when the news improvesAI capex split into writers and recipients — and then the recipients split too. Alphabet delivered the best cloud quarter of the AI era and lost 7.1% on a capex guide raised to $195– 205B with buybacks reportedly suspended. Tesla lost 14.5% burning cash on autonomy and robotics. The market has flipped from rewarding AI ambition to demanding AI return on investment, and that framing now governs how every hyperscaler print will be graded. The more useful observation is what happened on the recipient side, because it did not move together. Digital Realty posted a 34% core FFO beat, record backlog, and raised cash renewal spreads to +9–11%, and rose 11.0% on the same day memory lost 7–11%. SLB's data-center solutions line grew 80% with management guiding it to a $1B-plus annualised run-rate by year-end, and the stock rose 11.0%. Those are demand signals. What broke in memory was a supply event — a roughly $29B U.S. listing plan from the largest producer, layered on capacity additions — not a demand event. Data-center demand and memory pricing economics decoupled visibly this week, and they should be traded as separate exposures rather than as one "AI" position3. Oil became the Fed's policy variable, and therefore the equity multiple's. Crude rose 8.3% on the week, with Brent through $100 on Thursday before Friday's mediation headline pulled it back. In five sessions, market-implied odds of a hike at the July 28–29 meeting moved from under 12% to roughly 38%, with 82% by September, and the 10-year added 14bp to 4.68%. Nothing in the labour data caused this. The entire repricing came through the energy channel, and it landed on a Committee that is publicly split — the New York Fed arguing inflation has peaked and the Dallas Fed calling for modestly higher rates. The cross-asset evidence says the market is treating this as a multiple problem, not a cycle problem: high-yield credit fell only 0.53% on the week, investment grade's decline was duration rather than spread, and inflation-protected Treasuries outperformed nominals — breakevens widening, not growth expectations breaking. That distinction is the whole argument for why this remains a rotation to trade rather than a top to sell. It also means the fastest way to break the current setup is not an earnings miss; it is a Hormuz headline that stalls the mediation and re-accelerates the hike trade into Wednesday's statement.MACRO & RATES: THE GOVERNING FORCE The rate move this week was entirely a term-premium and inflation-expectation event, not a growth event. The 10-year rose 14bp to 4.68%, roughly 18-month highs, while the 3-month firmed to 3.81% as July hike odds re-priced — leaving the 10s-3s spread at about +0.87pp, a curve that steepened on the long end rather than inverting on the front. The dollar firmed 0.7% but gave no strong confirmation, and gold rose 0.95% into firmer real yields, which is a debasement and geopolitical bid rather than a rate trade. Credit is the control variable and it barely moved. High yield fell 0.53% across a week containing a 17.8% decline in Tesla, an 11% memory drawdown and a 7.8% fall in a $3.9T mega-cap. Investment grade's 1.24% loss maps almost exactly to the duration move. Inflation-protected Treasuries outperformed nominals. The composite read is unambiguous: the market is pricing an inflation and discount-rate shock, and explicitly not a credit or default cycle. The regime the market is currently priced for is a hawkish hold on Wednesday with a hiking bias attached — no change to the 3.50–3.75% funds rate, but statement language that removes the cutting option for 2026. Two things would flip it. Upside: a benign core PCE print on Thursday combined with successful U.S.–Iran mediation collapses the hike trade and unwinds the term-premium build, which is straightforwardly the pain trade higher given how little index protection is owned. Downside: crude reaccelerating through $100 before Wednesday, which would force a hawkish statement into a market where the entire year-to-date multiple was built on an easing bias, and where Monday's $139B of 2-year and 5-year supply provides an early tell on whether the front end can absorb it.FACTOR & STYLE + THEMATIC BASKETS The multi-week regime is now unambiguous and this week confirmed rather than countered it. Over one month, Momentum is -6.7% and High Beta -6.5% against Value +2.2%, Low Volatility +2.2% and Quality +1.7% — the textbook signature of a leadership handoff from crowded growth into cash-flow defensives. Over three months the picture is still the old regime (Momentum led), which is precisely why the unwind has fuel: the market remains positioned in the three-month winners while the one-month tape has already turned against them.Within the week itself, the honest reading is more two-sided than Friday suggested. Momentum actually finished the week up 1.4% because Tuesday's violent snap-back outweighed Friday's -2.4%, and High Beta was roughly flat. Equal-weight beat the cap-weighted index. Small caps lagged 0.98% but held their 50-day average, the only major index to do so. Emerging markets were flat on the week against -5.8% on the month, and long Treasuries fell 1.4% — no flight to duration, because this was an inflation shock rather than a growth shock. The single most important divergence remains credit versus equity dispersion: high yield unchanged while the most crowded factor in the market repeatedly gapped. Money rotating, not leaving. The thematic baskets tell the same story at higher resolution and add one correction. Defense was the week's cleanest winner — the aerospace and defense complex rose 4.1% on raised guidance from Lockheed and RTX arriving simultaneously with Middle East escalation, a rare case where geopolitical narrative and order books moved together. Software was the week's worst theme at -5.2% despite Friday's squeeze, and semiconductors finished +1.0% despite Friday's crash — the inverse of the impression the last session left. Biotech fell 2.5%. Friday's basket tape, taken alone, showed every longduration story theme liquidated together (Quantum -5.7%, Crypto -4.1%, Nuclear -3.3%, GLP-1 -2.9%) while near-term-earnings themes were bid, but across the full week the nuclear and power complex was actually up on AI power-demand headlines and Constellation gained 8.7%. The durable read-through: within AI, capital is migrating from second- and third-derivative story exposure toward assets with contracted revenue and near-term cash generation.EARNINGS SEASON PULSESeason scorecard. With roughly 27% of the S&P 500 reported, 86% have beaten on EPS against a 78% five-year average and 80% have beaten on revenue against a 70% average. Blended earnings growth is +37.9%, but that figure is distorted by a single company — excluding Alphabet's investment-gaininflated print it is +25.9%, which is the number to use. Blended revenue growth is +13.2%. Ten of eleven sectors are growing year over year, seven at double-digit rates, led by Energy, Communication Services, Information Technology and Materials; Health Care is the only sector with declining earnings, which sits oddly against its status as one of the strongest sectors in the tape and is worth watching as the hospital and tools cohort reports throughThe behavioural read: Three rules governed reactions this week and all three are about the marginal holder rather than the marginal dollar of earnings. First, crowded winners were sold on beats — Intel, MaxLinear, Teradyne's complex, T-Mobile. Second, de-rated names were bought on misses — SAP rose 9.3% on a 9.1% EPS miss, IBM rose after missing both lines because a pre-announcement had already reset expectations. Third, guidance beat results everywhere: Verizon missed revenue and rose on a raise, Tenet's move was about the full-year lift rather than the quarter, GE Vernova was forgiven a 19% EPS miss because it raised revenue and free-cash-flow guidance. Richly-valued compounders were punished asymmetrically — Rollins lost 16% after hours on a modest miss at roughly 45x earnings. The practical implication for next week: the estimate line is close to irrelevant for names that ran into their prints, and the guidance and capital-allocation lines are close to everything.WHAT THE SMART MONEY WAS DOING Positioning evolved through three distinct phases across the week. Monday opened with dealers short gamma across all three indices and spot roughly 3% below each flip, with index skew only modestly putbid (S&P risk reversal +4.3) — hedged but not braced. By Tuesday the vol complex went outright complacent: the VIX was crushed 8.6% to 17.05 and the weekly index straddle priced just ±1.0%, an extraordinarily cheap ask for a week containing four of the quarter's most consequential prints. Wednesday flipped it: the equity put/call ratio printed 1.60 — firmly in fear territory — while the VIX kept falling into the mega-cap night, a disagreement between options positioning and index volatility that resolved, as it usually does, in favour of the options market when Thursday arrived. Thursday's break brought the only genuine vol event of the week, a 12.4% VIX spike, but the threemonth contract barely moved and the curve held contango — the market treated it as an event-window problem, not a regime change. Then Friday delivered the week's most informative positioning fact: the VIX fell on a day the Nasdaq lost 1.1% and the memory complex lost 7–11%. Nobody paid up for index protection into a Fed meeting and four mega-cap prints. That is a cheap-hedge setup, not a comfortable one. Friday's settled positioning map is where next week starts. Net dealer gamma is -$2.5B on the S&P proxy and -$1.2B on the Nasdaq proxy, with spot 3.1% below the flip at 761.88 — hedging will accelerate moves rather than dampen them. Of the names scanned, nine carry negative gamma against eight positive, but the negative ones are the systemically important ones, which is precisely why single-name moves were so violent against a nearly flat index. Skew is bid at the index level and nowhere else: the S&P risk reversal is +4.6 and the Nasdaq +5.6, while the average across all scanned names is only +2.1, with the extremes isolated in single names — SanDisk's +11.4 risk reversal on a 136% implied-vol base is the most expensive protection in the scan. Two flow observations sharpen the picture. First, the institutional-versus-speculative split classified the week's aggregate flow as short-term speculation, weekly-dominated: roughly $144M of qualifying premium, of which only about $6M sat in long-dated institutional structures. Second, and more useful, the direction of that flow was cleanly bifurcated — index premium was overwhelmingly put-heavy (Nasdaq proxy roughly $68M of puts against $4M of calls; S&P proxy $36M against $5M), while every mega-cap scanned was call-heavy, with zero qualifying put premium in Microsoft, Apple, Meta or AMD. The market is hedging the index and speculating long the individual names reporting into it. That is not conviction; it is an attempt to own the upside of the prints while renting protection against the Fed.THE FORWARD VIEWNext week (tactical). The base case is a contained, violent-feeling week that resolves higher, and the specific reason is that positioning is braced for damage that consensus estimates do not support. The bar into all four mega-cap prints has been flat for thirty days — Microsoft's June-quarter consensus moved from $4.242 to $4.240 on 36 estimates, Meta's from $7.174 to $7.184 on 48, Apple's and Amazon's essentially unchanged on 33 and 47. Nobody has raised the hurdle, which means the EPS line is not the risk; the capex line is, because Alphabet reset that bar at $195–205B and neither Microsoft's $87.6B revenue consensus nor Meta's $60.2B captures it. Meta is the sharpest expression: consensus EPS of $7.18 against $60.2B of revenue implies roughly flat year-over-year earnings on mid-20s revenue growth — the cost line already carries the entire story, so the asymmetry into an in-line capex guide skews positive. Against that, the tape is genuinely fragile in one place. The Nasdaq proxy closed at 684.23 sitting directly on its 680 put wall — the largest put open interest in the chain and the last dealer-supported floor. A break below 680 with net gamma at -$1.2B hands the tape to trend-amplifying hedging with the next meaningful shelf far below, and the ±3.2% weekly expected move (663–706) says the options market has already sized that scenario. On the S&P side the first real floor is the 725 put wall, 1.9% below spot and just inside the ±1.8% weekly band, with max pain at 745 exerting a mild upward pull against the gamma regime. The pain trade is higher and it is not close. Index protection is objectively cheap — S&P implied vol at 15% against 14% realized, Nasdaq implied below realized at 26% versus 28%, the VIX falling on the week's worst dispersion day — while flow was overwhelmingly index puts against mega-cap calls. A hawkish-hold statement plus in-line capex from Microsoft and Meta would leave a market that is hedged, not positioned, chasing. The single event most likely to set the tone is Wednesday's 2:00 ET decision, not the earnings: the Committee is publicly split, hike odds moved from under 12% to 38% in five sessions on oil alone, and with dealers short gamma the statement language will be amplified whichever way it lands. The specific downside sequence to respect is crude re-accelerating through $100 into a hawkish statement, which breaks 680 and makes the ±3.2% band the floor rather than the range.Next few months (medium). The medium-term setup is a leadership handoff that has now persisted long enough to be treated as the operative regime rather than a bounce. One-month factor data shows Momentum -6.7% and High Beta -6.5% against Value +2.2%, Low Volatility +2.2% and Quality +1.7%; one-month sector data shows Energy +11.3%, Health Care +6.0% and Financials +4.8% against Technology -3.9% and Consumer Discretionary -4.9%. Three-month data still shows the old regime, which is the point: the market is positioned in the winners of a regime that has already turned. Until Momentum and High Beta turn positive on a one-month basis, the correct default is that semiconductor bounces are rallies to sell and value and defensive breakouts are the trend. What the market is underwriting over the next quarter is a benign resolution of the energy shock — a hold that never becomes a hike, crude normalising below $90, and AI capex that eventually produces visible return. The season's fundamentals support the first-order case: an 86% EPS beat rate against a 78% five-year average, 80% on revenue, +25.9% blended earnings growth excluding Alphabet, and ten of eleven sectors growing. The evidence that would confirm the underwriting is a soft core PCE print on July 30, a durable mediation outcome that keeps Brent below $95, and hyperscaler capex guides that come in at or below the reset bar with margin commentary attached. The evidence that would break it is a second-round wage signal in the July 31 employment cost index, or a hyperscaler raising capex again without a return framework — which would extend the "charge for capital intensity" regime from Alphabet to the whole complex and put the Nasdaq's 26% year-end implied vol, against 17% for the S&P, to work. The asymmetry worth holding through the quarter is the one the tape created this week and has not yet corrected: data-center demand and memory pricing decoupled visibly. Digital Realty raised renewal spreads to +9–11% with record backlog and SLB guided a data-center services line to $1B-plus, both on the same day the memory book broke on a supply-side listing event. Vertiv on Wednesday and Amphenol on Wednesday are the confirmations. If their orders corroborate, the expression is long the infrastructure and power chain against the commodity-memory chain, not short "AI." The volatility surface already agrees and prices it wrongly in a usable way: implied vol sits below realized in the names that actually moved on the catalyst (Micron -14.6 points, SanDisk -13.4, Western Digital -8.1) and far above realized in the sympathy names that moved on nothing (Lumentum +20.0, Coherent +15.1, Flex +13.7).THE DEBATE: WHAT'S PRICED IN? Debate 1 — Is a July hike a real risk, or has the market over-extrapolated an oil spike? What the price implies. Roughly 38% odds of a hike Wednesday and 82% by September, a 10-year at 4.68%, and a 3-month at 3.81%. Yet the S&P weekly straddle prices only ±1.8% and the VIX fell on the week — equity vol is not pricing a live policy surprise even as rates markets are. The bull case. The energy spike is already retracing — crude fell 3.1% Friday on reported mediation, and Brent's move off $100 removes the mechanical driver of the entire repricing. The Committee's own leadership says inflation has peaked with a glide to roughly 3.25% by year-end. Credit never confirmed the stress: high yield fell 0.53% across the week's worst dispersion. A hawkish hold, in this framing, is the ceiling of the hawkishness, and the pain trade is a relief rally into thin protection. The bear case. June CPI at 3.5% with wholesale prices +5.5% removes the doves' "transitory energy" argument entirely, and new 10% and 12.5% tariffs on 60 trading partners layer a second cost shock arriving days before Q2 GDP and PCE. A visibly split Committee raises the variance of Wednesday's outcome, and inflation-protected Treasuries outperforming nominals says breakevens are already widening. The entire year-to-date multiple was built on an easing bias. Our read. The hike itself is unlikely; the hawkish language is not, and it is the language that matters with dealers short gamma. The asymmetry favours owning cheap index protection through Wednesday rather than expressing a directional view — Nasdaq implied vol below realized into an FOMC plus four megacap prints is the mispricing, not the direction. The cleanest tell is not the statement but Monday's $139B of 2-year and 5-year supply at 1:00 ET: a weak tail there tells you the front end cannot absorb the hawkish scenario, and equities will find out before the Fed speaks.Debate 2 — Did AI capex just get repriced as a cost, or did only memory break? What the price implies. Alphabet lost roughly 7.8% on the week and $280B of market value on a capex raise despite the best cloud quarter of the AI era, while Digital Realty gained 14.5% and SLB 11.6% on data-center demand. The Nasdaq's year-end implied vol of 26% against the S&P's 17% prices an explicit AI-concentration premium. The bull case. Demand never cracked. Cloud accelerated to +82% at Alphabet, Digital Realty raised cash renewal spreads to +9–11% with record backlog, SLB guided a data-center services line from $186M toward $1B annualised by year-end, and Intel's Data Center and AI segment grew 59%. What broke in memory was a supply event — a ~$29B U.S. listing from the largest producer plus capacity additions — which is a pricing-curve problem, not an end-demand problem, and it does not touch power, cooling, connectivity or real estate. The bear case. The capital structure funding the build changed this quarter, and that is not a memoryspecific issue. Alphabet raised ~$70B of equity, preferred and debt and halted buybacks; combined bigfour 2026 capex is near $725B, up ~77% year over year. If the market has genuinely flipped from rewarding ambition to demanding return, then every recipient is only as safe as the writers' willingness to keep writing — and Microsoft, Meta and Amazon all guide within four sessions into a rate environment that has become hostile. Our read. The split is real and tradable, but its half-life is one week. Vertiv and Amphenol report Wednesday and are the cleanest test: if their order books corroborate Digital Realty and SLB, own the infrastructure and power chain against commodity memory. The volatility surface is currently paying you to do exactly that — implied vol below realized where the catalyst is (Micron, SanDisk, Western Digital) and far above realized where the move was borrowed (Lumentum, Coherent, Flex).Debate 3 — Is the momentum unwind a mean-reversion trade or a leadership handoff? What the price implies. The 25 best first-half Russell 1000 names, averaging +235%, have given back roughly a quarter of their value in under three weeks while the worst first-half names bounced ~10.6%. One-month Momentum is -6.7% against Value +2.2%. Yet the market still prices Nasdaq risk at nearly twice S&P risk, meaning it has not yet re-rated the concentration. The bull case for mean reversion. Momentum finished this week up 1.4%, semiconductors +1.0% and memory names up 6–9% — Friday's crash was a partial round trip of Tuesday's melt-up, not a new leg. Breadth improved across the week (58.8% above the 20-day, up from 53.5%), 52 new highs against 8 lows, credit unmoved, and the share above the 200-day flat at ~67%. Nothing structural broke; the winners are simply oversold, with Intel's RSI at 27 and Corning's at 25. The bear case for a handoff. The one-month rotation is now three weeks old and has survived four separate catalysts. Every attempt to buy it on fundamentals has failed: Intel beat by 91% and fell 7.9%, MaxLinear guided 22% above consensus and fell 21.5%. When beats of that magnitude are sold, the marginal seller is mechanical and mechanical selling does not care about the next print either. The sell side is rotating the same way — the week's upgrade board was entirely cyclical and hard-asset while software and IT services were being cut. Our read. Handoff, with a tactical bounce inside it. The evidence bar is explicit and worth writing down: this becomes mean reversion the moment Momentum and High Beta turn positive on a one-month basis while Value and Low Vol roll over — and the first concrete tell arrives Tuesday, when Seagate (±14.4%) and Teradyne (±16.3%) report into the same setup that buried Intel. The first semiconductor name that holds a beat marks the bottom of the unwind. Until then, treat bounces in the crowded complex as exits, not entries.NOTABLE MOVERS OF THE WEEKKEY SIGNALS FOR THE WEEK AHEADThe 680 put wall on the Nasdaq proxy. The index closed Friday at 684.23, directly on its largest put open-interest strike, with net dealer gamma at -$1.2B. Trigger: a close below 680. That removes the last dealer-supported floor and hands the tape to trend-amplifying hedging; the ±3.2% weekly expected move (663–706) becomes the realistic range rather than the tail. A hold above it, conversely, keeps 705 max pain as an upward magnet into Friday's expiry. Core PCE, Thursday July 30 at 8:30 ET. Landing the morning after the Fed, inside the most datadense hour of the quarter alongside Q2 GDP and personal spending. Trigger: a core print that annualises above the June CPI's 3.5% headline path. That validates the hike trade the oil shock started, extends the term-premium build past a 10-year already at 4.68%, and gives Friday's defensive rotation a second leg. A soft print collapses the September hike odds currently near 82% and is the cleanest path to the pain-trade rally. Whether any semiconductor name can hold a beat. Seagate (±14.4%) and Teradyne (±16.3%) report Tuesday after the close, Lam Research (±12.8%) and Arm (±15.4%) Wednesday, into the same setup that took Intel down 7.9% on a 91% EPS beat and MaxLinear down 21.5% on a 22% guide raise. Trigger: the first name to print a beat and close green. That marks the exhaustion of the mechanical seller and turns the crowded complex from an exit into an entry. Until it happens, treat every bounce as supply. Microsoft's and Meta's capex lines, Wednesday July 29 after the close. The EPS bar is not the risk — consensus has been flat for thirty days at $4.24 for Microsoft on 36 estimates and $7.18 for Meta on 48, against revenue of $87.6B and $60.2B. Trigger: a 2027 capex framework at or below the $195–205B bar Alphabet reset, with a return-on-investment framing attached. That confirms the recipients' demand case and squeezes a market positioned in index puts against mega-cap calls. Another unframed raise extends the "charge for capital intensity" regime from one name to the whole complex. Crude and the front end as a single signal. Brent's move through $100 and back is what drove hike odds from under 12% to 38% in five sessions; nothing in the labour data did. Trigger: Brent re-taking $95 before Wednesday, or a weak tail on Monday's $139B of 2-year and 5-year supply at 1:00 ET. Either one tells you the hawkish scenario is being funded and priced before the Fed speaks; the 2-year yield is the cleanest real-time vote on Wednesday's outcome.CALENDAR & CATALYSTSMonday July 27 - Durable Goods Orders and ex-Transport (June), 8:30 ET — first capex-intent read after the new 10% and 12.5% tariffs on 60 trading partners; core capital goods is the cleanest signal on whether industrial spending is absorbing input costs. - Dallas Fed Manufacturing (July), 10:30 ET — energy-weighted regional survey, the first datapoint to capture the $89–97 crude regime. - Treasury sells $69B 2-year and $70B 5-year, 1:00 ET — $139B of front-end supply into a market pricing rising hike odds. A weak tail is a direct equity risk with the 10-year at 4.68%. - Earnings (AMC): AstraZeneca ($263B) — oncology growth and US pricing under tariffs; Nucor ($56B) — steel realised pricing, the fundamental test of the materials bid; Celestica ($35B) — the first AI-hardware read since Friday's liquidation. Tuesday July 28 — the heaviest earnings day of the week - Consumer Confidence (July), 10:00 ET; CaseShiller 20-city (May), 9:00 ET — Confidence is the crosscheck on the American Express revenue miss: if the high-income consumer is softening, it shows here first. - Treasury sells $44B 7-year, 1:00 ET — completes the week's coupon supply ahead of the Fed. - Earnings (BMO): Coca-Cola ($354B, ±3.2%) — organic growth and pricing/mix as input costs rise; Boeing ($165B, ±6.0%); Corning ($126B, ±13.0%) — the direct AI-connectivity read after optical fell 8–10%; UPS ($98B, ±7.0%); PayPal ($50B, ±8.7%). - Earnings (AMC): Visa ($677B, ±3.9%) — the direct crosscheck on the AXP miss; KLA ($275B) — wafer-fab equipment outlook; Seagate ($193B, ±14.4%) and Teradyne ($55B, ±16.3%) — the first memory and semicap prints since Friday's liquidation, into negative-gamma positioning; Ford ($57B, ±6.7%). Wednesday July 29 — the week's dominant day - FOMC decision, 2:00 ET. Consensus is no change from 3.50–3.75%, but hike odds moved from under 12% to roughly 38% on the oil shock and the Committee is publicly split. With spot 3.1% below the gamma flip, dealer hedging amplifies the reaction in either direction. This is the single most important event of the month. - Earnings (AMC): Microsoft ($2,835B, ±7.2%) — consensus $4.24 EPS on $87.6B revenue, bar flat over 30 days; Azure growth and the FY27 capex guide against Alphabet's reset bar. Meta ($1,511B, ±8.4%) — consensus $7.18 EPS on $60.2B revenue, bar flat over 30 days; EPS implies roughly flat year-over-year on mid-20s revenue growth, so the cost line is the entire story. Also Arm (±15.4%, 140% ATM IV), Lam Research (±12.8%), Qualcomm (±9.7%), Starbucks (±6.5%) — the clearest discretionary read of the week, Fortinet, Chipotle, O'Reilly. - Earnings (BMO): Vertiv ($112B, ±11.9%) and Amphenol ($188B) — the cleanest test of the data-center-demand-versus-memory divergence; Boston Scientific — med-tech confirmation of the hospital utilisation cycle; SoFi, Generac.Thursday July 30 - Q2 GDP, Q2 and June PCE / Core PCE, Personal Income & Spending, Jobless Claims — all 8:30 ET. The most data-dense hour of the quarter, landing the morning after the Fed. Core PCE decides whether the hike trade extends or unwinds. - China NBS Manufacturing and Non-Manufacturing PMI (July) — read-through to China tech, industrial commodities and the materials bid. - Earnings (AMC): Apple ($4,891B, ±4.3%) — consensus $1.89 EPS on $108.8B revenue, bar flat over 30 days; the lowest implied move of the mega-caps. Amazon ($2,497B, ±7.2%) — consensus $1.81 EPS on $196.7B revenue, flat over 30 days; AWS growth and the 2027 capex guide are the single most important readthrough to the entire AI infrastructure complex. Also Roblox, Reddit. - Earnings (BMO): Mastercard ($477B, ±4.1%); Bristol Myers; Valero; Cigna (±6.4%) — the cost side of the utilisation cycle Tenet just monetised; International Paper (±9.6%) — must validate an 11.2% pre-print move made on target revisions alone.Friday July 31 - Employment Cost Index (Q2), 8:30 ET — the Fed's preferred wage gauge and the cleanest read on whether the energy and tariff shock is becoming a second-round wage problem. - Chicago PMI, 9:45 ET; UMich final including inflation expectations, 10:00 ET. - Earnings (BMO): Exxon ($651B, ±4.2%) and Chevron ($388B, ±4.0%) — whether $89–97 crude translated into upstream earnings or was offset by refining margins. Ongoing — Hormuz and U.S.–Iran mediation headlines. Brent above $100 is now the de facto Fed policy variable; Friday's 3.1% crude retracement on reported mediation is the only reason equities held, and a stall re-accelerates the entire hawkish repricing.