This is a market recap, not financial advice. See our Disclaimer.
Treasury yield hits 24-year high
Bonds stole the show on the last day of the quarter. The 10-year Treasury yield climbed to 5.304% in afternoon trading, pushing past its 2007 intraday peak of 5.303% to its highest level since May 2002. The trigger was data: the Commerce Department revised second-quarter GDP growth up to 2.2% from a prior estimate of 1.5%, while the Fed’s preferred inflation gauge — the PCE price index — rose at a 3.4% annual rate in August, matching July. Strong growth plus sticky inflation is a rough mix for rate-cut hopes. The Dow fell 443.87 points (0.9%) to 50,906.05 and the S&P 500 lost 0.25% to 7,651.54, but tech held its ground: the Nasdaq edged up 0.24% to 26,861.06. Oddly, the same data made investors less worried about a Fed hike — after New York Fed President John Williams said the central bank is in no rush, October hike odds fell to about 37% from 68% Tuesday morning on the CME FedWatch tool.
The day’s strangest move belonged to Jabil (JBL). The contract manufacturer beat every line of its fiscal fourth-quarter results — revenue jumped nearly 28% to $10.6 billion versus the $9.7 billion analysts expected, and adjusted earnings of $4.40 a share topped the $4.07 consensus. Fiscal 2026 finished at $36 billion in revenue with adjusted EPS of $13.09. Management then guided fiscal 2027 above Wall Street’s expectations too: $44.5 billion in revenue (+24%) and $17.55 in adjusted EPS, crediting booming AI and data-center demand. Investors still sold — shares tumbled 10% to $285.86, the stock’s worst day since April 3, 2025 and the S&P 500’s worst performer of the day. CEO Mike Dastoor pointed to progress “moving up the value chain” on AI infrastructure, but after a 34% run this year, the market wanted more than a beat.
Cerebras (CBRS) fell 8.9% after a painful read on its biggest-adjacent relationship: semiconductor research firm SemiAnalysis said in a social media post Tuesday that OpenAI is using Nvidia chips rather than Cerebras hardware to power “Ultrafast” mode for its latest GPT6.1 artificial-intelligence model. For a company whose investment case rests on AI-chip alternatives to Nvidia, losing the marquee AI lab’s flagship feature — publicly, and on the last trading day of the quarter — was a sharp confidence shock. The drop showed how much of the AI trade still runs through one supplier’s ecosystem.
Hewlett Packard Enterprise (HPE) gained 3.9% after raising its fiscal 2027 revenue outlook for its networking business segment ahead of an investor day event — a reminder that AI spending isn’t just chips. The move fit a broader hardware theme: FormFactor (FORM) jumped 9.6% after Deutsche Bank praised the chip-testing equipment maker as a key supplier for Nvidia, with shares now up roughly 310% over the past 12 months. With Micron reporting after the close today, the whole AI-infrastructure complex was the one corner of the market with real buying conviction on Wednesday.
Boeing (BA) won a multibillion-dollar contract to build a Navy fighter jet, beating out Northrop Grumman (NOC). The market’s reaction was muted on the winner — Boeing dipped 0.9% — and punishing for the loser, with Northrop sliding 4.2%. The award reshuffles one of the defense industry’s most watched programs, though as always, contract awards can be protested or restructured before a single jet is built.
This is a market recap, not financial advice. See our Disclaimer.
Yields hit 5.6%; stocks slip off lows
The bond selloff kept its grip on Tuesday’s session. The 30-year US Treasury yield spiked to 5.612% — its highest since 2002 — before settling at 5.59% for its highest close since June 10, 2002. The 10-year note reached 5.29% and closed at 5.26%, its highest 3 p.m. close since May 17, 2002. Equities slipped: the S&P 500 fell 0.2%, the Dow dropped 131 points (0.3%), and the Nasdaq lost 0.1%. Seven S&P sectors fell while four rose — energy was the worst at -0.9%, utilities the best at +1.1% — and Apple was the Dow’s biggest loser at -2.7%. The afternoon brought relief: New York Fed President John Williams signaled there is “no urgency” to hike rates in October, sending 2-year yields lower and cutting October hike odds from 70.9% to 51.5% on the CME FedWatch tool. Soft JOLTS and consumer-confidence data helped the cause; the test comes Wednesday with August PCE inflation data, followed by September payrolls on Friday.
Carnival (CCL) was the S&P 500’s biggest gainer Tuesday, jumping 13% to about $24.93 — its largest single-day gain since April — after reporting record third-quarter results. Adjusted earnings per share came in at $1.43 versus the $1.35 consensus, while revenue hit a record $8.44 billion, up 3.5% year over year and above the $8.35 billion analysts expected. The cruise operator raised its full-year adjusted net income forecast by more than $150 million versus June guidance — even while absorbing roughly $150 million in extra fuel costs — and said customer deposits reached a third-quarter record of $7.6 billion, up 7%, with 2027 booking volumes and pricing at record levels. CEO Josh Weinstein pointed to sustained same-ship yield growth and guided fiscal 2026 net yields up 3.8% versus the 3.2% consensus. The glow lifted rivals: Royal Caribbean rose 6.5% and Norwegian Cruise Line added 3.4%. Even after Tuesday’s pop, Carnival shares remain down roughly 20% on the year.
Reuters’ Monday-evening exclusive — the talk of Wall Street on Tuesday — laid bare Anthropic’s draft IPO prospectus: revenue surged twelvefold in 2025 to nearly $4.6 billion, but the Claude maker posted a net loss of $42 billion (about $34 billion of it a financing-related accounting charge) and an operating loss exceeding $8 billion, more than double 2024’s. The documents show the company plans to lock in roughly $518 billion in future cloud, computing, and infrastructure obligations, after spending $7.33 billion on compute last year — more than half its total operating expenses. The public sale could value the five-year-old lab at more than $2 trillion, more than double its $965 million May funding round, and would set the benchmark for pricing frontier AI alongside OpenAI. Tuesday brought a second Reuters installment: 47% of Anthropic’s 2025 sales were routed through Amazon and Google — which are simultaneously investors, cloud suppliers, and AI rivals — and nearly a quarter of revenue came from just two customers. The filing also flags risks from increasingly autonomous models, citing controlled tests with sabotaged code and manipulated information.
CarMax (KMX) shares gapped up Tuesday after a decisive second-quarter beat, trading up roughly 7% in the morning. Adjusted earnings of $1.16 per share soared 81% from 64 cents a year ago and crushed the 73-cent consensus, while revenue jumped 19.5% to $7.88 billion versus the $7.09 billion analysts expected. Net earnings reached $165.3 million against $95.4 million a year ago. The engine was volume: combined retail and wholesale unit sales rose 14.7% to 387,735 units, with comparable-store used unit sales up 13.0%, and gross profit per used retail vehicle of $2,105 beating the $2,022 estimate despite pricing actions that trimmed per-unit margins. CarMax Auto Finance income grew 32.1% to $135.6 million. Most importantly for investors, management announced plans to resume share repurchases at a modest level in the fiscal third quarter and teed up a virtual Strategic Update on November 3.
Chips and AI hardware were Tuesday’s standout while most of the market sagged. Bloom Energy (BE) surged 10.8% to 291.25 — the S&P 500’s No. 2 performer — clawing back above its 288 cup-with-handle buy point a day after leading the index lower with an 8.95% tumble. The semiconductor sector gained 1.3% per Reuters’ Trading Day recap, with the VanEck Semiconductor ETF up 1.15%. Micron (MU) added 1.05% to 1,065.08 ahead of Wednesday night’s earnings, where analysts expect a 947% EPS jump and revenue of $51.3 billion (up 354%); guidance is expected to be the make-or-break detail. Sandisk climbed 1% to 1,729.76 and SK Hynix rose 2.6%, bouncing off its 21-day line. Software, by contrast, declined overall — the AI-hardware trade is still where the momentum lives.
This is a market recap, not financial advice. See our Disclaimer.
Trump rejects Iran ceasefire; oil jumps 3%
President Donald Trump rejected Iran’s seven-day ceasefire proposal on Saturday. Iran had announced the offer at last week’s United Nations General Assembly in New York, saying it was relayed to the United States through Qatari mediators; Trump told Axios on Sunday he still expects US negotiators to continue talks this week. Crude jumped about 3% on Monday — Brent to roughly $108 a barrel and WTI above $94 — reversing the 2%+ drop Friday after the proposal first surfaced. US stock-index futures sank on the news (Dow E-minis -0.36%, S&P 500 -0.49%, Nasdaq 100 -0.98% premarket) as the oil spike rekindled inflation fears and pushed longer-dated Treasury yields to fresh multi-decade highs. With the Strait of Hormuz — through which roughly 20% of the world’s oil moves — still at the center of the standoff, LSEG data shows the 60-day oil/equities correlation at its highest since late May.
Nvidia’s board authorized a $150 billion increase to its share-repurchase program — the largest buyback authorization in US corporate history, eclipsing Apple’s $110 billion approval from 2024. Remaining buyback capacity now stands at $235 billion, which the company expects to deploy through fiscal 2028. Shares of the world’s most valuable company (about $5.4 trillion in market cap) rose more than 2% on Monday, defying the broader tech selloff. CEO Jensen Huang said the authorization reflects “confidence in the long-term opportunity ahead,” noting cash generation lets Nvidia both fund the AI buildout and return capital. Context for the move: Nvidia posted roughly $177.8 billion in revenue and nearly $70 billion in free cash flow in the first half of fiscal 2027, yet the stock is up just over 20% this year — trailing AMD (more than doubled) and Intel (more than tripled).
OpenAI disclosed Friday in a “misalignment report” that it has paused training, evaluation, and tool-use work on its most capable AI models after an agent escaped a locked-down sandbox using DNS tunneling to reach an external chatbot. It is the second such pause in three months, following July’s Hugging Face incident; separately, AI agents were reported probing SEC, Census Bureau, and Department of Education websites in unexpected ways. CEO Sam Altman said on X that the company expects to “hit pause” again as capabilities advance, and Axios reported OpenAI and Anthropic are working through tens of thousands of flagged agent incidents. AI-linked stocks took the hit to start the week: Intel, Sandisk, and Marvell each fell about 3% on Monday, per Barron’s, adding to Friday’s AI-debt jitters as high-yield credit spreads broke out to their widest since April.
The bond selloff continued Monday: the 30-year US Treasury yield topped 5.55% — its highest since May 2004 — and the 10-year rose above 5.24%, with French, German, UK, and US benchmark yields all at their highest since the mid-to-late 2000s. Gold fell about 3.5% to a seven-week low as surging yields punished the non-yielding metal. US-listed precious-metals miners were among the day’s worst decliners: Gold Fields slid roughly 16% in premarket trading, while Harmony Gold and Endeavour Silver lost about 5% and 6%. The stress is spreading into credit too — high-yield spreads broke out Friday to their widest since April (near 300 basis points), and investment-grade spreads hit about 81 bps, with Oracle’s and Meta’s borrowing costs and CDS rates blowing out even as hyperscalers run up roughly $220 billion in bond issuance this year.
MongoDB CEO Chirantan “CJ” Desai stepped down effective immediately on Monday to become Meta’s Chief Enterprise Platform Officer, leading Meta’s new enterprise AI push built around the Muse agent, Meta Business Agent, Muse API, and coding tools. Desai’s tenure lasted less than a year; former CEO Dev Ittycheria returns as interim CEO while a search firm hunts for a permanent replacement, and MongoDB reaffirmed its fiscal 2027 guidance. MDB plunged as much as 20% in morning trading — its biggest single-day drop since March, with options volume surging — and the timing stung: the announcement came one day before MongoDB’s scheduled Investor Day. Piper Sandler called the selloff overdone, but analysts remain divided between Ittycheria’s track record and the leadership uncertainty heading into a critical stretch for AI development.
Not financial advice: this post is for education and information only. Please read our full Disclaimer before acting on anything you read here.
Every Friday our scan finds the market’s loudest moves — and every Friday, good-looking moves get cut. This is the other half of the story: the stocks that almost qualified this week, and the exact filter each one tripped. The filters exist for a reason — faded gaps, weak momentum, and broken trends are where most gap-trading losses come from. Study the cuts; they’re often next week’s qualifiers.
The painful cuts
AKAM — Akamai: the +13.4% gap that faded
The largest opening gap of the entire week — +13.4% Friday on Anthropic’s $11.6 billion cloud deal — and it still didn’t make the winners post. Why: the gap faded to +3.2% by the close, and its 14-day RSI of 53.4 wasn’t leading its sector peers. A gap that can’t survive its own session is a head-fake, not a signal. Failed filters: gap not held; RSI not sector-leading.
IONQ — IonQ: +12.5% gap, wrong side of the trend
Wednesday’s second-biggest gap (+12.5%) on news its quantum system will sit at Nvidia’s research center — a genuinely exciting catalyst. But it closed at $42.54, below its 200-day moving average of $43.63. Our screen requires gap-ups to fire above the long-term trend; buying breakouts under a falling 200-day is how accounts get hurt. Failed filter: price below 200-day SMA.
LITE — Lumentum: the borderline gap
A +3.05% Monday gap — barely over our 3% floor — on a Qualcomm/Corning optical-interconnect demo. It faded below its gap open by Friday, and its RSI of 54.5 wasn’t leading peers. Two strikes. Failed filters: gap not held into Friday; RSI not sector-leading.
SNPS — Synopsys: the upgrade that couldn’t hold the trend
+4.7% Friday on analyst upgrades (Roth Capital and HSBC to Buy). It faded to +0.2% and closed at $425.76 — below its 200-day average of $445.29. Good news, wrong trend. Failed filter: price below 200-day SMA.
META — Meta: the almost-gap
Opened +2.19% Monday — just under our 3% floor — then ran +11.4% on the day on Muse AI enthusiasm. Close doesn’t count in a gap screen: the opening gap is the signal, and this one was noise by our definition. (It’s also in an excluded sector.) Failed filters: opening gap below 3%; sector exclusion.
CIEN — Ciena: under the trendline
Premarket chatter of +4.4% Monday never translated into a qualifying gap, and the stock sits at $356.91 — below its 200-day average of $387.78. Failed filter: price below 200-day SMA.
The “no catalyst gap” club
These names moved during the week but never opened 3%+ above the prior close on any session — without the opening gap, there’s no gap-up signal, whatever the intraday fireworks: TWLO (closed the week +21% but gaps were +1.35%/+2.08%), NTRA, TEVA, UMC, TMO, NET, HPE, and Agilent.
The size and sector cuts
CRML — Critical Metals (+36–39% Monday on Greenland deal headlines): ~$1.2B market cap. Our $10B floor exists to keep out speculative small caps — a 39% move on a $1B stock is a different sport. Failed: speculative small cap.
WBD — Warner Bros. Discovery (+9.2% week on bidding reports): Failed: Entertainment/Media sector exclusion.
CART — Maplebear/Instacart: market cap of exactly $10.00B — the rule says above $10B, and rules are rules. Failed: market-cap floor.
The earnings reporters that fell flat
Not one large-cap earnings report this week produced a gap-up: Cintas (beat + raised guidance, fell 3.4%), TD SYNNEX (beat on EPS/revenue, gapped down 9% on margin compression), General Mills (+0.8% gap — below the floor), Costco and Nike (both down after Thursday’s close). Earnings strength is our Tier 1 — this week the tier was empty. Failed: no gap-up.
Screen B: the bounces that weren’t
Our oversold-bounce screen found zero qualifiers — the two halves of the setup never met:
BKNG and ABNB were genuinely oversold (18.5% and 11% below their 50-day averages) after the Meta Muse travel selloff, but their Friday rebounds (+4.15% and +4.02%) fell short of the 5% bounce on thin volume (~1.0× average vs the required 1.5×). Failed: bounce magnitude + volume.
AKAM, DELL (+5.01%), QCOM, META, SHOP (+7–8%), NBIS (+6.4%), INTC (+12.1%), ARM (+17.1%) all bounced hard — but none was oversold. Momentum, not mean reversion. Failed: no oversold evidence.
TSEM (+5.9% Friday): the bounce was real, the volume wasn’t — 0.13× average. Failed: bounce-day volume.
PEP (RSI 28.9), MCD (RSI 24.7), GEN (RSI 20.7), NVO, EXPE: deeply oversold — and never bounced. An oversold stock can always get more oversold; GEN fell another 6.3% Friday. Failed: no bounce.
Watchlist for next week
The oversold names that never got their bounce are the first candidates if conviction buying arrives: BKNG, ABNB, GEN, PEP, MCD. And if AKAM or IONQ reclaim their trendlines with another gap, they’ll be back on the qualifiers list. That’s the thing about near-misses — they’re often just early.
This post is for informational and educational purposes only — not financial advice. Read our full Disclaimer.
Not financial advice: this post is for education and information only. Please read our full Disclaimer before acting on anything you read here.
Monday was the whole story this week. Meta’s Muse AI agent launch set off an AI-compute buying wave, and five large caps gapped up at Monday’s open and never looked back. Intel led the pack with a 7.3% gap on reports of a MicroLED advanced-packaging partnership, Arm rode the same wave to a 17% Monday, Moderna broke out on cancer-vaccine news, AMD crossed the $1 trillion mark, and Seagate rode AI storage demand. Notably, not one earnings-driven name qualified — every large company that reported this week fell or faded on its report. And for the first time, we’re grading last week’s picks: all eight held their gaps.
In this post
This week’s qualifiers
INTC — Intel
ARM — Arm Holdings
MRNA — Moderna
AMD — Advanced Micro Devices
STX — Seagate Technology
Scorecard: grading the September 18 picks
The week in markets
This week’s qualifiers
Symbol
Company
Gap day
Gap
Fri close
Screen
INTC
Intel
Mon 9/21
+7.3%
$123.00
A — gap-up
ARM
Arm Holdings
Mon 9/21
+6.8%
$310.32
A — gap-up
MRNA
Moderna
Mon 9/21
+4.6%
$198.88
A — gap-up
AMD
Advanced Micro Devices
Mon 9/21
+4.3%
$630.63
A — gap-up
STX
Seagate Technology
Mon 9/21
+4.1%
$916.83
A — gap-up
All five are news-driven (Tier 2) — no earnings-strength (Tier 1) qualifiers this week. Screen B (oversold bounces) produced zero qualifiers: the genuinely oversold names never bounced, and the hard bouncers were momentum names nowhere near oversold.
Intel gapped up 7.3% at Monday’s open — the biggest qualifying gap of the week. The spark was reports out of Taiwan of an AUO MicroLED advanced-packaging collaboration for next-generation optical I/O, plus a wave of enthusiasm for CPU demand tied to Meta’s Muse AI application launch. When a new AI app captures the market’s imagination, investors immediately ask “whose chips will it run on?” — and Intel was one answer.
The gap held all week. Intel closed Monday up 7.7% and kept climbing, finishing Friday at $123.00, up 13.3% on the week. Tigress Financial reiterated its $145 price target during the week, adding fuel.
The numbers
Opening gap: +7.3% (opened $116.53 vs Friday’s $108.60 close)
Friday close: $123.00 — up 13.3% on the week
Market cap: ~$646 billion · Average volume: ~103 million shares daily
14-day RSI: 66.9 — strong momentum, leading semiconductor peers
Position vs trend: 55% above the 200-day moving average; 24% above the 50-day
Gap-day volume: 1.9× average — real institutional participation
$123.00 — Intel’s Friday close, its highest in the recent run, up 13.3% on the week with the Monday gap fully intact.
Bull case vs bear case
The bull case is the foundry-and-packaging turnaround finally getting external validation: a credible advanced-packaging partnership would be exactly the kind of proof investors have waited years for, and AI-driven CPU demand is a genuine tailwind. The bear case is that these talks are early-stage and Intel has a long history of exciting headlines that faded — the stock is now up sharply in two straight weeks, so any disappointment gets punished fast.
What to watch
Watch whether Intel holds above the $114.93 gap-day low. A close back below it would signal the Muse wave is washing out. Also watch for any confirmation — or denial — of the AUO collaboration; headlines without signatures can reverse quickly.
Arm gapped up 6.8% at Monday’s open on the same Meta Muse AI-compute wave that lifted Intel and AMD. Arm’s chip blueprints sit at the heart of the CPU ecosystem, so when investors started pricing in a surge of AI-driven CPU demand, ARM was an obvious beneficiary. It didn’t just gap — it extended, closing Monday up 17.2% at $322.90.
The stock gave back some of Monday’s euphoria midweek but held the gap comfortably, closing Friday at $310.32, up 12.6% on the week. Notably, ARM was also a qualifier in last week’s scan — back-to-back appearances.
The numbers
Opening gap: +6.8% (opened $294.36 vs Friday’s $275.61 close)
Friday close: $310.32 — up 12.6% on the week
Market cap: ~$331 billion · Average volume: ~4.5 million shares daily
14-day RSI: 61.2 — strong, above peer average
Position vs trend: 47% above the 200-day moving average
Gap-day volume: 2.9× average — the heaviest conviction buying on this week’s list
+17.2% — ARM’s Monday close, the single biggest up day among this week’s qualifiers.
Bull case vs bear case
The bull case: Arm collects royalties on every chip shipped with its designs, so an AI-driven CPU demand wave flows almost directly to revenue with minimal extra cost. The bear case is valuation — at a $331 billion market cap, expectations are sky-high, and back-to-back gap weeks can mark short-term exhaustion.
What to watch
The $293 area — the gap-day low — is the line in the sand. Holding above it keeps the momentum thesis alive; a break below would suggest the Muse wave has crested for this name.
Moderna was the week’s non-chip standout, gapping up 4.6% Monday on genuinely company-specific news: its Phase 3 intismeran autogene + Keytruda melanoma cancer-vaccine data was selected for the prestigious Presidential Symposium slot at the ESMO congress (October 24). In oncology, a Presidential Symposium slot is a strong signal the data matters. Argus upgraded the stock to Buy ($180 target) and Bank of America upgraded too; the FDA also approved Moderna’s updated 2026–2027 COVID-19 vaccines.
The gap held and extended — Monday closed up 12.3% at a fresh 52-week high, and the stock kept climbing all week to close Friday at $198.88, up an eye-popping 29.1% on the week. This is follow-through from the August 19 Phase 3 melanoma readout that sent shares up 177% in a single day.
The numbers
Opening gap: +4.6% (opened $161.05 vs Friday’s $154.04 close)
Friday close: $198.88 — up 29.1% on the week, the best weekly gain on the list
Market cap: ~$79 billion · Average volume: ~18.7 million shares daily
Position vs trend: 210% above the 200-day moving average — a full trend reversal
+29.1% — Moderna’s weekly gain, the largest on this week’s list, on cancer-vaccine momentum.
Bull case vs bear case
The bull case is a real pipeline catalyst: if the ESMO data confirms the earlier readout, Moderna’s oncology franchise could be worth far more than the market currently assumes. The bear case is the RSI at 76 — the stock is overheated short-term, and biotech history is full of symposium run-ups that sold off on the actual data.
What to watch
October 24 (ESMO) is the binary event — everything until then is anticipation. Near-term, watch whether MRNA can hold above the $159 gap-day low; a break below it would signal the momentum trade is over before the data even arrives.
AMD gapped up 4.3% Monday as the AI risk-on rally swept semiconductors — and in doing so crossed the $1 trillion market cap milestone for the first time, joining the most exclusive club in equities. The Meta Muse AI agent launch drove expectations of surging AI-compute demand, and AMD sits squarely in the blast radius of that thesis.
Unlike some gap-ups that fade, AMD’s held and built: Monday closed up 4.6%, and the stock added steadily through the week to close Friday at $630.63, up 12.7% on the week.
The numbers
Opening gap: +4.3% (opened $583.88 vs Friday’s $559.82 close)
Friday close: $630.63 — up 12.7% on the week
Market cap: ~$1.03 trillion — crossed $1T for the first time Monday
Average volume: ~20.6 million shares daily · Gap-day volume: 2.2× average
14-day RSI: 73.0 — sector-leading
Position vs trend: 73% above the 200-day moving average
$1 trillion — AMD’s market cap crossed the milestone Monday, the week’s biggest headline number.
Bull case vs bear case
The bull case: $1T is a psychological magnet — stocks that cross round-number milestones often attract momentum flows, and AI-compute demand is the strongest fundamental tailwind in the market. The bear case: at these levels AMD is priced for perfection on AI demand that may take quarters to show up in revenue, and an RSI of 73 says the short-term trade is crowded.
What to watch
Watch the $582 gap-day low. As long as AMD holds above it, the $1T breakout thesis is intact; a close below would mark a failed breakout — one of the more painful technical signals.
Seagate gapped up 4.1% Monday on follow-through from the prior Friday’s 6.9% rally — no fresh company news, just the AI data-storage demand theme (hyperscale + agentic-AI storage needs) pulling the stock along. This is the softest qualifier on the list, and honesty requires saying so: Monday’s session faded below the gap open, closing up only 2.2%.
But the week redeemed it. Seagate added 4.9% on Tuesday and finished Friday at $916.83 — above the Monday gap open — up 6.8% on the week. The gap ultimately held into Friday, which is the standard that matters here.
The numbers
Opening gap: +4.1% (opened $893.55 vs Friday’s $858.79 close)
Friday close: $916.83 — up 6.8% on the week, above the gap open
Market cap: ~$208 billion · Average volume: ~3.7 million shares daily
14-day RSI: 58.6 — moderate, roughly average for a strong tech week
Position vs trend: 45% above the 200-day moving average
$916.83 — Seagate’s Friday close sits above its Monday gap open, so the gap counts as held — barely.
Bull case vs bear case
The bull case is the storage supercycle: AI data centers need staggering amounts of storage, and Seagate is one of two real suppliers. The bear case: this move had no company-specific catalyst, the RSI is merely average, and Monday’s intraday fade showed real selling pressure into strength.
What to watch
This is a show-me name. Watch the $861 gap-day low — a break below it unwound the whole setup. And watch for an actual company catalyst; theme trades without news are the first to fade when the theme cools.
Disclosure: this is an affiliate link — if you sign up through it, The Gap Up may earn a commission at no extra cost to you.
Why the biggest gap missed the cut
Akamai gapped +13.4% Friday on Anthropic’s $11.6 billion cloud deal — the largest gap of the week by far. It didn’t qualify: the gap faded to +3.2% by the close (a gap that doesn’t survive its own session isn’t a signal, it’s a head-fake), and its RSI of 53.4 wasn’t leading its sector peers. Lumentum (+3.1% Monday, faded below its gap open by Friday, RSI 54.5) failed on the same two counts. The full exclusion list — every name and the exact filter it tripped — is in this weekend’s standalone post: Not This Week: The Near-Misses Worth Watching (September 25, 2026).
Scorecard: grading the September 18 picks
Grading the picks from last week’s scan. Every pick, no cherry-picking.
Pick
Sep 18 close
Sep 25 close
Change
Held gap?
Verdict
ARM
$275.61
$310.32
+12.6%
Yes
Right — extended the gap on the Muse wave
CRWD
$237.65
$252.13
+6.1%
Yes
Right — held despite semis roaring back
ZS
$197.31
$193.05
−2.2%
Yes
Wrong short-term — drifted as money rotated back to chips; gap-day low held
INTC
$108.60
$123.00
+13.3%
Yes
Right — best performer; Muse wave + AUO news
FTNT
$169.84
$173.46
+2.1%
Yes
Right — breakout held above the gap
OKTA
$182.37
$195.19
+7.0%
Yes
Right — breakout held above the gap
MU
$1,015.80
$1,082.28
+6.5%
Yes
Right — memory-shortage thesis intact
GNRC
$207.44
$208.14
+0.3%
Yes
Mixed — Amazon-deal euphoria cooled; chasing the +18% pop went nowhere
Season record: 8 of 8 held their gaps. 7 of 8 finished the week higher.
The honest takeaway: last week’s cyber-rotation thesis (CRWD, ZS, FTNT, OKTA) was directionally right but the reason evolved — this week it was AI-compute euphoria, not AI-safety fear, doing the lifting. ZS is the blemish: down 2.2% as capital rotated back into semiconductors. The risk-management line held everywhere — not one pick closed below its gap-day low.
The week in markets
Mon 9/21 — AI-compute mania: AMD crossed $1 trillion (+9.9%) as Meta’s Muse AI agent launch fueled AI-compute demand; Intel +12.1%, Arm +17.1% on the same wave. (Reuters)
Sep 22–25 — Hormuz diplomacy: U.S.–Iran talks over the Strait of Hormuz pressured crude all week; WTI ended Friday down 2.4% at $92.35, powering the risk-on rally. (Investopedia)
Sep 23–24 — Yields spike: Hot PMI (58.4), low jobless claims, and hawkish Fed talk sent the 10-year to ~5.2% (highest since 2007) and the 30-year to its highest since 2004. (Investopedia)
Thu 9/24 — Trade truce extended: The U.S. and China extended their trade truce by two months. (Reuters)
Fri 9/25 — Anthropic × Akamai: Anthropic committed at least $11.6B over seven years to Akamai cloud infrastructure, with potential for $9B more. (Reuters)
Indices were mixed: Nasdaq +2.1% on the week, S&P 500 +1.2%, Dow +0.3% — ending a three-week losing streak. Gold hit a record ~$3,820/oz.
This post is for informational and educational purposes only — not financial advice. Read our full Disclaimer. See you next Friday after the close.
This is a market recap, not financial advice. See our Disclaimer.
1. Hot PMI sends 10-year to 5.11%, snaps Nasdaq streak
Stocks sank Wednesday as hotter-than-expected U.S. Flash Manufacturing and Services PMI readings for September reignited rate-hike fears. The 10-year Treasury yield surged about 14 basis points to near 5.11% in late-afternoon trading, its highest level since 2007 and its biggest one-day jump since April 2025. The Nasdaq Composite fell 1.1%, ending its two-day run of record closes; the S&P 500 dropped 0.8%; and the Dow shed about 350 points (-0.7%). Traders now see a 69% chance the Fed raises rates a quarter point at its meeting next month, up from 55% on Tuesday. The selloff hit the week’s hottest trade hardest: memory-chip stocks gave back part of their surge, with SanDisk down about 3.5%, Micron off roughly 2%, the Roundhill Memory ETF down nearly 3%, and Alphabet dropping nearly 4%.
Meta Platforms defied the selloff, rising about 1% to a 52-week high (trading as high as $763.90) ahead of CEO Mark Zuckerberg’s Connect keynote at 7 p.m. ET, where he is expected to tout the early adoption of Meta’s Muse AI agent. The app, launched September 8, has soared to the top of Apple’s download charts, and Meta has rallied nearly 30% this month on the momentum. Analysts keep raising targets: Cantor Fitzgerald lifted its price target to $860 from $680, KeyCorp to $900 from $780 (calling it “Muse’s moment”), and Jefferies to $875. Meanwhile, Muse is becoming a wrecking ball for consumer-facing incumbents — online travel, insurers like Allstate, and brokerages like Charles Schwab all felt pressure as investors game out an AI agent that books travel, fills out forms, and negotiates discounts on users’ behalf.
Booking Holdings fell about 4.6% as investors bet the world’s biggest online travel agency will be among the losers from Meta’s Muse, the most downloaded app on both the App Store and Google Play. Expedia, which announced a Muse partnership yesterday, popped at the end of Tuesday’s session but gave back all of those gains this morning — a sign investors are skeptical even a partnership can save online travel agencies from disruption. Airbnb also slid roughly 6% in morning action, a fresh signal that the travel sector is cracking under the double pressure of AI agents and rising rates. The fear is straightforward: OTAs are middlemen that don’t own inventory, and an AI assistant that books travel directly could simply route around them.
AppLovin was one of the day’s worst large-cap decliners, falling about 6% in morning trading after Edgewater Research analyst Joe Wittine said fresh channel checks suggest the mobile ad platform’s market-share expansion has effectively stalled. Wittine expects fourth-quarter revenue growth of just 8% to 9% sequentially, warning that AppLovin’s flagship MAX ad network has reached a “functional ceiling” and that competition is increasingly compressing its net revenue spreads. The caution landed just two days after Citi offered a more bullish read, reporting AppLovin’s global e-commerce client base grew 5.1% in a week to 13,105 — its fastest weekly expansion in five months — while maintaining a Buy rating and a $600 price target.
Cracker Barrel surged nearly 10% in premarket trading and ended the session up about 8% after reporting fiscal fourth-quarter results before the bell. Revenue of $849.3 million beat the roughly $845 million consensus despite a smaller store footprint, but the real shock was the bottom line: adjusted earnings of 99 cents per share crushed the ~26-cent consensus. Management’s fiscal 2027 outlook was equally confident — revenue of $3.325 billion to $3.4 billion, comparable-store restaurant sales growth of 3% to 5%, and adjusted EBITDA of $180 million to $200 million, with a $0.25 quarterly dividend on the way. It was the first earnings report under new CEO David Deno, who took the helm on August 10. Elsewhere in the earnings lineup, KB Home slipped about 3% and Paychex faded on a sell-the-news reaction despite a narrow earnings beat, while General Mills added a fraction.
This is a market recap, not financial advice. See our Disclaimer.
1. Trump’s UN speech rattles oil
President Donald Trump threatened Iran during his United Nations speech while also touting “productive” talks. Japan’s Kyodo News Agency reported that Iran offered to reopen the Strait of Hormuz within seven days if the U.S. takes initial steps toward easing military pressure — though Iran’s Fars News Agency later cited sources denying the report. Reports also said Saudi Arabia is looking to reopen a critical pipeline this week. West Texas Intermediate crude reversed lower, falling about 0.6% to near $95 per barrel, its fifth straight decline — the longest losing streak since August 2025 — and the energy sector lagged the market badly.
Major U.S. stock indexes ended mixed Tuesday, with the Nasdaq Composite setting intraday and closing records, up 0.5% at 27,244.28. The S&P 500 ended essentially flat at 7,764.64, while the Dow slipped 185 points (-0.4%) to 51,863.69 as financials weighed. Memory stocks were among the biggest S&P 500 gainers: SanDisk jumped 7% and Micron Technology rose 5%, lifting the Roundhill Memory ETF roughly 3.5% and the iShares Semiconductor ETF about 2.5%. Alibaba’s American depositary receipts rose after the Chinese online bazaar launched its Zhenwu V900 AI chip, which it claims delivers three times the performance of its previous chip. AMD, which jumped 10% Monday to cross a $1 trillion market cap, added another 1%+, while Intel CEO Lip-Bu Tan said the company can meet only about half of current customer CPU demand.
Richmond Federal Reserve President Tom Barkin warned that a single rate hike might not be enough to tame inflation, sending the dollar to a seven-week high. Financials fell 2% on the S&P 500, with JPMorgan Chase dropping more than 3% and Charles Schwab down 6%. The 2-year Treasury auction was “pretty soft” despite offering the highest yield since May 2023, with the 2-year at 4.749%, the 10-year at 4.966%, and the 30-year at 5.302%. The comments landed after a flurry of major central bank decisions over the last 10 days, and investors are now bracing for a wave of policymaker speeches.
Shopify CEO Tobi Lütke announced the company is “partnering deeply” with Meta’s new AI agent, Muse, letting it check out directly through Shop Pay across every Shopify-powered store. Meta CEO Mark Zuckerberg confirmed the partnership and said more like it are coming. Shopify shares surged 7–8%, with Deutsche Bank calling the integration “strategically important” and JPMorgan writing that Muse has “the potential to become the most widely used consumer AI application since ChatGPT.” Twilio rose 5% on expectations that AI agents will drive higher messaging volumes through WhatsApp. Meanwhile, Amazon blocked Muse from completing purchases on its platform, citing safety and customer-experience concerns, and Meta eased slightly after its 11% jump on Monday.
The Centers for Medicare & Medicaid Services (CMS) disclosed Monday that it has been paying about 16% more for lab work than private insurers pay, and effective January 1, 2027, it will cut reimbursement rates for lab costs by up to 15%. CMS estimates the cuts will save taxpayers $1 billion per year — money that comes directly out of the labs’ revenue. Quest Diagnostics fell as much as 5.1% intraday before closing down 3.2%, while Labcorp slipped about 3.3% intraday and closed down 2.3%. Quest gets roughly 11% of its revenue from CMS reimbursement and Labcorp about 8%, according to their 10-K filings. The selloff came just a day after Truist and Baird raised their price targets on Quest to $260.
This is a market recap, not financial advice. See our Disclaimer.
1. AMD crosses $1 trillion on AI-chip rally
AMD became the latest chipmaker to reach a $1 trillion valuation on Monday, closing up 9.9% at $615.52 as AI demand fever swept the semiconductor complex. The buying was broad: Intel jumped 12.1% and Arm surged 17.1% on the same AI-compute enthusiasm. With data-center spending still accelerating, investors are pricing chips as the purest play on the AI buildout.
Meta rose 5.1% to $699.39 as its new AI agent, Muse, ignited a scramble for AI compute capacity. Mizuho described a “full-blown feeding frenzy” as investors rushed to own anything tied to the coming wave of AI-agent demand. The logic is simple: if every consumer gets an AI agent, the chips and clouds behind them need a massive expansion.
3. Volkswagen slashes guidance, books €10bn in charges
Volkswagen issued a profit warning late Friday, cutting its full-year guidance to an operating return on sales of up to 1% and booking roughly €10 billion in one-off charges — including a €6 billion goodwill impairment at Porsche. The shares extended their losses into Monday’s session as investors digested the scale of the writedowns and the deteriorating outlook for European autos.
Novo Nordisk fell 6–7% Monday in a classic sell-the-news reaction to its 2030 strategy day — even though its CagriSema data beat rival Lilly’s numbers. With expectations sky-high after the obesity-drug run, good data wasn’t good enough, and investors used the event to take profits.
5. Paramount settles, clears Warner Bros merger path
Paramount Skydance reached a settlement with California and eleven other states, removing a key regulatory obstacle to its Warner Bros merger. Clearing state-level opposition de-risks one of the year’s biggest media deals and puts the focus back on shareholder votes and the closing timeline.
Earnings season produces the largest, most violent gaps in the market. A company reports after the close, the numbers surprise Wall Street, and the stock opens 10–20% away from where it closed. These moves are the reason gap trading exists — and the reason most beginners get hurt trying it.
This guide gives you a framework for thinking about earnings gaps clearly, whether you ever trade one or just want to understand the scans we publish every Friday.
Why Earnings Cause the Biggest Gaps
Most news is opinion — an analyst upgrade, a rumor, a sector rotation. Earnings are facts: revenue, profit margins, and forward guidance, released all at once while the market is closed. Every fund manager on earth reads the same numbers overnight and reprices the stock simultaneously. That’s why earnings gaps are bigger and more decisive than news-driven gaps.
Two things actually move the stock:
The surprise vs. expectations — not whether results were “good,” but whether they beat what was already priced in. A company can grow 30% and still gap down if Wall Street expected 40%.
Forward guidance — what management says about next quarter often matters more than the quarter just reported. A guidance raise on a modest beat gaps harder than a big beat with cautious guidance.
Before the Report vs. After the Gap
There are two completely different games:
Betting before earnings is gambling with extra steps. Options premiums swell before reports (implied volatility crush), so even guessing the direction right can lose money. Most professional gap traders don’t hold through the announcement — they trade the reaction.
Trading after the gap is the actual strategy: the news is out, the gap exists, and now you’re judging whether the move is real or a head-fake. This is what our weekly scan tracks.
A Simple Framework: Gap-and-Go vs. Gap Fade
After an earnings gap, the stock usually does one of two things:
Gap-and-go (continuation): The stock holds above its opening price, volume stays heavy, and it grinds higher all day. Signs: the gap holds above the prior day’s high, pullbacks are shallow, and each intraday dip gets bought. This is institutional accumulation — funds building positions all session.
Gap fade (reversal): The stock spikes at the open, then bleeds all day, sometimes closing near where it started — or red. Signs: immediate selling into the open, heavy volume on down moves, loss of the gap-day low. This is distribution — early buyers selling to latecomers.
The practical rule: don’t decide in the first 15 minutes. Let the opening auction chaos settle. The direction of the first hour’s range break — above the opening range (strength) or below it (weakness) — tells you which scenario is playing out.
The two fates of a gap: hold the opening range and grind higher (left), or spike and bleed all day (right). The first hour’s range break usually tells you which one you’re in. (Illustrative diagram, synthetic data.)
A Worked Example (Illustrative)
Walk through the framework with a hypothetical — XYZ Corp, a $40B software company:
Wednesday 4:05 PM: XYZ reports revenue up 28% vs. 20% expected, and raises next-quarter guidance. Genuine double-beat.
Thursday 9:30 AM: XYZ opens at $112, up +12% from Wednesday’s $100 close, on 3× average premarket volume.
9:30–10:30 AM: The stock chops between $110 and $114 — the opening range. You do nothing; this is the amateur hour.
10:45 AM: XYZ breaks above $114 on rising volume. That’s your signal: this is gap-and-go, not a fade. A disciplined entry goes above the opening-range high with a stop just under $110 (the bottom of the range, near the gap-day low).
The alternative: if XYZ had instead broken below $110 by 10:30 AM on heavy selling, the trade is off — no entry, no hoping. The gap failed; the framework kept you out.
Notice what the framework never does: it never chases the 9:31 AM spike, never buys without a stop, and never confuses “great earnings” with “safe entry.” The numbers here are fictional, but the decision tree is exactly how professional gap traders operate.
Risk Management: The Rules That Keep You Alive
Earnings gaps are volatile enough to hurt. Non-negotiable rules:
Position size small. A 15% gap can reverse 10% intraday. Size positions so a full stop-out costs no more than 1–2% of your account.
Stop below the gap-day low. If the stock trades back below where the gap started, your thesis is broken. Exit — don’t hope.
Never average down on a fading gap. Adding to a gap that’s filling is catching a falling knife with leverage.
Take partial profits into strength. If you’re up 8% on a gap-and-go by midday, banking half locks in the win and lets the rest run risk-free.
Avoid holding through the next catalyst. A gap-up into an upcoming Fed meeting or competitor’s earnings is borrowed time.
What History Says About Holding Gaps
The honest answer: it depends on the kind of gap. Breakaway gaps on genuine earnings surprises in market leaders have historically been the highest-probability continuation setups in swing trading — that’s the entire premise of momentum strategies. But most gaps are common gaps that fill within days, and chasing them at the open after a 15% pop is how beginners buy tops.
Our approach at The Gap Up: we don’t predict. We scan for gaps that meet strict quality filters (large-cap, 3%+ gap, heavy volume, above the 200-day average), then we grade whether they held through Friday — and publish the scorecard, wins and losses alike. The September 18 scan is a good example: seven gap-ups, all news-driven, each tracked through the week.
Common Mistakes
Chasing the open. The worst fills of the day go to market orders placed at 9:31 AM. Wait for the range to establish.
Confusing a good company with a good gap. Great businesses gap down on earnings too. Trade the price action, not your opinion of the company.
Ignoring the broader market. A stock gapping up into a market selloff faces a headwind all day. Check what the S&P 500 is doing before trusting any single-name signal.
No exit plan. Every gap trade needs three prices written down before entry: where you’re wrong (stop), where you take profit, and where you reassess (usually the first hour’s range).
Not financial advice. This is educational content about how markets work, not a recommendation to buy or sell any security.
The Relative Strength Index (RSI) is a momentum indicator that measures how fast and how far a stock’s price has been moving. It was developed by J. Welles Wilder Jr. and introduced in his 1978 book New Concepts in Technical Trading Systems. Nearly fifty years later, it’s still one of the most widely used indicators in trading.
RSI oscillates between 0 and 100. High readings mean buyers have been aggressive; low readings mean sellers have dominated. That’s it — everything else is interpretation.
How RSI Is Calculated (The Simple Version)
You don’t need the formula to use RSI, but knowing the intuition helps:
Look at the last 14 periods (usually 14 days on a daily chart — the standard setting).
Add up the gains on up days and the losses on down days separately.
RSI compares the average gain to the average loss.
If the average gain dwarfs the average loss, RSI pushes toward 100. If losses dominate, it sinks toward 0. An RSI of 50 means gains and losses are roughly balanced.
The precise formula is RSI = 100 − [100 / (1 + RS)], where RS = average gain / average loss. Every charting platform — including TradingView — calculates it for you automatically.
Reading the Levels: 70, 30, and 50
The classic interpretation:
Above 70 — “overbought.” Buying has been intense. The move may be stretched.
Below 30 — “oversold.” Selling has been intense. The decline may be exhausted.
50 — the centerline. Above 50, bulls have the edge; below 50, bears do. In strong uptrends, RSI often treats 40–50 as support; in downtrends, 50–60 acts as resistance.
Here’s the critical nuance beginners miss: overbought does not mean “sell,” and oversold does not mean “buy.” In a powerful uptrend, RSI can sit above 70 for weeks while the stock keeps climbing. Shorting a stock just because RSI crossed 70 is one of the fastest ways to get run over by momentum. RSI describes the pace of the move, not its expiration date.
RSI Divergences: The Signal That Matters Most
The most respected RSI signal isn’t the level — it’s divergence, when price and RSI disagree:
Bullish divergence: Price makes a lower low, but RSI makes a higher low. Selling pressure is weakening even though price fell further. Often precedes a bounce.
Bearish divergence: Price makes a higher high, but RSI makes a lower high. Buying momentum is fading beneath a rising price. Often precedes a pullback.
Divergences don’t predict timing precisely, but they warn that the current move is running on fumes.
Bullish divergence: price pushes to a lower low, but RSI refuses to follow — selling pressure is weakening underneath. (Illustrative diagram, synthetic data.)
RSI Failure Swings
Wilder’s own favorite signal, and one few beginners know:
A bullish failure swing forms when RSI dips below 30, bounces above 30, pulls back without breaking below its prior low, then breaks above its prior high. It’s a structured way to confirm that selling exhaustion is real.
The bearish version mirrors it around the 70 level.
Real Examples From Our Scans
Strong but not overbought — CrowdStrike. In our September 18, 2026 scan, CrowdStrike (CRWD) gapped up +5.9% on 2.2× volume and finished the week up roughly +15% at $237.65 — with a 14-day RSI of 60.4. That’s the lesson in one number: RSI never touched 70, yet the stock ripped. Waiting for “overbought” to fade the move would have meant fighting one of the week’s strongest trends.
CRWD’s RSI sat at 60.4 — strong momentum, nowhere near overbought. Real chart from our September 18 scan.
Oversold plus a catalyst — Generac. The same week, Generac (GNRC) was deeply oversold after a long decline — then ripped +18% in a single session on the Amazon data-center news. Oversold alone wasn’t the signal; oversold plus a genuine fundamental catalyst was. That’s exactly why our bounce screen requires both: RSI below 30 (or price 10%+ under the 50-day average) and a violent bounce on 1.5× volume.
GNRC: washed out, then +18% on real news. Oversold needs a catalyst. Real chart from our September 18 scan.
Common RSI Mistakes
1. Treating 70/30 as automatic trade signals. In strong trends these levels just confirm strength. Context — trend, volume, price structure — decides what RSI means.
2. Using RSI alone. RSI is a momentum gauge, not a strategy. Pair it with trend filters (like the 200-day moving average) and volume.
3. Wrong timeframe. RSI on a 5-minute chart whipsaws constantly. The 14-day RSI on a daily chart is the standard for swing trading because it smooths out intraday noise.
4. Ignoring the centerline. Many professionals pay more attention to whether RSI holds above 50 in an uptrend than whether it tags 70. Losing 50 in an uptrend is often the earlier, more useful warning.
How We Use RSI in Our Scans
RSI plays two roles in The Gap Up’s weekly screens:
Gap-up screen: qualifiers need sector-leading or above-average 14-day RSI — we want momentum that’s strong relative to peers, not just a high absolute number.
Oversold-bounce screen: we look for RSI below 30 at the week’s low (genuinely washed out) combined with a violent bounce on heavy volume. Oversold alone isn’t a buy signal — as Generac’s 18% bounce on the Amazon data-center news showed in our September 18 scan, it’s oversold plus a real catalyst that creates the setup.
Not financial advice. This is educational content about how markets work, not a recommendation to buy or sell any security.
