This cycle's dominant pattern is beat-then-sell-off, confirmed on three separate names. CRDO reported a confirmed EPS beat ($1.16 actual vs. $1.03 consensus) on September 1, and Yahoo Finance's own coverage — carried in this desk's news pull — reports the stock fell roughly 19% the next session "as investors worried about margin pressures and cautious future guidance"; the chart cracked hard as a result, RSI collapsing to 34.2 from 49, 30-day print to -24.07% from +31.16%, and vs-SMA200 flipping negative to -2.7% from +33.4% — a trend that was solidly intact two cycles ago is now genuinely under test. MDB reported fiscal Q2 revenue of $771.8 million (+30% year over year) and an EPS beat of $1.90 versus $1.61 consensus (+18.1% surprise) on September 1, with raised full-year guidance — and MarketBeat's coverage in the same pull reports the stock "fell sharply... despite beating revenue and earnings expectations"; RSI fell to 42.2 from 63, 7-day print to -17.44%, and the 200-day cushion compressed hard to +7.7% from +30.9%. PANW also reported a confirmed beat (revenue $3.41B vs. $3.35B consensus, EPS $1.02 vs. $0.98, +34.5% YoY revenue growth) on September 1, and its stock fell on the week too, though far less violently. SMCI's one-year deficit nearly closed entirely this cycle — from -15.67% to -2.73% — while its 30-day bounce cooled to +30.57% from +44.28%, the clearest structural improvement in the "strength on a negative year" cohort this scan tracks. ESTC's scheduled print, unconfirmed last cycle, is now confirmed as a beat (EPS $0.70 vs. $0.58 consensus, +19.85% surprise, reported August 27) — and the stock still cooled hard out of parabolic, RSI falling to 60.1 from 77.5, joining the beat-then-cool pattern rather than extending on the news. MSFT's one-year return flipped back negative — to -1.63% from +0.76% — reversing last cycle's flip-to-positive after its 30-day print cooled sharply to +2.51% from +31.49%. OKTA's reversal continued into a second confirming cycle, one-year extending to +90.1% from +79.53%. HUBS's near-flip of its 200-day reversed back down — vs-SMA200 widened to -5% from -1.3%, undoing last cycle's near-crossing.

Data as-of the 2026-09-04 close. Source: tech-insider-buys — price/technical fields only; this scan does not carry independent insider dollar-flow figures for this universe this cycle (see companion insider-scan for the overlapping names' Form-4 data). Prior scan: 2026-08-28-tech-insider-buys (2026-08-28 close).

CRDO's confirmed earnings beat did not stop the chart from breaking, and the break was severe. ($170.57, RSI 34.2, was 49; 7D -26.74%, 30D -24.07%, was +31.16%) — 2026-09-01-crdo-earnings confirms revenue of $479 million (a roughly $3.7 million beat) and EPS of $1.16 versus a $1.03 consensus, a genuine beat. But Yahoo Finance's own coverage, carried in the same pull, reports the stock fell approximately 19% the following session "as investors worried about margin pressures and cautious future guidance" — the only cause this pull offers, and it is sourced directly to the event record rather than inferred. The chart damage that followed is real: vs-SMA200 flipped to -2.7% from +33.4%, one-year return collapsed to +27.29% from +76.6%, and the regime downgraded from pullback to collapse. This is now the sharpest single-cycle deterioration tracked in this universe.

MDB reported a strong confirmed beat and sold off anyway. ($368.74, RSI 42.2, was 63; 7D -17.44%, 30D -2.02%, was +40.4%) — 2026-09-01-mdb-earnings confirms fiscal Q2 revenue of $771.8 million (+30% year over year, Atlas up ~29%, enterprise/other up ~36%) and EPS of $1.90 versus a $1.61 consensus (+18.09% surprise), with raised full-year guidance driven by Atlas and AI-workload demand. Despite that, this desk's news pull (fetched 2026-09-05, ok, not stale) carries a headline reading "MongoDB's stock fell sharply after strong fiscal Q2 results, despite beating revenue and earnings expectations." The 200-day cushion compressed hard to +7.7% from +30.9%, and one-year return cooled to +14.5% from +40.4% — a real, confirmed beat that the market sold anyway.

PANW also beat and pulled back, more mildly than CRDO or MDB. ($333.26, RSI 43.5, was — comparable cycle-over-cycle read not separately tracked in this universe's own file; 7D -10.36%, 30D -8.11%) — 2026-09-01-panw-earnings confirms revenue of $3.41 billion versus $3.35 billion consensus (+34.5% year-over-year growth) and EPS of $1.02 versus $0.98 (+4.35% surprise), a genuine beat, alongside an announced acquisition. The stock still fell on the week — a third confirmed-beat-sells-off case this cycle, though softer than CRDO's break or MDB's compression. Read together, three separate confirmed beats produced three separate sell-offs this cycle — a pattern worth naming directly rather than treating each as an isolated, unrelated move.

ESTC's scheduled print, unconfirmed last cycle, is now confirmed as a beat — and the stock cooled hard out of parabolic regardless. ($91.81, RSI 60.1, was 77.5; 30D +31.68%, was +55.04%; vs-SMA200 +43.8%, was +56.6%) — 2026-08-27-estc-earnings confirms EPS of $0.70 versus a $0.58 consensus (+19.85% surprise), reported August 27, resolving last cycle's overdue-unconfirmed flag cleanly. The regime downgraded from parabolic to uptrend and one-year cooled to +8.38% from +13.8% — a confirmed beat followed immediately by a cooldown, the same shape as CRDO, MDB, and PANW, just less severe.

SMCI's one-year deficit nearly closed entirely this cycle — the clearest structural improvement in the "bounce vs. broken trend" cohort this scan tracks. ($39.59, RSI 62.7, was 56.9; 30D +30.57%, was +44.28%, decelerated) — one-year change improved sharply to -2.73% from -15.67%, and vs-SMA200 widened to +26.2% from +18.3%. Last cycle this desk flagged SMCI's strength as sitting on top of a negative one-year return — the same read holds directionally, but the deficit itself has narrowed to nearly nothing, meaning a continuation next cycle could plausibly flip it positive for the first time in this tracked window.

MSFT's one-year return flipped back negative, reversing last cycle's flip to positive. ($499.70, RSI 59.4, was 73.1; 30D +2.51%, was +31.49%, cooled sharply) — one-year change is now -1.63%, down from +0.76% last cycle, as the trailing comparison window rolled forward past the strongest part of the prior run. Vs-SMA200 held at +15.9% (was +19.1%), still comfortably positive, so this reads as the one-year window's base effect catching up rather than a fresh breakdown — but it is a genuine reversal of last cycle's headline improvement and worth tracking directly rather than assuming the flip to positive was durable.

HUBS's near-flip of its 200-day reversed back down. ($247.60, RSI 54.3, was 61.7; 30D -1.04%, was +3.81%) — vs-SMA200 widened back to -5% from -1.3%, undoing the near-crossing flagged last cycle, while one-year stayed deeply negative at -46.67% (was -45.27%). This is the second straight cycle this name has moved toward, then away from, its own 200-day line without completing the flip either direction.

OKTA's reversal continued into a second confirming cycle, and ARM's deep pullback extended with the 30-day window reversing negative even as the longer trend strengthened. OKTA ($170.60, RSI 63.4, was 64.7 — essentially flat; 30D +16.02%, was +21.42%; one-year +90.1%, was +79.53%) held its reversal from two cycles ago. ARM ($252.09, RSI 48, was 40; 30D -8.19%, was +6.3%, flipped negative; one-year +86.07%, was +67.7%; vs-SMA200 +24.9%, was +19.9%) shows a mixed but net-positive picture — the short window reversed while the longer trend and cushion both extended, still a deep pullback inside a strengthening trend rather than a break.

VEEV held its positive one-year flip from last cycle, and TEAM continued cooling off its multi-week extension. VEEV ($275.09, RSI 65.9, was 74.9; one-year +1.86%, was +1.6% — held) stayed marginally positive on the year for a second straight cycle. TEAM ($189.58, RSI 70.3, was 79.1; 30D +67.3%, was +82.6%, decelerating as the initial spike rolls out of the window; one-year +12.79%, was +7.72%, improved; -4.54% off its 52-week high, was within 2.33%) continued the cooldown flagged last cycle, still without a fresh catalyst identified in this desk's corpus beyond the original August 6 beat.

MRVL's scheduled print, unconfirmed last cycle, is now confirmed as a narrow beat, and the chip-adjacent complex broadly cooled or held. 2026-08-27-mrvl-earnings now shows status: captured, EPS $0.94 versus $0.93 consensus (+1.0% surprise). MRVL ($223.55, RSI 51.1, was 47.2) has recovered since, one-year now +248.75%. INTC ($95.80, RSI 51.7; one-year +289.27%, vs-SMA200 +29.2%) and ASML ($1,714.88, RSI 48.3; one-year +127.61%, vs-SMA200 +16.1%) both stayed roughly steady. AVGO ($357.90, RSI 38.1; one-year +16.92%, vs-SMA200 -3%) crossed below its 200-day this cycle — full detail in the companion bargain-bin scan.

IONQ improved modestly while staying broken, and INTU's broken-thesis flag survived a tenth cycle unmoved. IONQ ($39.52, RSI 46.4, was 44.5; one-year -6.15%, was -9.5%; vs-SMA200 -10.6%, was -11.9%) narrowed its deficit slightly, still below its 200-day. INTU ($332.70, RSI 46.6, was 60; one-year -50.22%, was -46.4%, worsened) carries the same thesis: broken flag across ten straight cycles now — full detail in the companion bargain-bin scan; this override does not clear on any bounce or any print.

The table labels are technical states, not trade recommendations. This cycle's real lesson generalizes the desk's standing rule that no earnings beat clears a broken thesis (INTU, ten cycles running): a beat does not automatically protect an otherwise-healthy chart either. CRDO, MDB, and PANW all reported confirmed beats this cycle and all three sold off — CRDO's break below its 200-day, sourced directly to Yahoo's own reporting on margin and guidance concerns, is the sharpest single-cycle deterioration tracked in this universe, and it happened on a beat, not a miss. ESTC's resolution from overdue-unconfirmed to a confirmed beat followed the same shape — a real, sourced print, followed by a cooldown rather than extension. Read against that backdrop, SMCI's near-closed one-year deficit and MSFT's reversal back to negative are the two cleanest structural swings this cycle that had nothing to do with an earnings print — both are base-effect and momentum stories the desk can trace directly to the summary's trailing windows rather than to any single event. thesis: broken (INTU, tracked in the companion bargain-bin scan) remains a hard override that a confirmed earnings beat did not clear this cycle either.

Confirmed beats sell off anyway — three names, one cycle — CRDO (broke its 200-day on margin worries), MDB (sold off despite a strong beat and raised guidance), and PANW (mild pullback on a real beat) all confirm the same shape.

Confirmed print resolved, cooled hard regardless — ESTC's scheduled print, unconfirmed last cycle, is now a confirmed beat, and the stock cooled out of parabolic anyway rather than extending.

One-year returns swinging on base effects, not events — SMCI's deficit nearly closed and MSFT's flipped back negative, neither tied to a dated catalyst — both are the trailing-window math catching up with the underlying price action.

Deep pullback, trend intact or strengthening — ARM, AAPL, AMD, ASML, INTC, MRVL, ANET, CRWD, NET all cooled or reversed on the short window while remaining well above their 200-days with strongly positive one-year returns.

Near-flip, reversed — HUBS moved toward its 200-day, then back away from it, for a second straight cycle without completing the flip.

Broken thesis, unmoved by anything — INTU. Ten straight cycles, no bounce or print has cleared the flag, and the chart worsened further this cycle.

• Tape data (price/RSI/SMA/returns/1-year change/vs 200-day): full-scan pre-computed summaries as of the 2026-09-04 close — tech-insider-buys.
• CRDO earnings: 2026-09-01-crdo-earnings (confirmed beat, EPS $1.16 vs $1.03; Yahoo Finance coverage attributes the ~19% post-print drop to margin/guidance concerns).
• MDB earnings: 2026-09-01-mdb-earnings (confirmed beat, revenue $771.8M +30% YoY, EPS $1.90 vs $1.61); news pull latest (fetched 2026-09-05, ok, not stale) — "stock fell sharply... despite beating expectations."
• PANW earnings: 2026-09-01-panw-earnings (confirmed beat, revenue $3.41B vs $3.35B, EPS $1.02 vs $0.98).
• ESTC earnings: 2026-08-27-estc-earnings (confirmed beat, EPS $0.70 vs $0.58).
• MRVL earnings: 2026-08-27-mrvl-earnings (confirmed narrow beat, EPS $0.94 vs $0.93).