Last scan's two flagged "new rollover risk" names both broke down hard this cycle: BE's 30-day change deepened from -1.7% to -34.6%, and CRDO's fell from -2.6% to -25.4%. The read that both were genuine trend reversals, not just digestion, has been confirmed. The other side of the story: PANW, OKTA, and CRWD — flagged last scan as "cooling from overbought" — instead reaccelerated, all climbing back into the mid-60s RSI and extending further vs their 200-day, even as the broader semis/index tape cooled hard (see this cycle's ETF-universe scan). SMCI's tentative stabilization did not hold — it deepened instead. HUBS and INTU both continued bouncing in price while their explicit broken-thesis flags persisted — still avoids regardless of the chart. IONQ's feed refreshed to current and confirms real, ongoing weakness (RSI 25.7, 30D -38.5%).

Data as-of 2026-07-17 close. Source: summaries/tech-insider-buys.json — price/technical fields only, no live insider transaction data in this snapshot; cross-reference against the companion insider-scan report (the desk's own tooling) for actual insider dollars where the two universes overlap. Prior scan: 2026-07-11-tech-insider-buys (2026-07-10 close).

BE and CRDO are the scan's confirmed story this cycle. BE ($214.96, RSI 39.0) had already shown a rough week last scan (7D -18.9%); this cycle its 30-day change deepened from -1.7% to a genuine -34.6%, with its price down from $244.61 to $214.96 and its vs-SMA20 gap widening to -21.6%. Its thesis stays flagged intact and the golden cross holds, but the near-term trend has clearly broken — this lines up with the companion insider-scan cut, which shows BE's insider-selling figure (-$444.5M) essentially unchanged, meaning the price damage isn't being offset by any fresh buying. CRDO ($202.68, RSI 39.4) shows the same pattern even more sharply: 30-day change fell from -2.6% to -25.4%, 7-day -21.4%, after a huge multi-month run that's now compressed from +91.9% to +16.13% on the 3-month window. Both names confirm last scan's read that these were genuine reversals, not routine digestion.

The cyber/AI-infra cluster went the opposite direction from what was flagged last scan. PANW ($358.68, RSI 67.4, was 58.6) and OKTA ($149.35, RSI 64.8, was 62.2) both reaccelerated rather than continuing to cool, and CRWD ($203.08, RSI 62.9, was 57.2) did the same — all three now sit further above their 200-day than last scan (PANW's cushion widened from +59.2% to +72.1%). This is a notable divergence given that this cycle's ETF-universe scan shows the broad index/semis sleeve (QQQM, SMH, VGT) cooling hard over the same window — the security-software names are bucking that pullback entirely.

SMCI's tentative stabilization from last scan did not hold. RSI fell back from 43.3 to 34.7, and its 30-day change deepened from -11.5% to -21.14% — still death-crossed and in downtrend regime, now clearly weaker rather than basing. HUBS and INTU both continue the "don't be fooled by the chart" pattern: both carry thesis: broken flags that have now survived two consecutive scans of real price improvement — HUBS's 30-day gain accelerated from +11.3% to +27.4%, INTU's from -0.3% to +9.51%. Per desk doctrine, neither bounce clears the flag. IONQ's feed refreshed to current this cycle (was stale at 07-09 last scan) and confirms genuine, ongoing weakness: RSI 25.7, 7D -18.85%, 30D -38.5% — the deepest technical damage in the whole universe now that the data is live. CEG's mixed read from last scan resolved to the downside: its 30-day change flipped back negative (-7.91%, was +1.9%), still collapse regime and death-crossed.

The table labels are technical states, not trade recommendations. "Confirmed rollover" (BE, CRDO) means last scan's flagged 30-day trend reversal has now deepened into a genuine breakdown across a second consecutive cycle — a materially more cautious read than a single-cycle wobble. "Reaccelerating" (PANW, OKTA, CRWD) means RSI and vs-SMA200 extension both climbed further, the opposite of digestion — worth tracking whether this divergence from the broader tech pullback (this cycle's ETF-universe scan) persists or reverts. thesis: broken is a hard override that a price bounce does not clear (HUBS, INTU), regardless of how many consecutive scans the bounce continues. This watchlist references "insider buys," but the JSON summary contains price/technical fields only — cross-reference the companion insider-scan report for live insider dollar flow before treating any RSI print here as a buy signal.

Confirmed rollover (both flagged names broke down) — BE and CRDO both went from a single-cycle wobble to a genuine, multi-cycle trend break. BE's is corroborated by unchanged insider selling in the companion insider-scan cut; CRDO's is a pure momentum break after a huge run that's now largely unwound on the 3-month window.

Reaccelerating, not cooling — PANW, OKTA, CRWD all climbed further this cycle instead of continuing last scan's digestion pattern, and did so while the broad index/semis tape (this cycle's ETF-universe scan) cooled hard. A real divergence worth tracking.

Broken thesis, bounced again — HUBS and INTU. Both improved on price for a second consecutive cycle; both remain thesis: broken. Treat the improvement as noise until the flag lifts.

Stabilization reversed — SMCI's tentative improvement from last scan gave way to renewed weakness (RSI 43.3→34.7). IONQ's feed refreshed to current and confirms it's genuinely the weakest name in the universe now, not a stale print.

Extended strength — FTNT, DDOG, OKTA, AAPL, NET, SNOW all sit 22-99% above their 200-day. AAPL's move into RSI 71.6 is notable given how hard the rest of the broad tape cooled this cycle.

• Tape data (price/RSI/SMA/returns): full-scan pre-computed summaries as of the 2026-07-17 close — *.