Today's research notes

Do not double down on speculative unprofitable holdings with unfunded backlog and active dilution based on intraday VWAP holds, nor allow ambiguous profit trails to let large excursions round-trip.

FROG (+18.8% peak to .76 vs target) round-tripped to -1.11% raw / +0.16% net; selecting it over NTRA (+4.66%, 0.85R) for aggressive target EV was flawed.

CART (+12.4% gap, -8.7pp SPY trend) failed -6.63% raw, stopping at -6.23% net; picking it over WAT (+1.74%, 2.9% stop risk) for a 0.11pp EV edge was flawed.

May-June IOT/INCY/TGTX/NBIS/KPTI cases show event catalysts need confirmation, profit protection, and cooldown discipline after target-zone runs.

May 2026 ABNB/HALO/XYZ/HWM/AMAT failures show real post-earnings catalysts still need range acceptance, shelf/reclaim confirmation, and feasible stop risk.

June 2026 AI/software cases show event-asymmetry rechecks, completed-close validation, stop enforcement, and fast profit protection matter more than a fixed 7-day hold.

May 2026 stale post-earnings cases failed when old catalysts lacked reclaim, volume, sector alpha, or same-day participation confirmation.

Differentiate distant legal or regulatory milestones from immediate binary gates; treat mid-horizon trials as path risk modifying sizing rather than hard rejections.

Oil spikes and rising 10Y yields create risk-off tape favoring energy cyclicals, but require sector-relative volume confirmation over unconfirmed raises or multi-year projects.

Verify binding economics, delivery timelines, and revenue thresholds for AI hardware and datacenter infrastructure before treating partnership announcements or milestone reviews as revenue.

High-beta optical and semicap extensions driven by analyst chatter or unconfirmed rumors face distribution and wide stop risk unless backed by new issuer economics.

Normalize consumer earnings by stripping discrete tariff refunds, debt-repurchase gains, and share-count reductions from core comps and recurring operating margins.

Enterprise tech recovery setups require warmup-inclusive volume-complete event VWAP clearance, while debt refinancing and stale M&A recaps must not simulate operating catalysts.

Defense awards require separating undefinitized program ceilings from funded near-term revenues, while isolating restatement and preferred dilution effects.

Sep 10 final picks (ASO 11:55 ET .09, DELL 14:18 ET .95, GME 16:43 ET .39) and deterministic first-barrier exit plans.

Sep 10 alternate slates and exact numeric reconsider conditions across consumer, hardware, software, and mobility candidates.

LDOS was marginally correct only under its shelf-break exit; the later target hit and contract were non-executable.

Do not infer durable five-week momentum from a same-day VWAP hold after an extreme gap when guidance and earnings quality remain mixed.

UNH’s low-volume post-earnings reclaim failed quickly because nearby resistance and prior event-day rejection left weak path reward.

CLS reached its declared target before a post-close equity offering caused the later raw loss.

Negative catalyst reaction plus a 0.58R first path left FITB’s distant measured-move target carrying an EV that failed.

CGAU was correct; policy captured +15.40% net, while later raw gains were non-executable and gold beta explained much of the payoff.

AMLX choice was correct; pivotal-data momentum hit target next session and beat every finalist.

AVGO was the correct actionable pick; completed-bar execution and catalyst attribution reduce the headline raw return.

Cash beat the slate under planned stops even though PM finished slightly positive; endpoint rankings overstated PM's executable outcome.

Cash underperformed four of five finalists; BBIO and PODD advanced while preserving plausible structural stops.