Acadia Pharmaceuticals (ACAD) — Core Intelligence

Current read
“Acadia enters the RADIANT readout with a durable commercial base, a credible Alzheimer's disease psychosis mechanism and a trial that is more likely than not to succeed statistically. The harder question is magnitude. An observed effect near d≈0.45, coherent across doses and supported by responder analyses and global clinical measures, would validate the Clinaptis forecast; only a stronger result meaningfully improves our estimate of the effect Phase 3 can reproduce. We are constructive into the catalyst and favor long exposure, while recognizing that the stock has already moved above the reference price used in our original work.”

The maintained investment case

RADIANT is not testing the 5-HT2A mechanism from scratch. Pimavanserin established randomized human activity in Alzheimer's disease psychosis, but its disease-specific evidence remained marginal and analysis-sensitive.

Remlifanserin attempts to repair both the pharmacologic and experimental weaknesses through greater exposure headroom, a cleaner cardiac profile, biomarker-confirmed Alzheimer's disease, dose ranging and an architecture that feeds independent Phase 3 studies. Clinaptis centers on roughly three SAPS-H+D points of placebo-adjusted separation, or d≈0.45 under the central variance assumption, making ACAD a magnitude thesis rather than a simple positive-or-negative binary.

Where the market may differ

The market is inclined to treat RADIANT as a binary direction event. Clinaptis expects statistical success to be the likelier outcome and asks whether magnitude, dose behavior, trajectory, responders and global measures sufficiently improve the effect Phase 3 is likely to reproduce.

A reported effect near d≈0.45 is consistent with the house forecast; a coherent result above approximately d=0.55 is where conviction should increase.

What matters most

01
Magnitude, not headline direction: statistical success is the first hurdle, while the observed effect determines how much Phase 3 risk actually falls.
02
Dose direction, onset, trajectory, responder analyses, CGI-S concordance and safety must form a coherent efficacy package.
03
RADIANT Phase 2 is de-risking rather than registrational; the economic question is what effect Phase 3 can reproduce.
04
NUPLAZID and DAYBUE provide real downside support, but failed high-profile data could temporarily overwhelm standalone value.
05
The post-readout stance depends on evidence quality and the opening valuation; a positive endpoint is not an automatic Add at any price.

Assets, indications, next decision point

Acadia Pharmaceuticals programs: asset, indication, stage and next event
Asset	Indication	Stage	Next event
NUPLAZIDpimavanserin	Parkinson's disease psychosis	Commercial	Quarterly demand, prescription and guidance updates
DAYBUEtrofinetide	Rett syndrome	Commercial	STIX uptake and international expansion
Remlifanserin	Alzheimer's disease psychosis	Phase 23	RADIANT Phase 2 topline expected September–October 2026
Remlifanserin	Lewy body dementia psychosis	Phase 2	ILLUMERA development update

Published work behind the current view

Supporting evidence

Selective 5-HT2A modulation has direct randomized human activity in Alzheimer's disease psychosis, although the prior evidence was not approval-quality.
RADIANT prospectively targets a biomarker-confirmed Alzheimer's population rather than extracting evidence from a heterogeneous dementia study.
Remlifanserin provides greater exposure headroom with a cleaner demonstrated cardiac profile.
The trial is powered around a moderate d≈0.40 design assumption, close to the independently reconstructed Clinaptis case.
Phase 3 screening and enrollment began before Phase 2 topline, reducing calendar loss after a positive result.
NUPLAZID and DAYBUE provide a commercial base that funds pipeline development.

Tempering evidence

The historical pimavanserin evidence establishes plausibility, not a reliable effect-size benchmark.
The modeled d≈0.45 depends on trial variance not yet disclosed; roughly three raw SAPS-H+D points is the more durable forecast.
Transferability across prior instruments and populations is the least empirically anchored model input.
A roughly 106-patient-per-arm Phase 2 study cannot precisely estimate the effect Phase 3 will reproduce.
Conditioning on statistical success creates winner's-curse inflation, especially when the underlying effect is weak.

Strongest opposing view

The strongest bear case is that a cleaner molecule and trial design do not create stronger biology: pimavanserin already achieved high receptor occupancy, and a positive Phase 2 could be a favorable small-trial draw. The strongest bull response is that RADIANT fixes the population, endpoint, exposure and operational problems together, so waiting for mature confirmation could miss the principal re-rating.

Note 01
Jul 29, 2026
Acadia's Alzheimer's Psychosis Bet: Direction Is Priced, Resolution Is Not
We're constructive into Sep/Oct: ~70–80% odds of success, and we like the setup — the alpha's in the magnitude: fade a thin print, add on a strong one.
→

Acadia Pharmaceuticals

The maintained investment case

Where the market may differ

Assets, indications, next decision point

The events that can move the file

Published work behind the current view

Acadia's Alzheimer's Psychosis Bet: Direction Is Priced, Resolution Is Not

What changes the view

Principal risks

Strengthens the thesis

Weakens the thesis

Break conditions: what would end the thesis