CNS drug development is often viewed in terms of binary outcomes. In actuality, they also are probabilistic, and those probabilities can be influenced by design. The distinction between a typical Phase 2 program and a ‘better engineered’ one is often subtle, but it matters.

MapLight Therapeutics – which trades as MPLT on Nasdaq – appears to have built its pipeline with that distinction in mind, and the market doesn’t yet seem to recognise the structure.

From Base Rates to an Engineered Framework

Historically, Phase 2 CNS trials succeed roughly 30–40% of the time. That base rate assumes first-in-class biology, variable translational exposure, heterogeneous trial design, and limited capital flexibility.

MapLight’s structure differs along several of those axes.

Between late 2026 and 2027, the company expects Phase 2 readouts in schizophrenia (ZEPHYR), Alzheimer’s disease psychosis (VISTA), and autism spectrum disorder (IRIS). Superficially, that resembles three independent coin flips. Structurally, it functions as a partially correlated portfolio with targeted risk mitigation.

Updated Probability Framework

If we take into account how Maplight is designing the advancement of their three programs – As well as how they’ve organized their capital runaway, and created optionality for the future, we can tune standard ‘base case’ rates of success, to ‘Maplight specific’ outcome probabilities.

Probability Lift 1: The Biological Layer — Mechanism and Exposure

Translational failure in CNS typically arises from two sources: inadequate central exposure or incorrect biological mechanism. MapLight addresses both.

Across multiple Phase 1 studies — including repeated-dose designs and cohorts of healthy adults and healthy elderly participants — ML-007 achieved steady-state cerebrospinal fluid (CSF) concentrations of approximately 140 ng/mL. Preclinical modeling indicates receptor engagement between roughly 14 and 27 ng/mL, implying a five- to ten-fold exposure margin within tolerable dosing. That reduces delivery risk.

Mechanism risk has also been materially reduced. The M1/M4 muscarinic pathway has already been clinically validated in humans. This does not guarantee success for every molecule. But it lowers the probability that Phase 2 failure would reflect an invalid biological target.

ML-007 therefore enters Phase 2 not as a first-in-class experiment, but as a differentiated molecule within a validated pathway.

Together, exposure margin and mechanism validation reduce fundamental translational uncertainty

Probability Lift 2: The Methodological Layer — Preserving Assay Sensitivity

Placebo inflation remains a structural risk in CNS trials.

Maplight is deliberately addressing this risk. The schizophrenia study (ZEPHYR) limits enrolment to a small number of high-quality U.S. sites, excludes patients with large screening-to-baseline symptom swings, and tightly controls dosing and pharmacokinetic synchronisation.

These measures aim to preserve assay sensitivity — ensuring that if a drug signal exists, the trial is capable of detecting it.

This does not eliminate risk. But it addresses a common cause of false negatives.

Probability Lift 3: Correlation and Portfolio Structure

Schizophrenia and Alzheimer’s disease psychosis share muscarinic pharmacology but differ meaningfully in patient biology and endpoints. Outcomes are partially correlated, not identical.

Autism spectrum disorder introduces a different receptor system, reducing stacked mechanism risk.

This ‘three indication’ structure increases the probability of at least one success an improves the odds of two independent readouts.

Probability Lift 4: The Financial Layer — Time as a Risk Variable

Small-cap biotech programs often fail because capital runs out before data are available.

MapLight exited its IPO with approximately $500 million in pro-forma cash, sufficient to fund operations through major Phase 2 readouts. This reduces the probability of forced dilution or strategic compromise before key inflection points.

Probability Lift 5: The Discovery Layer — Extending the Right Tail

A collaboration with google spinoff SandboxAQ introduces longer-term optionality through AI-enabled GPCR / compound modelling. MapLight retains development and commercial rights to any advanced compounds while SandboxAQ participates via milestones. This doesn’t impact the success probability of the 3 programs underway. But it does add an accelerated path to future, more accurate shots on goal.

The Implication of Probability Improvements

Based on Maplight advancing 3 programs at the same time, with probabilities of success enhanced, I hypothesise that there is an 89% likelihood of at least one indication advancing to P3, a 49% of two, and a small but noteworthy chance of all 3 advancing.

The difference between baseline vs. Maplight Engineered probabilities reflects reduced translational uncertainty, improved trial design, partial correlation awareness, and capital runway.

The valuation implications for these probability sets are beyond the scope of this post, however it is likely a safe assumption that MPLT’s current Enterprise Value does not reflect the likelihood of the base probability set, let alone the probability adjusted case. And that discrepancy is where opportunity may lie.

The 2026–2027 Inflection Point

By late 2026, data will replace speculation. A single positive readout would likely justify a meaningful value re-rating by confirming clinical viability. Two positive readouts would materially reduce single-asset risk and shift valuation from a binary development story to a multi-indication franchise. The valuation implications in that case is significant. Three successes would suggest the emergence of a broader CNS platform – and further justify even higher valuation.

Authors note: I am intrigued by what Maplight has launched. Aside from this ‘probability engineering’ map, I am also a fan of the cap table, management, strategic corporate relationships, the quantitative AI tie in, and the economic and social wins of developing treatment in the CNS / psychiatric space. Perhaps I’ll follow up this post and address some of those strategic / structural factors soon.

Disclosure – The Author owns and is acquiring MPLT shares at the date of this blogpost. The Author is not compensated in any way by MPLT and has no commercial or financial relationship with the company except as an investor. Readers are required to do their own due diligence and not rely on any aspect of this blogpost for advice.