Time Horizon: 12-18 months. Phase 3 design announcement (Q3 2026) is the near-term forcing function. Vertex AMPLITUDE data (late 2026/early 2027) is the structural resolution. MAZE's clinical trajectory plays out over 2027-2029 if Phase 3 proceeds.

Base Rate: Clinical-stage biotech, Phase 2 to approval.

Base rate: Phase 2 biotech → approval = 22-32% (small molecule, rare disease)
MAZE-specific: FSGS path ≈20%, broad AMKD ≈12-15%
Prior odds: 0.25 (using 20% midpoint)

Alpha vs Beta:

Factor regression (trailing 250d):
  β_SPY  = -0.03  (zero market)
  β_QQQ  = -0.09  (zero tech)
  β_XBI  = +1.18  (biotech sector)

Variance decomposition:
  XBI sector:      10.3%
  Idiosyncratic:   89.7%  ← well above 75% target

This is a pure company bet. No factor exposure to separate.
All return — up or down — is idiosyncratic.

No alpha/beta decomposition is meaningful here because there is no forward alpha estimate. The stock is binary on clinical outcomes. At 90% idiosyncratic variance, the factor profile is clean — what you see is what you get.

B — Business Model

Clinical-stage genetics-first biotech. Zero product revenue. Binary CF structure: all-or-nothing on clinical success.

Compass Platform: Mines protective genetic variants from large biobanks (UK Biobank, Million Veteran Program), determines molecular mechanism, designs small molecules that mimic the protection pharmacologically. Produced three clinical candidates. Scientifically sound, not unique — Regeneron and deCODE/Amgen do the same. Patent protection covers molecules, not genetic insights. A factory, not a moat.

Three Programs:

MZE829 (APOL1 inhibitor) — Lead. Targets kidney disease caused by APOL1 G1/G2 variants (toxic gain-of-function in podocytes). Designed to mimic N264K protective variant, which reduces ESKD risk 81% through ion channel blockade. Phase 2 complete. Phase 3 design expected Q3 2026.

Phase 2 HORIZON results (8-K 2026-03-25):

The company reported means for favorable subgroups (non-diabetic 48.6%, FSGS 61.8%) and withheld the diabetic mean, disclosing only the responder rate (2/5 = 40%). Back-calculation from disclosed data yields ≈17.4%, well below the 30% clinically meaningful threshold. The 35.6% broad AMKD headline is an artifact of the FSGS subset pulling up the average.

MAZE claims a "dual mechanism" — blocking both ion conductance and pore assembly. The 10-K hedges this three times in one sentence: "Preclinical studies suggest that MZE829 may disrupt APOL1 pore assembly... a so-called dual mechanism" (line 541, q=0.95). The same-day press release says "uniquely inhibiting" with zero qualifiers. The binding document under SOX was weakened by counsel. N264K achieves 81% ESKD reduction through channel blockade alone — it does not prevent pore assembly. If channel blockade is sufficient, the "dual mechanism" adds nothing proven. Vertex published data (Nature Communications, January 2025) describes inaxaplin as a "highly specific channel blocker," contradicting MAZE's characterization of it as "pore only."

Safety was clean (no SAEs, no severe TRAEs) but n=15 enrolled with 3 excluded for non-compliance (20% exclusion rate in a 12-week, once-daily oral trial).

MZE782 (SLC6A19 inhibitor) — Pipeline. Possibly more interesting than MZE829. Targets PKU (phenylketonuria, 60K patients) and CKD (37M patients). Genetic validation is stronger than MZE829: bidirectional GWAS signal at genome-wide significance (p < 5x10⁻⁸), with a single loss-of-function allele conferring ≈50% renal progression reduction. Phase 1 showed 42x increase in phenylalanine excretion (vs 10x benchmark). Well tolerated. Yellow flag: vasculitis-like finding at supratherapeutic doses. PKU Phase 2 by mid-2026 with fast biomarker readout (plasma Phe). CKD Phase 2 in H2 2026.

The market prices MAZE almost entirely on MZE829. MZE782 provides a floor that is largely ignored.

MZE001 (GYS1, Pompe Disease) — Licensed. Licensed to Shionogi for $150M upfront + $605M in milestones + royalties. Shionogi dosed first Phase 2 patient March 2026 ($20M milestone expected Q2 2026). Validates the Compass platform as a drug-production engine. No longer on MAZE's P&L.

Φ — Financial Trajectory

The strongest dimension. Financial position is well-understood by the market and consensus.

Liquidity stack:

Runway: ≈2.8 years conservative (existing cash + Hercules 1A + Shionogi) at $150M/yr burn. Up to 4.6 years with full facility access.

What's changing:

• Burn accelerating 30-40% YoY: $111.9M operating cash used in FY2025, estimated $150-170M in FY2026 (Phase 3 prep + dual MZE782 trials)
• SBC growing 70% YoY (vs 30% opex growth): $16.4M FY2025, $28.4M unrecognized over ≈3 years
• Hercules debt at 7.95% floor adds $3.2M/yr interest (new in FY2026)
• ATM overhang: $200M at $29 = 14% dilution if fully drawn; at $16 (IPO price) = 25%
• Clean EY audit, no going concern, no restatements, single operating segment
• Quarterly burn: Q1 $32.8M → Q4 $34.6M (slight acceleration)
• Tax shield: $283.6M federal NOLs (no expiry) + $15.4M R&D credits

Expectations gap: ZERO. Financial position is well-covered by 12 analysts and accurately reflected in consensus. No edge here.

K — Competitive Position

The weakest dimension. Three findings from primary sources that consensus underweights:

1. Vertex is 18-24 months ahead and building a nephrology moat.

Vertex ($120B) has AMPLITUDE Phase 3 for inaxaplin (APOL1 inhibitor) with data expected late 2026/early 2027. Breakthrough Therapy Designation. AMPLIFIED Phase 2 enrollment complete, testing in diabetics with moderate proteinuria — the exact population where MAZE showed ≈17.4% mean uACR reduction (below threshold). Vertex CEO stated renal franchise will "ultimately rival CF business" (≈$12B revenue). Povatacicept BLA submission for IgAN initiated December 2025, building a nephrology salesforce that inaxaplin can leverage. Four renal programs across multiple diseases. MAZE has zero commercial infrastructure.

2. MAZE removed the Vertex comparison from its 10-K.

FY2024 10-K (lines 997-1013): "MZE829 was approximately 100 times more potent than an independently synthesized sample of inaxaplin." FY2025 10-K: zero mentions of Vertex, inaxaplin, or VX-147. This is a material disclosure change — either legal caution or reduced confidence in the differentiation claim.

3. MAZE is invisible to competitors.

Cross-corpus search across all earnings transcripts for "MAZE Therapeutics," "MZE829," "MZE-829," and "APOL1 inhibitor" returned zero mentions by any company. No competitor names MAZE on earnings calls. This is informative: at $1.4B cap, MAZE is a rounding error for Vertex's competitive monitoring.

Third competitor: AstraZeneca/Ionis (AZD2373, antisense oligonucleotide) in Phase 2b with 67% APOL1 protein reduction but subcutaneous injection (vs MAZE/Vertex oral) and no patient efficacy data yet. Completion expected August 2027.

FSGS regulatory white space is the differentiation thesis: Vertex targets broad AMKD, MAZE can target APOL1-FSGS specifically. This is real but unquantifiable. FSGS addresses ≈25-40K patients vs ≈250K for broad AMKD. TVTX FILSPARI PDUFA (April 13, 2026) may validate the proteinuria endpoint for FSGS via a different mechanism (dual endothelin/ARB), establishing regulatory precedent.

Moat: None today. Potential K_reg (FDA exclusivity) upon approval. No network effects, no switching costs, no scale advantage, no brand. The Compass platform is a factory, not a moat — its genetic insights are not proprietary (same biobanks are available to Regeneron, deCODE/Amgen).

G — Governance

Bearish signals dominate. Board quality partially offsets.

Insider transactions (12 months): Zero buys. $11M in sales.

All under 10b5-1 plans. The CMO's selling is the most informative signal: Bernstein served as Head of Translational Medicine at Vertex from October 2017 to September 2021 (10-K line 10596). He has firsthand knowledge of inaxaplin's development. He then sold 100% of direct shares in the weeks before MAZE's Phase 2 data release ($2.5M across March 10 and March 20, data on March 25). Post-sale direct holdings: zero. This is not generic insider liquidation — it is the person with the deepest competitive benchmark knowledge exiting completely.

VC funds exiting aggressively: Third Rock Ventures (-26% QoQ), Matrix Capital (-71%), Alphabet/GV (-42%), Foresite Capital (-75%). Four of seven independent board members represent funds actively reducing. Officers + directors + affiliates hold ≈52% of voting stock, but this is declining rapidly.

Board quality (partial offset): Andre Hoppenot (ex-Incyte CEO, added October 2025) as independent chair. Rajeev Kumar (BridgeBio CEO, added March 2026). Neil Spiegelman (ex-BioMarin CFO) on audit committee. These additions signal commercial maturity preparation — but the people who know the science best are selling.

Compensation: All options repriced to $1.08 (December 2024). CEO has 8.2M options with ≈$28/sh intrinsic at current price. Management has massive embedded gains regardless of future performance. Double-trigger CIC with moderate multiples (1.5x CEO, 1.0x others). No dual-class. Classified board with supermajority amendment requirement (standard Delaware).

β — Factor Profile

Clean. No factor management required.

r_MAZE = α + β_SPY·r_SPY + β_QQQ·r_QQQ + β_XBI·r_XBI + ε

β_SPY  = -0.03   (zero market exposure)
β_QQQ  = -0.09   (zero tech exposure)
β_XBI  = +1.18   (biotech sector, 10.3% of variance)
Idio:    89.7%    (well above 75% target)
R²:      9.8%

σ_total = 103.1% annualized
σ_idio  = 97.5% annualized

Recent path is entirely idiosyncratic. XBI was +2% while MAZE was -36% in the past month. Zero correlation to biotech sector during the crash. The stock moves on clinical data, not market or sector direction.

Short interest: 8.9% of float, 7.6 days to cover. Moderate — not a squeeze candidate, not a crowded short.

Options positioning: OI P/C = 0.34 (legacy bullish calls, 99.7% OTM — tombstones from pre-crash). Volume P/C = 1.54 (current flow is puts). May $20 puts: 312 volume vs 6 OI (52x unusual) — directional bearish bet. IV term structure 105% → 81% → 70% (steep contango, market expects vol compression). ATM IV 70-80% vs 93% realized — options are cheap relative to actual moves. Total OI across all expirations: ≈6,000 contracts. No meaningful gamma structure — too thin to create floors or ceilings. Stock moves on news flow, not options plumbing.

Δ — Expectations Gap

The gap that matters: TAM shrinkage

Street models broad AMKD. The drug works in FSGS only.

Pipeline value implied by market:
  EV = ≈$1.35B → pipeline = ≈$1.32B (after cash floor $7/sh)

Street assumption (12 analysts, 100% Buy, $46-110 targets):
  Broad AMKD, peak $1.5-2B, P(success) 40-50%

Primary source finding:
  FSGS only, peak $400-800M, P(success) 12-20%

The derived diabetic mean (≈17.4%) and non-FSGS non-diabetic mean (≈15.6%) eliminate ≈70% of the addressable population the analysts are modeling. Management is guiding "pivotal program in patients with AMKD" (broad). The 12 analysts follow management's guidance, not the data.

Current $29.05 sits between FSGS fair value ($11-18) and broad AMKD fair value ($42+). The market is pricing a blend — roughly 60% weight on FSGS outcome, 40% on broad AMKD possibility. The data says broad AMKD weight should be near zero.

Ranked gaps

All four negative gaps are high-q (SEC filings, Form 4s). The one positive gap (MZE782) is lower q (Phase 1 only) and smaller magnitude. Net expectations gap is materially negative.

Mispricing assessment

The mispricing is real but not actionable today. The forcing function (Phase 3 design, Q3 2026) is five months away. RSI 12.6 at time of research means a mechanical bounce is the near-term move. 103% annualized vol makes position sizing negligible at any responsible Kelly fraction. No options liquidity at the right expiry — August (141d) has 309 total OI and may expire before the catalyst; November (232d) has 45 OI, effectively dead. A gap without a near-term forcing function and no liquid vehicle to express it is not a trade.

Key Risks

Clinical risk (binary). MZE829 is a Phase 2 asset. The FSGS efficacy signal (61.8%, n=5) is encouraging but statistically meaningless — a 95% CI at n=5 spans roughly 20-100%. Phase 3 enrollment, execution, and endpoint selection are all undetermined. Management has not disclosed the FDA End-of-Phase-2 meeting outcome, meaning Phase 3 design is not settled.

Regulatory risk. Zero FDA designations (no BTD, Fast Track, Orphan Drug, RMAT). Vertex has BTD. If FDA requires eGFR slope rather than proteinuria for APOL1-AMKD/FSGS, MAZE needs a longer, larger, more expensive trial — the 12-week proteinuria data becomes insufficient. Vertex CEO stated (Q4 2025 transcript): "recall FDA Accelerated Approval AMKD based on 48-week eGFR."

Competitive risk. Vertex is better capitalized ($120B vs $1.4B), further advanced (Phase 3 vs Phase 2), has regulatory advantages (BTD), and is building commercial infrastructure (nephrology salesforce from IgAN). VRTX AMPLIFIED targets the diabetic population where MAZE failed. If AMPLITUDE is positive and Vertex captures broad AMKD, MAZE's addressable market narrows to FSGS niche (≈25-40K patients).

Dilution risk. $200M ATM at current $29 = 14% dilution. Accelerating burn ($150-170M FY2026E) with Phase 3 costs ahead. VC funds exiting aggressively.

MZE782 tox risk. Vasculitis-like finding at supratherapeutic doses in preclinical. If confirmed in Phase 2, kills the pipeline floor.

What to Watch

Steelman Bear Case

The strongest argument against a long position is not that the Phase 2 data was disappointing — the market has already repriced that (−36%). The strongest argument is that the stock is still overvalued even after the crash.

At $29.05, the market pays $22/share ($1.32B) for the pipeline after backing out the cash floor. This requires either broad AMKD success at 35-44% probability or FSGS success at an implausible 73%. The data says the drug does not work outside FSGS. The 12 analysts model broad AMKD because management hasn't conceded FSGS-only — but the derived subgroup means (≈17.4% diabetic, ≈15.6% non-FSGS non-diabetic) are below the 30% clinical threshold. When Phase 3 design forces the issue (Q3 2026), every analyst target gets cut.

The CMO's selling makes this worse, not better. Bernstein is not a generic insider — he spent four years at Vertex as Head of Translational Medicine (2017-2021) with firsthand knowledge of inaxaplin's development. He sold 100% of direct shares before MAZE's own data readout. The person with the deepest competitive benchmark knowledge chose to exit completely.

The "dual mechanism" — MAZE's primary competitive differentiator — is marketing, not science. The 10-K's own language hedges it three times ("suggest...may...so-called") while the press release omits all qualifiers. N264K achieves 81% ESKD reduction through channel blockade alone. Vertex's published data describes inaxaplin as a "highly specific channel blocker" that prevents cell death — contradicting MAZE's claim that inaxaplin is "pore only." No clinical evidence can distinguish the two mechanisms. The differentiator does not differentiate.

This is not a noisy signal. The bear case is that $29 is above fair value for a company whose lead drug works in a 25-40K patient niche, competing against a $120B competitor with an 18-24 month head start, Breakthrough Therapy Designation, a nephrology salesforce, and clinical data coming in the exact populations where MAZE failed.

Concession: MZE782 is a legitimate asset with stronger genetic validation than MZE829, and the market underprices it. If PKU Phase 2 delivers (mid-2026), it provides a meaningful floor. But MZE782 is Phase 1 complete with a tox flag — it cannot carry a $1.4B valuation by itself.

Kill Criteria

Thesis dies if (bearish thesis weakened):
  - Insider buying: any C-suite open-market purchase → reassess governance signal
  - FSGS Phase 3 data shows n>20 with mean uACR >50% and tight CI → drug works, TAM may expand
  - Vertex AMPLITUDE fails → MAZE becomes class leader, transforms competitive position
  - FDA grants BTD or Orphan Drug → regulatory path accelerates, differentiates from Vertex

Thesis strengthened if (bearish thesis confirmed):
  - Phase 3 design = FSGS-specific → analyst TAM models collapse, targets cut to $20-30
  - Phase 3 design = broad AMKD → P(success) drops to ≈12%, worse risk-adjusted outcome
  - Vertex AMPLITUDE positive → MAZE niche or acquisition at discount
  - MZE782 tox confirmed in Phase 2 → pipeline floor removed
  - ATM drawn at depressed prices → dilution accelerates

LR Signal

Memo LR: 0.4 (strong bearish — market overpricing)

The divergence from market pricing is large and supported by high-credibility evidence. The TAM shrinkage (broad AMKD → FSGS only) is derived from the company's own 8-K data at q=0.95. The CMO's selling and Vertex background are from Form 4 and 10-K at q=0.95. The "dual mechanism" hedging is from the 10-K at q=0.95. All four bearish signals come from Tier 1 sources.

The market (12 analysts, 100% Buy, $46-110 targets) has not incorporated the subgroup analysis, the CMO's Vertex connection, the 10-K language hedging, or the Vertex competitive positioning. The cumulative evidence LR from primary source research was 0.10 — a 10x bearish shift from prior. The street hasn't caught up.

LR is not lower (more bearish) because: (a) n=5 is genuinely noisy — the FSGS signal could strengthen with more data, (b) MZE782 provides a non-zero floor the market may be partially pricing, (c) management hasn't conceded FSGS-only — the broad AMKD path remains their stated intention, and until Phase 3 design is announced, there's a non-zero probability they have unpublished data supporting broader efficacy.

Evidence