Thesis

CABA is a $310M clinical-stage biotech that the market gives a 22% chance of surviving. We think it's 60%. The gap is the entire trade.

Three RMAT designations. Registrational trial enrolling in an indication with zero approved treatments. 100% Phase 1/2 hit rate. Insider buying at $2.26. Options chain 590x call-skewed into the H1 data window. Stock down 12.5% this week because Ernst & Young wrote "going concern" in the audit opinion.

The market is pricing two risks as independent: clinical data and capital survival. They are not independent. They are linked at ≈0.85 correlation through a reflexive warrant mechanism. Positive data mechanically resolves the balance sheet. Negative data kills both. This is a single-factor bet on H1 2026 data quality, purchased at a going concern discount.

LR 1.7. Market-implied survival is 22%; conditioned survival given the warrant mechanics is 58%. The correlation between data and capital is the alpha.

The Warrant Reflexivity

This is where the thesis gets structural. Start here because everything else is context for understanding this mechanism.

53 million warrants at $2.50 expire September 12, 2026. Stock is $2.80. The warrants are $0.30 in the money. If exercised, they generate ≈$132M -- enough to extend the runway past the 2027 BLA submission.

The reflexive loop:

Positive H1 data --> stock rises --> warrants exercised --> $132M cash
  --> going concern resolved --> stock rises further --> BLA path funded

And the reverse:

Negative data --> stock falls below $2.50 --> warrants expire worthless
  --> $132M lifeline evaporates --> going concern terminal --> death spiral

The data IS the financing. The market doesn't need to "decide" to fund CABA. Warrant holders exercise mechanically when they're in the money. If the H1 2026 readouts are strong, the capital problem solves itself without a single board meeting or term sheet.

Back out the market-implied survival probability:

P(survive) x $12.00 + P(die) x $0.25 = $2.80

P(survive) = ($2.80 - $0.25) / ($12.00 - $0.25) = 21.7%

The market says 22%. But condition survival on data quality -- which the warrant mechanics require:

P(survive | good data) = 0.90  (warrants exercise, ATM available, going concern lifts)
P(survive | bad data)  = 0.10  (maybe a distressed deal saves equity, probably not)
P(good data)           = 0.60  (based on 100% Ph1/2 hit rate, mechanism validated in PV)

P(survive) = 0.90 x 0.60 + 0.10 x 0.40 = 0.58

Fair value at 58% survival: 0.58 x $12.00 + 0.42 x $0.25 = $7.07. That's 2.5x current.

The market models data risk and cash risk as independent. They're correlated through price. This correlation is the alpha.

Why the Clinical Assets Matter

The warrant reflexivity only works if the clinical data has a real shot at being good. Here's why we assign 60% probability to positive H1 2026 readouts.

Myositis registrational trial (initiated December 2025). 17 patients. Single-arm. 16-week primary endpoint: TIS response off immunomodulators. FDA agreed to this design under RMAT. Background rate of spontaneous response: under 10%. Phase 1/2 hit rate in evaluable patients meeting registrational criteria: 100%. The target -- antisynthetase syndrome -- has zero FDA-approved treatments. Not "competitive." Zero. 15,000 patients with nothing.

The RMAT density is anomalous. CABA has 3 RMATs across myositis, SLE/LN, and systemic sclerosis. Kyverna has 2 (SPS and MG). Fate has 1 (SLE). Adicet and Allogene have zero. Three RMATs across three distinct rheumatologic indications in a $310M company is the cheapest regulatory portfolio in autoimmune CAR-T. Kyverna trades at $500M with two.

No-preconditioning is the sleeper. In pemphigus vulgaris, 2 of 3 patients achieved complete B cell depletion without lymphodepleting chemotherapy. No ICANS. Similar expansion kinetics to conditioned patients. SLE no-preconditioning data is expected H1 2026. If it replicates, CABA has something that fundamentally changes the commercial equation: outpatient CAR-T for autoimmune disease without chemo. Rheumatologists who won't refer for fludarabine/cyclophosphamide plus infusion might refer for an outpatient injection.

This is an industry-wide direction -- Fate has 2 conditioning-free patients in SLE, Adicet and Allogene are exploring reduced conditioning -- but CABA and Fate are furthest along. Kyverna is the only major competitor NOT pursuing conditioning reduction. That's telling.

Cellares manufacturing. CABA is the only autoimmune CAR-T company with a Cellares Cell Shuttle deal. BMS validated the platform with $380M. Cellares raised $612M total. We searched KYTX, FATE, ACET, and ALLO 10-K filings -- none mention Cellares. INDa cleared January 2026. First clinical data H1 2026. If it works, autologous CAR-T manufacturing goes from $350-500K per patient artisanal process to scalable automated production.

Why the Going Concern IS the Entry

E&Y issued a formal going concern. Management confirmed: "substantial doubt about our ability to continue as a going concern for at least twelve months following the filing." Cash $133.6M. Burn $131.1M in 2025, up 49% from 2024. Runway: "into the fourth quarter of 2026."

The burn is structural. R&D jumped 47% to $142.7M -- manufacturing scale-up (+$22.7M), clinical enrollment across five simultaneous trials (+$17.3M), headcount (+$7.7M). This accelerates as registrational enrollment ramps and Lonza/Cellares manufacturing builds out. Accumulated deficit: $517M.

This is genuinely existential. The June 2025 financing tells you where capital markets see this company: 50M shares plus 50M warrants at $2.00. Distressed pricing. Shares outstanding: 111.3M. More dilution is coming regardless.

But here's what the market misses: the going concern was already known. Management disclosed substantial doubt in prior quarters. The E&Y formalization triggered another wave of headline selling -- down 12.5% this week. Same clinical assets, lower price. The going concern is the discount mechanism that creates the entry. Everyone sees the headline. Not everyone does the conditional probability math on the warrants.

Eight insiders bought $296K at $2.26 in January. CEO put in $100K. Directors and officers bought below the warrant strike. They can see the pipeline data. They bought personal money into a going concern.

The Bear Case (40% Probability)

Not a straw man. The going concern kills companies.

Capital markets are hostile to going concern names. Institutional mandates prohibit holding auditor-qualified going concerns. The buyer pool for any equity raise is restricted. The $117M ATM remaining could be used, but selling into a thin market at $2.80 risks cratering below the $2.50 warrant strike, triggering the exact death spiral the company needs to avoid.

The $2.50 floor is one bad week from breaking. Beta 3.28. A 4% market decline translates to 13% CABA decline -- below $2.50 in one session. Max pain across all expirations: $2.00. The warrant lifeline is fragile.

Big Pharma is buying in vivo, not autologous. $7.35B in autoimmune CAR-T M&A went to in vivo platforms (AbbVie/Capstan $2.1B, BMS/Orbital $1.5B, Lilly/Orna $2.4B, Gilead/Interius $350M, AZ/EsoBiotec $1B). Roche/Poseida ($1.5B) went allogeneic. Zero went to autologous. CABA appears in zero Big Pharma earnings transcripts. No cavalry is coming unless data forces it.

IASO license: latent China risk. Rese-cel's core CD19 binder is licensed from Nanjing IASO Biotherapeutics. BIOSECURE primarily targets manufacturing, not IP licenses, and IASO isn't on any sanctions list. But one executive order could change this. The IP foundation of the entire program runs through a Chinese entity.

FDA policy is unstable. The myositis BLA strategy depends entirely on RMAT-enabled accelerated approval: 17-patient, single-arm, surrogate endpoint. DOGE efficiency mandates and RFK's FDA influence could tighten standards. The Feb 2026 "Plausible Mechanism Framework" guidance was encouraging. Policy reversals happen.

The Trade

Position: Long CABA common stock, 1-1.5%. Entry: Split -- 0.75% at current ($2.80), add 0.75% at $2.50 if available. Average ≈$2.65. Stop: Below $2.00 (-28%). Warrant lifeline dead, thesis break. Targets: $5.00 (+79%), $8.00 (+186%), $12.00 (+329%). Vehicle: Common stock. Options are overpriced at 110-163% IV. Common captures the same asymmetry without theta decay or IV crush. Horizon: 6 months. By September, everything resolves.

Scenarios

12-month scenarios compress to: Bull 20% / $8.00, Base 40% / $3.50, Bear 40% / $0.75. EV: $3.30 (+18%).

Risk/reward to first target: 2.8:1. To second: 6.6:1.

What the Options Market Is Telling You

P/C open interest ratio in August: 0.002. That's 590 calls for every put. The $2.50 call strike has 14,968 OI in May and 13,804 in August -- the entire options market is a bet on whether the stock holds above the warrant strike through the data window.

Call IV trades 15-63% above put IV across all expirations. Normally calls trade cheaper than puts. Here the market is paying up for upside exposure. Unusual for a going concern name. Whoever is accumulating these calls sees what we see.

September $8 calls: 1,005 OI at $0.40. Someone paid $40K+ for a 3x move. Small dollars but directional conviction.

Timing

NOW (Mar 23)     10-K filed. Going concern headline. Stock oversold. ENTRY WINDOW.
Apr-May          Conference abstracts. Data teasers. Market starts pricing catalyst.
Jun (EULAR)      PRIMARY CATALYST. SLE no-preconditioning data. SSc/LN/gMG updates.
Jul-Aug          Cellares manufacturing data. Full H1 data package.
Sept 12          HARD DEADLINE. 53M warrants expire. $132M capital or zero.
Q4 2026          Cash cliff. Must have raised by now.
2027             BLA submission (if funded).

The entry window is now through April. By June, you're right or late. Do not wait for data confirmation -- by definition, positive data gaps the stock and the entry is gone.

What Breaks It

Bearish: stock below $2.30 for 5+ days (warrant lifeline jeopardized). Disappointing SLE no-preconditioning data. Distressed raise below $2.00. FDA RMAT policy reversal. IASO license complication.

Bullish revision triggers: positive SLE data leaks (abstracts, KOL commentary). Named pharma partnership. Capital raise at $4+. Warrant exercises accelerating.

Open Questions

1. Evaluable patient N in myositis Phase 1/2. "100% hit rate" -- but what's the denominator? ACR Convergence 2025 abstract has this data. We haven't found it.
2. Nature Biotechnology Feb 2026 comparative review. Peer-reviewed assessment of rese-cel vs all CD19-CAR-T constructs. Could validate or undermine competitive positioning.
3. FDA RMAT stability under DOGE/RFK. Entire BLA strategy rests on this pathway. Unquantifiable policy risk.
4. Partnership conversations. Zero pharma mentions in transcripts. Is anyone talking to CABA, or is the going concern forcing distressed deal discussions?

Conviction

This is a single-factor bet disguised as a multi-factor stock. Clinical data is 35% of variance. Capital survival is 30%. They're correlated at ≈0.85 through the warrant mechanism. It's the same bet. Edge is ≈35-42% (below the 45% threshold for full sizing -- hence 1-1.5%, not 3-5%).

The honest read: the clinical assets are genuinely strong and genuinely underpriced. The going concern is genuinely existential. The warrant reflexivity is the structural insight that the market may not be pricing. The options chain and insider buying suggest informed participants see it.

A call option on H1 2026 data quality, purchased at a going concern discount, where positive data mechanically resolves the capital crisis. Size for surviving the bear case. Entry window is the 10-K selloff, before catalysts price in.

Evidence