Catalyst Pharmaceuticals [CPRX] — Company Research

Tier-2 memo · framework: Criteria, 2026-07-29 · analysis date 2026-07-29 This document issues no position verdict. It scores Criteria and outputs an analysis. The book decides.

0. The finding that governs every other number in this memo

CPRX no longer exists as a listed security. Catalyst Pharmaceuticals was acquired by Angelini Pharma S.p.A. for $31.50 per share in cash. The merger closed on 2026-07-15. Every share outstanding was cancelled and converted into the right to receive cash.

The Tier-1 screen was run on 2026-07-28 — thirteen days after the merger closed and four days after deregistration. Its spot: 31.49 is not a price; it is a frozen last-trade tick from a security that had already stopped trading. Independently verified against Alpaca:

trades/latest   p=31.49  t=2026-07-14T19:59:54Z
quotes/latest   bid 31.48 / ask 31.51  t=2026-07-14T20:19:55Z
daily bars      no bar exists after 2026-07-14

There are no bars after 2026-07-14 on any feed. The quote is stale by the same 10 trading sessions.

The screen's headline — "+30.8pp margin, 40.1% demonstrated CAGR vs 9.3% required" — is therefore a valuation of a company whose equity had already been extinguished at a known, fixed, realised cash price. The implied-path test is not merely wrong here; it is answering a question that has no remaining referent.

The screen's own data_quality_ok field is false. That flag was set and then overridden by the ranking, which is the failure mode worth recording: the pipeline carried a self-declared quality failure through to a +30.8pp buy signal without the flag blocking anything.

Why "low volatility" made it worse, not better

The brief's framing — "low volatility plus a high margin makes this a natural candidate for real size" — is exactly the trap. The screen's vol_252d_pct: 32.0 reproduces on our own bar history (32.1%, 2025-07-14 to 2026-07-14), so the input is arithmetically correct. But it is a regime average across a structural break:

From 2026-05-07 the stock traded in daily ranges of one to five cents, drifting from $31.15 to $31.49 as the arbitrage spread decayed. Its volatility was 1.2% because it was a pinned deal stock, and then zero because it ceased to exist. Inverse-volatility sizing fed a 32% number would have sized this position up — into a security with no remaining upside above $31.50 and no ability to be sold after 2026-07-14.

Recommended pipeline control: a screen must assert that the latest trade timestamp is within N sessions of the screen date, and must check for Form 25 / Form 15 in the filing index. Both checks are one API call each and both would have rejected this record outright.

1. Mechanism — the named, evidenced driver

Setting the merger aside and underwriting the operating business as at the last reported quarter, there is a real, specific, dated mechanism, and it is not the one the screen found.

The FIRDAPSE upstream royalty step-down, effective January 2026.

Catalyst licensed North American FIRDAPSE rights from BioMarin in October 2012 (rights later sold to SERB S.A.). That agreement carried royalties for seven years from first commercial sale, tiered at 7% of US net sales up to $100m and 10% above. First US commercial sale was January 2019. The seven-year term therefore expired in January 2026. Simultaneously, under the July 2022 Jacobus settlement, the amifampridine royalty stepped from 1.5% to 2.5% for calendar 2026 onward.

Company statement (Q1-26 earnings release, 2026-05-11), verbatim:

"in January 2026, the Company completed seven years from the date of first commercial sale of FIRDAPSE in the U.S. On that date, the royalty on net U.S. sales that the Company previously paid to Catalyst's licensor at a tiered rate of 7% to 10% of net U.S. sales of FIRDAPSE expired… the overall royalty rate that the Company pays to its upstream licensors for net U.S. sales of FIRDAPSE is 6%, which is down from a previous maximum rate of 18.5%."

This is evidenced, dated, quantified by the company, and already visible in the P&L:

Scaled: on the Q1-26 FIRDAPSE run-rate of ~$395m annualised, a 12.5pp reduction in the royalty rate is up to ~$49m/yr of cost removed, or ~7.8pp of group revenue at the FY26 guidance midpoint. This is the single largest earnings driver in the business and it is a contract clock, not a commercial achievement. It is non-recurring in the sense that matters: it happens once, it cannot repeat, and it does not compound.

Offsetting mechanism, in the opposite direction. AGAMREE — the fastest-growing product — carries the heaviest royalty load in the portfolio: a base 7% of North American net sales owed to the licensor for its upstream obligations, plus an additional 7% on net sales between $100m and $200m, escalating with scale (10-Q Q1-26 Note 13). AGAMREE passed $100m of annual net sales in 2025. Every incremental dollar of the portfolio's growth therefore arrives at a materially lower contribution margin than the FIRDAPSE dollar it is notionally replacing. The mix shift is margin-dilutive at the gross line even while the royalty expiry is margin-accretive.

2. Accounting quality — is the reported growth real?

This is the section the brief asks for, and the answer is: substantially less real than 40.1% suggests.

2.1 Product-level revenue, from the audited disaggregation tables

Source: FY2023 10-K, FY2024 10-K and FY2025 10-K, Note 2 "net product revenue disaggregated by product" ($ thousands).

¹ from 2023-01-24, the FYCOMPA acquisition date. ² from 2024-03-13, the AGAMREE commercial launch date.

2.2 The organic-versus-acquired split — the answer

The screen's 40.1% is the FY2022 → FY2025 total-revenue CAGR. It reproduces exactly (589.0/214.2)^(1/3) − 1 = 40.1%. Decomposing the $374.8m of revenue added over that window:

61.5% of the growth the screen credited to CPRX was bought, not built.

Same-product CAGR, FY2022 → FY2025:

The screen's required_cagr_pct was 9.3%. Against the correct organic denominator the margin is +9.5pp, not +30.8pp — before any of the corrections in §2.3 and in the Valuation document.

Counterfactual: absent both transactions, FY2025 revenue would have been ~$358.6m, not $589.0m.

RUZURGI is a fourth acquired asset that produced no revenue at all. Catalyst licensed and acquired RUZURGI from Jacobus in a July 2022 settlement following its successful Eleventh Circuit challenge to the FDA's approval of Jacobus's competing amifampridine. RUZURGI appears in zero rows of the product-revenue disaggregation table in any year. It was bought to remove a competitor and to buy a covenant — Jacobus agreed not to exploit any product competitive with FIRDAPSE or RUZURGI in the territory through the later of the royalty term or 2034-12-31. That is a legitimate defensive purchase; it is not growth, and no part of it appears in the 40.1%.

2.3 The growth rate is not just acquired — it has already stopped

The deceleration is violent and the current-portfolio rate is nowhere near 40%.

Q1-2026 by product:

The acquired asset that supplied 30.2% of the three-year growth is in outright collapse. FYCOMPA's Q1-26 annualised run-rate is $55.1m against FY2025's $113.3m — a 51% decline. The cause is named and dated in §3.

A screen that treats "40.1% demonstrated" as the hurdle reference is comparing a five-year forward requirement against a growth rate the company had already told the market, in writing, it would not repeat — three months before the screen ran.

2.4 "Promoted products" — a metric that appears exactly when the reported metric breaks

Management's mention-frequency series (§5) surfaces this generatively, and it is the disclosure-quality catch of this memo.

The term "promoted product" has never appeared in a Catalyst quarterly earnings release in the fifteen releases from 2023-02-07 to 2025-11-05. It appears for the first time on 2026-02-25 — the release in which total revenue growth decelerated to +19.8% and Q4 to +7.6% — and again on 2026-05-11:

• 2026-02-25 headline: "Promoted Product Revenue, Net Expected to Grow by 20% or More in 2026"
• 2026-05-11 headline: "Results include 28% YoY Increase in Net Product Revenues from FIRDAPSE & AGAMREE"

"Promoted products" is FIRDAPSE + AGAMREE. It is the portfolio with the declining asset removed. In the same release in which total revenue grew +5.6%, the headline number offered to the market is +28.2%.

This is not a misstatement — the total is disclosed on the same page, in the same table, and the definition is plain. But it is the same structural pattern the brief's precedents describe (Twist retiring orders; TXG's settlement-inflated line): a new, more flattering denominator introduced in the precise quarter the reported one turned. Any process reading headlines rather than the disaggregation table would have taken +28% as the growth rate. The screen took +40%. The reported number is +5.6%.

2.5 Reported net revenue was purchased with working capital — and the benefit is unquantified

FY2025 10-K, MD&A, verbatim:

"The increase in accounts receivable, net, primarily relates to a change in the payment terms resulting from the renegotiation of a contract between us and a customer. Under the revisions to the contract with this customer, among other changes, we are now paying reduced fees to the customer (which are recorded as a reduction in gross-to-net expenses), but the customer is paying amounts due on its obligations to us on a monthly basis rather than a semi-monthly basis… the payment of approximately $27.0 million that we would have received… at the end of December 2025 under the previous contract terms was received on January 2, 2026."

Read carefully, this says: Catalyst reduced the fees it pays its distributor, which mechanically reduces the gross-to-net deduction and therefore increases reported net revenue, in exchange for being paid more slowly.

The consequences are visible and large:

• AR grew +83.6% YoY at Q1-26 against revenue +5.6%.
• FY2025 operating cash flow fell to $208.7m from $239.8m despite net income rising from $163.9m to $214.3m — the cash flow statement carries a −$61.0m accounts-receivable drag (FY2024: −$12.0m) and a −$18.1m inventory drag.
• Inventory rose +76.6% YoY ($19.8m → $34.9m) on +5.6% revenue.

Catalyst does not quantify how much reported net revenue the fee reduction added. That number is not in the 10-K, the 10-Q or any earnings release. It is a real, disclosed, unquantified tailwind sitting inside the FIRDAPSE growth rate for at least the last two reported quarters. It is not fraud and it is not hidden — but it is not organic demand either, and no part of the 17–18% FIRDAPSE growth can be cleanly attributed to volume without it. The company attributes FIRDAPSE growth "primarily" to sales volume; that word is doing work it is not asked to defend.

2.6 Customer concentration

FIRDAPSE and AGAMREE are sold in the US through a single exclusive distributor to specialty pharmacies. FYCOMPA is sold through wholesalers.

Assessed against the AAOI precedent in the brief, this is materially cleaner: the concentration is disclosed with a number in every filing, the channel type is described, the counterparty is a specialty distributor rather than an unnamed private stocking entity, and the 74% → 90% rise is arithmetically explained by mix — FYCOMPA (wholesale channel) collapsed while the exclusive-distributor products grew. The DSO extension is the real flag, not the concentration — and it has a disclosed contractual cause.

2.7 Non-GAAP excludes the cost of the acquisitions that produced the growth

FY2025 non-GAAP net income of $346.2m versus GAAP $214.3m — a 61.5% uplift. The exclusions are stock-based compensation, depreciation, amortisation of intangible assets and the income tax provision. Amortisation of intangibles was $37.5m in FY2025 and $9.7m in Q1-26: that is the amortised cost of the FYCOMPA and AGAMREE rights. The adjusted measure removes the cost of the two assets that supplied 61.5% of the growth while retaining all of their revenue. Adding back the tax provision as well is aggressive by any standard. GAAP figures are used throughout this memo.

2.8 What is genuinely clean

Stated for balance, because the record should be accurate rather than uniformly damning:

• No funded debt, company-stated, at 2026-03-31. Total liabilities of $134.9m are operating.
• $755.9m of cash, up from $517.6m a year earlier, and the balance has compounded every quarter.
• Product disaggregation is disclosed by product every year, with acquisition dates and partial-period footnotes. The organic/acquired split in §2.2 is computable entirely from the company's own tables — this memo did not have to reconstruct anything. Catalyst told the market plainly what was bought.
• The FY2025 release itself says growth was driven by "organic growth of our promoted products" — a claim that is true of FIRDAPSE and AGAMREE individually and false of the 40.1% figure the screen used.
• Segment disclosure, royalty terms, patent expiries and litigation status are all disclosed at a level well above the small-cap norm.

Catalyst's disclosure is good. The screen's reading of it was not.

3. Exclusivity runway — the patent cliff question

The brief asks whether a rare-disease franchise is a patent cliff in slow motion. Product by product, from the FY2025 10-K:

FIRDAPSE (amifampridine) — 59% of Q1-26 net product revenue

• Orphan Drug Exclusivity expired 2025-11-28 (the 10-K also cites 2025-11-26 in its forward-looking-statement list; both dates appear and the discrepancy is the company's, not this memo's).
• Six Orange Book patents, latest expiring February 2037.
• Four ANDA filers, all four now settled:

The 30-month Hatch-Waxman stay expired 2026-05-26; trial against Hetero was scheduled first for 2026-03-23, then 2026-05-18. Catalyst settled instead, on the day the merger was announced, and paid $11.0m to do so.

Assessment: this is the opposite of a cliff. Orphan exclusivity has already lapsed and it did not matter, because the patent estate held and every challenger took a 2035 date. Effective US generic entry for FIRDAPSE is January 2035 — roughly nine years of runway. The hypothesis in the brief is not supported for FIRDAPSE.

One caveat that is not cosmetic: all four settlements are reverse-payment-adjacent agreements subject to FTC and DOJ review, and the Hetero settlement carries an explicit $11.0m payment. The 10-K's own risk factors ask "whether there will be a post-closing review by antitrust regulators of our previous or future Paragraph IV patent settlements." That is a tail risk on the 2035 date, disclosed by the company.

FYCOMPA (perampanel) — 9% of Q1-26 net product revenue, down from 25%

• US Patent 6,949,571: expired 2025-05-23 (including patent term extension).
• US Patent 8,772,497: expired 2026-07-01.
• FY2025 10-K, verbatim: "Following the expiration of the '571 patent and the '497 patent, three generic versions of the tablets and one generic version of the oral suspension have come onto the market."
• A settled Paragraph IV filer was free to launch from 2025-12-15.

Assessment: the cliff already happened. It is not prospective, it is not a risk, it is in the printed numbers: −17.4% in FY2025, −61.3% in Q1-2026, run-rating to ~$55m against $138m in its first year of ownership. Catalyst paid ~$198m of asset-acquisition consideration for FYCOMPA in January 2023 and the asset had lost half its revenue by Q1-2026.

This is the correct answer to the brief's trap, and it is stronger than the brief anticipated: the acquisition-driven growth was not merely inorganic, it was purchased in an asset whose exclusivity was already running out at the time of purchase. The '571 patent's May-2025 expiry was public and in the Orange Book when the deal was signed in January 2023.

AGAMREE (vamorolone) — 25% of Q1-26 net product revenue

• New Chemical Entity exclusivity to October 2028.
• Orphan Drug Exclusivity to October 2030.
• Seven Orange Book patents, earliest 2029-05-28, latest 2040-07-16.
• Patent Term Extension requested on three patents (US 8,334,279; 10,857,161; 11,833,159); FDA published the regulatory-review-period determination in the Federal Register on 2025-06-25, opening a 180-day comment window. PTE outcome not yet determined — this memo does not assume it.

Assessment: clean runway to at least 2030, plausibly to 2040 depending on the patents. But it is licensed, not owned: royalties of 7% + 7% (escalating) flow to Santhera and its upstream licensors, and rights are North America only.

Franchise-level conclusion

Weighted by Q1-26 revenue, ~91% of the portfolio has an exclusivity horizon beyond 2030. The patent cliff is not in front of this business — it is behind it, and it already took a third of the acquired growth with it. Note also that every one of the three products is licensed or acquired. Catalyst has never discovered or developed a product. FY2025 R&D was $12.7m on $589.0m of revenue — 2.2%. This is a commercialisation platform with a capital-allocation function attached, not a research company, and it should be underwritten as one: its terminal value depends entirely on continuing to buy assets at prices that work.

Management is explicit about that dependency. FY2025 10-K / release: "Catalyst reviewed and evaluated over 100 prospective clinical stage acquisition targets during 2025… However, no such agreements were entered into in 2025." Eighteen months of searching, over 100 targets, zero transactions, and a $755.9m cash balance accumulating against a business whose only growth engine is acquisition. Angelini bought it four months later.

4. Product-cycle intelligence

AGAMREE (vamorolone), DMD — the only live adoption curve in the portfolio.

Sequentially $35.3m → $36.7m (+4.0% q/q) — a launch curve entering its flattening phase, still growing 67% YoY off the easy base. Supporting evidence, dated: Health Canada approved sub-licensee KYE Pharmaceuticals' NDS (first DMD therapy approved in Canada); the SUMMIT five-year open-label registry continues enrolling; a two-part Phase 1 comparing vamorolone to prednisone and deflazacort read out 2026-06-30 (8-K Item 8.01: "Vamorolone Demonstrates On-Target Glucocorticoid Activity Without the Immunosuppression Characteristic of Traditional Corticosteroids") — the immunosuppression-dose work is aimed at positioning AGAMREE alongside approved DMD gene therapies that require concurrent corticosteroid immunosuppression. That is a genuine, named, forward indication-expansion path. It is also now Angelini's.

FIRDAPSE, LEMS — a slow, real, diagnosis-limited ramp. FY2023 +21%, FY2024 +18.4%, FY2025 +17.1%, Q1-2026 +18.1%. Remarkably stable at ~18% for three years. Named drivers with dates: FDA approval of an increased maximum daily dose (80mg → 100mg) on 2024-05-30; NCCN Clinical Practice Guidelines for Small Cell Lung Cancer updated in 2025 to include VGCC antibody screening of SCLC patients and amifampridine as supportive care — a diagnosis-rate mechanism in the largest LEMS-adjacent population; Japanese launch by sub-licensee DyDo Pharma (Jan 2025, royalty-bearing, immaterial to revenue).

FYCOMPA, epilepsy — a managed decline. No cycle. See §3.

5. Transcript mention-frequency — required core metric

Source limitation, stated plainly

Alpha Vantage's 25/day quota was exhausted after one call. EARNINGS_CALL_TRANSCRIPT returned 2024Q1 and then the rate-limit Information payload for every subsequent quarter. A one-quarter series cannot support a time-series claim, so the transcript-based metric is INDETERMINATE for this name and is recorded as such.

Substitute corpus, clearly labelled: the fifteen quarterly earnings press releases filed as 8-K Exhibit 99.1 from 2023-02-07 to 2026-05-11, pulled from EDGAR. These are the company's own prepared written communication, quarterly and complete, and they are not transcripts. Counts are normalised per 1,000 words (word counts range 1,269 to 3,420 — the ISRG length-artifact failure mode is real and is controlled for here). There is no prepared-remarks / Q&A split available in this corpus; that decomposition is not claimed.

The series (mentions per 1,000 words)

Column labels are the reporting period of each release; the 2023-02-07 preliminary release is shown first.

Open questions generated, before interpretation

Recorded in the order the method requires — anomalies first, explanations after.

1. royalt* emerges from nothing. Zero across the first nine releases, then 2.3 and 4.8 in the two most recent — a 0 → 4.8 emergence, the sharpest move in the table.
2. promoted product is a first-ever appearance in 2026-02-25, after fourteen releases at zero.
3. generic emerges from zero in FY2024 and peaks at 2.2 in 25Q2.
4. FYCOMPA decays monotonically 11.0 → 1.1, an order of magnitude.
5. lifecycle first appears 2026-02-25 at 0.3.

Resolution against the independent record

1. royalt* → confirmed and material. Resolved in §1: the FIRDAPSE upstream royalty expired January 2026, cutting the blended rate from a maximum 18.5% to 6.0%. The language moved one to two quarters before the COGS line did — the 2026-02-25 release (2.3) precedes the Q1-26 gross-margin step from 87.3% to 90.3%. This is the method working exactly as designed, and it surfaced the memo's genuine mechanism.
2. promoted product → confirmed and adverse. Resolved in §2.4: a new, more flattering revenue denominator introduced in the quarter the reported one decelerated.
3. generic → confirmed. Tracks the FYCOMPA '571 patent expiry (2025-05-23) and the settled filer's 2025-12-15 entry date. Rising mentions preceded the −61.3% print by three quarters.
4. FYCOMPA decay → confirmed, and it is de-emphasis of a real decline, not a disclosure choice: the product line remains in the disaggregation table every quarter.
5. lifecycle → weak, single observation, not pursued. Recorded so the record is complete, with no claim attached.

Provenance note per the reference: hypotheses 1 and 2 were generated by this process, not confirmed by it. Neither the FIRDAPSE royalty expiry nor the "promoted products" redefinition was in view before the counts were run. Both survived checking against the primary record.

6. Screen input validation — verified against primary filings

Per the brief, every material input was re-derived. Results:

Scale cross-check (net income ÷ shares ≈ filed EPS), per the brief: FY2025 net income $214,326k ÷ basic WASO 122,290,866 = $1.7526 vs filed basic EPS $1.75 ✓ FY2025 net income $214,326k ÷ diluted WASO 127,257,929 = $1.6842 vs filed diluted EPS $1.68 ✓ Share count and income scale are internally consistent. No 4.1x or 54% error of the kind found elsewhere.

The one genuine share-count discrepancy — reported, not silently adopted

The screen uses 122,380,992 basic shares outstanding. The transaction was struck on a fully-diluted basis:

"$31.50 USD per share in cash, for a total equity value of approximately 4.1 billion USD"

$4,100m ÷ $31.50 = ~130.2m fully-diluted shares — reflecting options and RSUs that vested and cash-settled at the effective time (8-K 2026-07-16). Against 122.4m basic, that is ~7.8m shares (6.4%) of dilution the screen does not carry.

Neither number is adopted silently. The screen's EV understates the enterprise value actually transacted by $246m (7.9%). Both are carried through the Valuation document.

7. Criteria scoring

No position verdict is issued. Types are as assigned in criteria.md for the long-only absolute-return strategy; the memo blocks on none of them.

8. Downside Criteria — the permanent-loss case with a named cause

Type: MEASURED. Logged and scored; it rejects nothing.

Realistic permanent-loss scenario, and its cause. Underwritten as at the last quarter for which the business was a going public concern, i.e. ignoring the merger:

Named cause: failure of the acquisition engine, with FIRDAPSE as the only load-bearing asset.

Catalyst has no discovery capability (R&D = 2.2% of revenue) and every product is licensed or acquired. Its growth is a function of doing deals. In 2025 it reviewed over 100 targets and did none. FYCOMPA demonstrated that its deal underwriting can be wrong: ~$198m paid in January 2023 for an asset whose primary Orange Book patent expired May 2025 and whose revenue halved within three years of purchase. If no further asset is acquired, the business converges to FIRDAPSE (+18%, decelerating with diagnosis-rate saturation) plus AGAMREE (a launch curve flattening, royalty-burdened at 14%+ above $100m of sales) — a mid-single-digit grower with a January 2035 patent wall and $755m of undeployed cash. On its own historical p25 EV/EBIT of 8.61x applied to guided FY26 EBIT, that is ~$25.6–26.5/share.

Estimated permanent impairment from the pre-deal unaffected price of $25.94: ~0% to −20%, the low end requiring a further de-rating toward the bottom decile of its own history (EV/EBIT 7.34x → ~$22.6/share, −13%). The severe case requires a second FYCOMPA-style acquisition writedown, which is the specific, demonstrated failure mode.

Probability assigned: 30% for a −13% or worse outcome over 24 months from the unaffected price. This is not a going-concern case and is not flagged as one. Zero funded debt, $755.9m of cash and $208.7m of annual operating cash flow make insolvency implausible on any horizon considered here.

Realised outcome, for the ledger. The scenario did not get to run. Shareholders received $31.50 cash on 2026-07-15, a +21.4% return from the unaffected close of $25.94 on 2026-04-22. The bear case is logged for Brier scoring but is unresolvable — the path was truncated by an acquisition, and it should be recorded as unresolvable_acquired, not as a miss. Scoring it as a loss would corrupt the calibration record.

9. Unsupported, missing or unverifiable

Stated per the brief's rule: omit and say what is missing rather than fabricate.

1. Consensus estimates: not obtained. Alpha Vantage quota exhausted after one call. No NTM revenue or EPS consensus, no analyst count, no revision history. The 12-month target in CPRX_Valuation.md is built on company guidance instead, which is disclosed there.
2. Earnings-call transcripts: one quarter only (2024Q1). The mention-frequency series in §5 is built on quarterly earnings press releases, is labelled as such throughout, and the transcript metric is INDETERMINATE. No prepared-remarks / Q&A split is claimed.
3. Peer multiple spread: not computed. HRMY is named as the correct comparator; no figure is asserted.
4. The revenue benefit of the distributor fee renegotiation is not quantified by the company and is not estimated here. It is a real, disclosed, unsized tailwind inside FY2025 H2 and Q1-2026 reported net revenue.
5. The AGAMREE Patent Term Extension is pending, not granted. The 2040 patent horizon is not assumed.
6. FIRDAPSE ODE expiry date is internally inconsistent in the FY2025 10-K — 2025-11-26 in the forward-looking-statements list, 2025-11-28 in the litigation note. Both are reported; neither is resolved.
7. No options chain was pulled. The Liquidity Criteria requires the actual chain before proposing any structure; no structure is proposed because the underlying does not trade. This is a genuine absence, not an omission.
8. No 12-month forward operating margin was independently modelled. The no-deal counterfactual target holds the TTM operating margin (44.8%) flat, which is stated at the point of use and is an assumption, not a forecast.
9. The own_multiples history uses IEX-feed daily closes (SIP bars returned HTTP 403 on this entitlement). IEX closes can differ marginally from consolidated closes; the latest-trade and snapshot endpoints, which are SIP, corroborate the final price to the cent.
10. Angelini Pharma is private. No post-close information on the asset is or will be available. This name is permanently uncoverable from public sources after 2026-07-24.

10. What this name should change in the process

Filed as calibration evidence, per the standing instruction that defects are the primary output.

1. A screen must verify that the security still trades. Two checks, one API call each: (a) latest trade timestamp within N sessions of the screen date; (b) absence of Form 25 / Form 15 in the recent filing index. Either alone would have rejected this record. This is the highest-value control this batch produced, because it is cheap, universal and currently absent.
2. data_quality_ok: false must block ranking, not annotate it. The screen set this flag on CPRX and then ranked it at +30.8pp anyway. A quality flag that does not gate is decoration.
3. Trailing volatility must not be used for sizing across a structural break. 32.1% trailing versus 1.2% post-announcement is a 27x error in the sizing input, and it errs toward more size. A regime test — e.g. ratio of last-quarter to trailing-year realised vol — should force manual review.
4. "Demonstrated CAGR" must be decomposed before it becomes a hurdle reference. Product-level revenue disaggregation is a required XBRL/filings disclosure for multi-product issuers, and 61.5% of this name's "demonstrated" growth was acquired. A CAGR computed across an acquisition is not a demonstration of anything about the business. Proposed rule: where revenue growth spans a business combination or in-licensing, the demonstrated rate must be recomputed on a same-product basis, or declared UNIDENTIFIED.
5. The exit multiple must be sanity-checked against the name's own history. The screen anchored at 23.4x "GROWTH_MATCHED" from n=188 peers. CPRX's own six-year EV/EBIT history has a median of 11.26x, a p90 of 19.38x, and an all-time maximum of 28.64x. 23.4x sits above the 90th percentile of everything this stock ever traded at — and an informed strategic acquirer, with full diligence and control, paid 12.49x. Anchoring above a name's own p90 requires an argument; here there was none. This is the single largest contributor to the spurious +30.8pp margin.

Sources: SEC EDGAR (CIK 0001369568) — FY2022/2023/2024/2025 10-K, Q1-2026 10-Q, 8-K filings of 2026-05-07, 2026-05-11, 2026-06-23, 2026-06-30, 2026-07-08, 2026-07-16, Form 25-NSE 2026-07-15, Form 15-12G 2026-07-24, and the 8-K Ex-99.1 earnings-release corpus 2023-02-07 → 2026-05-11. Prices: Alpaca Markets (IEX daily bars; SIP latest trade/quote/snapshot). Reverse DCF: investment-memo/assets/reverse_dcf.py. Every figure in this document is traceable to one of these; none is estimated except where explicitly labelled.