Phase Space AI

Trade Construction

Halozyme Therapeutics [HALO]

Halozyme Therapeutics [HALO] — Trade Construction & Risk

2026-07-29 · framework: Criteria, 2026-07-29

This document issues no position verdict. It states what a position would have to look like if a book chose to take one, what would invalidate the analysis, and what the actual tradable instruments are.

1. Vehicle — the chain was pulled, not assumed

criteria.md, Liquidity Criteria: "Any proposed options structure requires the actual chain pulled first — open interest and quoted size for the specific strikes and expiry. On HCA the maximum open interest across an entire March-2027 chain was 18 contracts; the default defined-risk spread was uninvestable at any size and nothing tested for it. A vehicle that cannot be filled is not a vehicle."

HALO reproduces the HCA failure almost exactly.

Common stock — viable
63-day median volume 1,751,574 shares
63-day median dollar volume $128.2m
A $10m position 7.8% of one day's volume
A $25m position 19.5% of one day's volume
20-day realised volatility 30%

A $10m position exits in under a day at 20% of ADV. A $25m position is a two-to-three-day exit in a normal tape and materially longer in a stressed one. PASS at $10m; sized attention required above ~$20m.

Options — FAIL. Not investable at any size worth taking.

March 2027 calls (233 days), spot $83.62 — the only listed expiry beyond twelve months:

Strike Open interest Bid Ask Spread as % of mid IV Delta
70 2 17.38 20.98 18.8% 39.8% 0.79
75 10 14.11 17.66 22.3% 39.1% 0.72
80 2 11.63 15.30 27.3% 40.5% 0.65
100 8 3.25 6.63 68.4% 36.3% 0.35
105 4 1.88 5.62 99.7% 35.9% 0.29
110 2 0.58 4.61 155.3% 34.5% 0.22
Total across the entire $70–110 range 28 contracts

Total open interest across the whole usable March-2027 strike range is twenty-eight contracts, with a maximum of ten at any single strike. The HCA precedent was eighteen. This is the same defect and it is being reported rather than papered over.

December 2026 calls (142 days) — better, but still not usable:

Strike Open interest Bid Ask Bid size Ask size Spread % of mid
80 423 9.89 12.52 1 110 23.5%
85 654 6.37 9.95 238 164 43.9%
90 380 4.06 7.45 190 127 58.8%
100 215 1.23 4.78 154 117 118.1%
115 283 0.00 2.22 0 85 no bid
Total 2,514

Quoted size in December-2026 is real (100–240 contracts). The problem is the spread: 44% of mid at the money, 59% one strike out, and no bid at all at $110 and $115. Round-tripping an 85/100 call spread costs roughly 60–80% of the premium in friction.

Conclusion: no options structure in HALO is investable. If a position is taken, the vehicle is the common stock. No options recommendation is appended to trade_recommendations.jsonl, because there is nothing fillable to append.

Note on implied volatility. Where a two-sided quote exists, IV is 34–41% against 30% realised. Long premium is expensive here too, which is a second and independent reason to prefer the common.

2. Sizing inputs

Input Value
20-day realised volatility 30%
Scanner volatility (screen) 31%
Volatility tier Moderate
Beta not regressed — no figure asserted
Leverage $2,179.6m convertible principal; net debt $1,823.8m; book equity $48.8m at 2025-12-31
Sub-sector for concentration Pharma (drug-delivery platform / royalty)

A concentration point the book should note explicitly. HALO's revenue is a derivative of its partners' product sales. Its economically relevant correlation set is J&J, Roche, argenx, Takeda and BMS, not the biopharma peer group. A book already long argenx or J&J is more exposed to HALO's revenue driver than a correlation matrix computed on equity returns would suggest, because 41% of HALO's revenue is one partner's product.

Inverse-volatility sizing at 30% realised sizes HALO larger than ALNY at 42%. That is the correct mechanical output and it is worth flagging that it runs opposite to the qualitative risk ranking: HALO's five-year terminal value depends on an undisclosed royalty rate and a pending patent-validity proceeding, which is a fatter left tail than the volatility implies. criteria.md states the interim control is inverse-volatility sizing precisely because "a fat-left-tail name is almost always a high-volatility name." HALO is the counter-example: a moderate-volatility name with a genuinely fat left tail. Recorded here so the book can override rather than discover it later.

3. Entry timing — what Momentum Criteria says, and what it does not

Momentum is MEASURED. It governs when, never whether.

Metric Value
12-1 momentum +47.8%
12-month price change +41.6%
% of 52-week high 100.0%
RSI-14 69.2
20-day realised vol 30%
12-1 cross-sectional percentile INDETERMINATE — no HALO record in the universe scan

INDETERMINATE is not FAIL. The absolute figures are strong; the cross-sectional read that criteria.md actually requires is unavailable and is declared unavailable rather than substituted with the absolute number.

Read. The stock is at a 52-week high with RSI-14 at 69.2, and simultaneously at the 19th percentile of its own EV/sales range. Price has risen 42% while the multiple has fallen, because revenue rose faster. That is the signature of a name where the fundamentals are outrunning the re-rating.

Timing implication, not a veto. Momentum favours entry; RSI near 70 argues against paying up in a single clip. A book scaling in has a dated checkpoint: Q2-2026 results, expected early August 2026 (not confirmed — see the calendar).

4. Invalidation — dated, falsifiable, specific

# Invalidation condition How it is observed When
1 The magnitude of the post-2027 rHuPH20 royalty rate reduction is disclosed and is severe. This is the number the whole terminal value turns on and it has never been published. FY2026 10-K; 2027 guidance; any 8-K most likely Jan–Feb 2027, as the 2027 US expiry enters the guidance window
2 An adverse USPTO post-grant review decision on the MDASE patents. Removes the Merck optionality and the 2032–2034 estate at once. USPTO PTAB decision; 8-K no date disclosed
3 The German preliminary injunction against Keytruda SC is vacated on appeal, or the D.N.J. case is dismissed. 8-K; court docket no date disclosed
4 FY2026 total revenue lands below the guided $1,710m. Q1 delivered $376.7m, about 21% of the midpoint — the ramp is second-half weighted and untested. Q2, Q3, Q4-2026 earnings quarterly through Feb-2027
5 Bulk rHuPH20 sales decline sequentially for two consecutive quarters. This is the forward indicator of partner manufacturing demand; it grew 172% year on year in Q1-2026. A reversal would lead a royalty slowdown by two to three quarters. quarterly revenue disaggregation footnote quarterly
6 A partner terminates. Agreements are terminable "for any reason … generally upon 90 days prior written notice." With Partner A at 41% of revenue, this is a live contractual risk, not a theoretical one. 8-K any time
7 The $400m 2026 buyback is not executed. It is the most reliable component of the 12-month EPS bridge, and it is projected rather than committed. quarterly cash-flow statement and share count quarterly

What is explicitly not an invalidation. A drawdown; a quarter where milestone revenue is zero (it was zero in Q1-2026 and revenue still grew 42%); or a GAAP net loss caused by intangible amortisation, of which $118.0m per year is scheduled for 2026, 2027 and 2028.

5. Downside Criteria — logged for the ledger

Field Value
Bear case USPTO post-grant review invalidates the MDASE estate; the German injunction against Keytruda SC is lifted; competing recombinant hyaluronidases (Alteogen is named in the 10-K's competition section) enter across the partner base; and the DARZALEX royalty rate step-down from the 2027 US / 2029 EU rHuPH20 expiry lands at the harsher end. Royalties plateau near $1.2bn instead of compounding, and the market reprices the stream from a growth multiple to a run-off multiple of roughly 4.0–4.5x sales.
Named cause Patent expiry and patent-validity loss on a royalty stream whose duration is contractually tied to those patents, with the largest counterparty's own patents explicitly not extending it
Probability 15%
Drawdown if it occurs −60%, to approximately $34 (EV ≈ $1.4bn × 4.25 = $5,950m, less $1,824m net debt, ÷ 122.9m shares)
Secondary bear Partner A (41% of revenue) terminates or materially renegotiates on 90 days' notice. Probability 5%, drawdown −35% to −45%
Tertiary bear FY2026 guidance miss on second-half-weighted revenue. Probability 20%, drawdown −15% to −20%, recoverable
Upside optionality not in the target A HALO win against Merck converts SC Keytruda into a royalty-bearing product. No value for this is included anywhere in the 12-month target or the implied-path test.
Going concern No. FY2025 operating cash flow $651.6m, free cash flow $644.6m, FY2026 adjusted EBITDA guided $1,125–1,205m. This is a de-rating case, not a solvency case, and it is described as such.

Calibration note, given the principal's recorded objection that bear cases are often used "without much backbone or rigorousness … to exclude good investments": the bear case here is not used to exclude anything — the Valuation Criteria passes with a +15.0pp margin. It is recorded because the flip point is identified and quantified (8.0x EBIT, a 4.0x exit sales multiple, a 54% de-rating), which is a falsifiable statement a book can disagree with, rather than a paragraph of caution.

6. Position construction, if a book takes one

Element Specification
Direction of the research Positive — Quality PASS (COMPOUNDER, ~23% ROIC with an evidenced redeployment mechanism), Valuation PASS with a +15.0pp margin
Vehicle Common stock only. Options FAIL the Liquidity Criteria outright (§1)
Entry Momentum favours entry; RSI-14 at 69.2 argues against a single clip. Scale, with Q2-2026 results as the checkpoint
Sizing Inverse-volatility on 30% realised — moderate tier. Note the explicit override flag in §2: the volatility understates the left tail on this name
12-month target $107.5 (+28.6%); bear $86 (+3.1%); bull $126 (+50.6%)
Invalidation level (price) None asserted as a stop. Invalidations in §4 are evidentiary
Hard evidentiary invalidation A disclosed severe royalty step-down, or an adverse PGR decision
Ledger entry No options recommendation appended to trade_recommendations.jsonl — nothing fillable exists. The bear case above is logged for Brier scoring

7. What a book should watch that this memo cannot resolve

  1. The royalty step-down magnitude. Undisclosed. This is not a modelling gap that more work would close; the number does not exist in the public record. A book underwriting HALO's terminal value is underwriting an unpublished contractual term. That should be stated at the investment committee, not hidden inside a WACC.
  2. The USPTO post-grant review docket. HALO discloses that PGRs have been instituted; it does not disclose which patents, the schedule, or the claims at issue. PTAB filings would resolve this and were outside this run's scope.
  3. Consensus. No Street numbers obtainable (Alpha Vantage quota exhausted by parallel agents). If the Street is already at $2.0bn for FY2027, the 12-month target here is a consensus target, not a variant view. Unknown, not guessed.
  4. Partner-level royalty rates. Disclosed only as "an on average mid-single digit percent rate." The rate on DARZALEX specifically — 34.6% of revenue — is not public.
  5. Correlation to the partners. HALO's revenue driver is J&J, Roche, argenx, Takeda and BMS product sales. portfolio-book should treat exposure to those names as correlated exposure to HALO.