Oncology Platforms

The question

Are big-cap pharma oncology franchises priced for a successful succession off the 2028-2030 Keytruda cliff, or for a near-total revenue gap with no offset from the subcutaneous reformulation, combination label extensions, or the successor IO and ADC pipelines?

Base rates

Healthcare has trailed SPY materially over the trailing twelve months, and large-cap pharma has been the worst-performing sector cohort within healthcare. The gap is the opportunity, not the warning. The reference class is prior blockbuster patent cliffs where the incumbent franchise holder owned the successor molecule or reformulation: Lipitor to Crestor-era statin succession, Humira to IL-23 and IL-17 class succession, and Gleevec to second-generation TKIs. In each case, the incumbent lost thirty to fifty percent of peak revenue on the core molecule and retained sixty to eighty percent of peak franchise economics through reformulation, combination, and successor label.

The relevant conditional probability is that a sponsor with an approved anti-PD-1 backbone and an active ADC or bispecific combination program will generate at least one successor approval within the thirty-six months surrounding loss of exclusivity. Across the last two decades of oncology data, that conditional probability has been roughly sixty-five to seventy-five percent. The consensus discount embedded in Merck's current multiple implies a probability closer to twenty-five percent.

For the basket, median forward P/E sits roughly four turns below the fifteen-year median, dividend coverage is intact, and free-cash-flow conversion remains above eighty percent across all seven large-cap holdings. The base rate for a diversified pharma basket at a four-turn discount with intact coverage and visible pipeline is a low-teens three-year annualized total return, versus high-single-digits for the market.

Why consensus is wrong

The sell-side oncology model is built around single-asset succession probabilities applied independently. That structure understates the franchise economics in two ways.

First, Keytruda's subcutaneous reformulation, if approved on the current timeline, retains a meaningful share of the injectable franchise through post-2028 formulation exclusivity. The market is modeling a binary cliff; the actual outcome is a staircase. Staircases are worth substantially more than cliffs in a DCF, and the re-rating when the timeline becomes visible to the sell-side is mechanical.

Second, combination-with-successor labels are a well-established path to extend franchise economics. AstraZeneca's Tagrisso-plus-chemo, Merck's Keytruda-plus-Lynparza, and Bristol's Opdualag all demonstrate that the incumbent franchise holder has structural advantages in combination development: real-world data on the backbone, established prescriber relationships, and commercial infrastructure. The probability of at least one major combination approval extending franchise economics past 2030 is substantially higher than the probability applied to any individual successor asset.

Third, the ADC and bispecific pipelines at AstraZeneca, Johnson & Johnson, and Merck have advanced far enough that several readouts in 2026-2027 will force a re-rating regardless of Keytruda timing. The market has priced those readouts as low-probability binary events; the clinical data support a higher base rate for at least one approval across the cohort.

Position construction

The book has three twenty-percent anchors and two sub-books.

Anchors. Johnson & Johnson at twenty percent captures the broadest oncology franchise exposure across Darzalex, Carvykti, and the emerging bispecific pipeline, with the consumer-health separation providing a cleaner pharma-only valuation look. AstraZeneca at twenty percent is the purest ADC play at scale -- Enhertu, Datroway, and the broader Daiichi partnership produce the strongest pipeline depth in the cohort. Novartis at twenty percent owns Pluvicto, Kisqali, and a focused radiopharmaceutical build-out that is still under-appreciated by the market.

Core large-cap, roughly twenty-five percent. Merck at ~17.0% is the central Keytruda-succession position; Bristol-Myers Squibb at ~7.1% adds Opdivo, Breyanzi, and a cardiovascular-adjacent pipeline at the steepest discount in the basket.

Diversifiers and specialty, roughly fifteen percent. Vertex at ~8.6% is the cystic-fibrosis monopoly and emerging pain and nephrology expansion, uncorrelated to IO dynamics. Takeda at ~3.6% adds a hematology and rare-disease franchise with a dividend floor. Neurocrine at ~1.1%, Jazz at ~0.7%, Exelixis at ~0.9%, and Roivant at ~1.0% are specialty-pharma positions with their own catalysts. Denali at ~0.2% is a small asymmetric exposure to neurodegeneration.

Asymmetric payoff

If Keytruda subcutaneous is approved on timeline and at least one ADC or combination label expansion clears in the basket over the next eighteen months, the book returns roughly 15-24% annualized over three years as the multiple re-rates toward the fifteen-year median. If the Keytruda cliff plays out closer to consensus and two or more pipeline assets fail, the book returns roughly -2% to +4% with dividend income absorbing most of the multiple pressure. If radiopharma or ADC data read through to a category-wide re-rating, the right tail is 28-38%.

At a 55% base, 25% bear, 20% bull weighting, expected value is roughly +12 to +16% annualized against an SPY base rate near +8%. The asymmetry comes from the fact that the current multiple already assumes the bear case.

Three things that would change our mind

1. The FDA declining to approve the Keytruda subcutaneous reformulation or issuing a narrow label that caps pricing parity with the IV form, which would convert the staircase back into a cliff.
2. Two or more top-ten ADC or bispecific pipeline readouts in the basket reading out below their Phase 2 efficacy benchmarks by more than thirty percent on the primary endpoint, indicating the successor probability we are betting on is materially below the historical base rate.
3. A durable change in IRA-driven pricing rules that extends negotiation to the first nine years of branded exclusivity rather than the current window, which would compress franchise economics on every molecule in the book simultaneously.

What we're explicitly NOT betting on

We are not betting on a single molecule approval or a single readout. We are not betting on a specific M&A target being acquired at a specific premium. We are not betting on cell and gene therapy economics improving on the current trajectory -- that is a separate idea with different reference classes. We are not betting on small-cap biotech re-rating broadly; the XBI-cohort recovery is not the claim. The thesis requires only that the implied successor probability embedded in large-cap pharma multiples compresses toward the long-run reference class. That is a narrower claim than picking winners, and the book is sized for it.