A friend asked me a while back what I think of Novo Nordisk (NVO). The honest answer is that I don't know.
I made NVO my #2 pick in the March 2025 edition of this newsletter, at $90.65 with a $134 price target, about 48% upside. My thesis was that the stock was in a temporary pullback after underwhelming CagriSema trial results and conservative 2025 guidance, and that it was a better value than Eli Lilly at the time.
That call has not aged well.
The stock is now down more than half from the price I quoted, which puts it near the top of my loser list. I've already used it as a case study in straying outside my circle of competence.
But my friend’s ask made me think of a better question: is there a way to value a pharmaceutical company without pretending to be a pharmaceutical expert?
Why my valuation method breaks on pharma
My circle of competence is tech, oil and gold. I like businesses where I can judge how durable the cash flows are and then put a price on them.
Pharma breaks that.
Most of a drug company's value depends on things I can't judge: which platforms are working, which late-stage trials will read out well, how fast a patent cliff will bite, and whether the research team is any good.
The usual answer is to go learn the industry. That takes years, and I'd still be guessing at the parts the specialists get wrong.
So I tried something that sounds a little crazy: build a valuation that doesn't require me to know any of that. I thought it was crazy too. But it made a fun thought exercise, and it forces you to separate what you can know from what you can only guess.
Here’s my take.
The framework: three layers of value, three levels of risk
The idea is to split a pharma company's value into three layers, ordered from most knowable to least.
Layer | Question it answers | Risk | What it rests on |
|---|---|---|---|
Dead R&D | What is the company worth if it never discovers another drug? | Lowest | Current sales, patent expiry dates, generic erosion |
Current Pipeline | What is the pipeline worth? | Medium | Historical odds of success at each trial phase |
Future R&D | What is the ability to keep inventing worth? | Highest | Industry-average returns on R&D |
A: Dead R&D. Pretend the company can't spend another dollar on new technology. Its value comes entirely from the drugs already on the market, and we know how that income decays: patents expire, generics arrive, and prices fall hard. This is the layer with the most real data behind it.
B: Current Pipeline. Now value the drugs currently in the development pipeline. Predicting any single trial is a crapshoot, but the averages are well documented. Of the programs that enter Phase 1, only about 8% ever reach approval, and the biggest cut comes at Phase 2.
Stage | Chance of moving to the next stage |
|---|---|
Phase 1 to Phase 2 | 52.0% |
Phase 2 to Phase 3 | 28.9% |
Phase 3 to filing | 57.8% |
Filing to approval | 90.6% |
Phase 1 to approval | 7.9% |
Source: BIO / Informa / QLS, 2011-2020. These are industry-wide averages, so they say nothing about any one company's drugs, but they beat guessing.
Turning the odds into dollars. The odds only tell you how likely a drug is to make it, so you still need to know what it's worth if it does. For each drug in the pipeline, size the market it would sell into using markets that already exist, such as what current drugs for that disease sell today, and assume a modest share for a new entrant. Multiply that potential by the chance the drug reaches market from its current stage, then discount it for the years until launch. Add those up across the pipeline and you have Layer B. It works best on big, diversified pipelines, where hits and misses average out, and worst when the pipeline is really one big bet.
C: Future R&D. This is the value of the company's talent, its platforms and its ability to recruit. It's the riskiest layer and the one where my lack of expertise hurts most. So I cheat: I swap in long-run industry data and assume the company is average. That's an outrageous assumption. But depending on your discount rate, this layer is usually the smallest piece of the total, so being wrong about it costs less than being wrong about the other two.
The cool part about this approach is that it’s built to avoid double counting, so you can add each layer to get the total value of the business.
Turning it into a signal
For any company, compare today's price to each cumulative part of the business. For shorthand, let’s take:
Dead R&D and rename it “A”
Current Pipeline can be called “B”
Future R&D can be called “C”.
Using this notation, you can set up ratios versus current price to see your margin of safety. For example (not hard-and-fast rules):
Price / A < 1: you're paying less than the existing drugs are worth. That's bargain territory, and the full pipeline and future R&D are free upside.
Price / (A+B) ~ 0.8: you're getting the existing drugs and the pipeline at a 20% discount. That would be an interesting buy.
Price / (A+B+C) > 1: the market is paying for more than an average R&D machine. Ask yourself where you really think price appreciation will come from.
Building the workflow
The trend can matter more than any single reading. So while running this by hand on one company is a fun exercise, running it every week across a whole watchlist is not.
I'm building an AI workflow to do it for me. The goal is a workflow that updates A, A+B and A+B+C for each company and tracks the three ratios against the price over time.
NVO is my first test case. I ran a first pass, but the inputs need some tweaking before I'd trust them, so I'm hoping to share the NVO results next week. So I still don’t have an answer for my friend’s question, but I’m working on it 😁
If I get the workflow up and running, let me know if you'd like a weekly report.
Sources:
