A DCF asks what a business will earn. A biotech valuation asks a different question first: whether the product will exist at all. Most of a development-stage company’s value sits in programmes that have not been approved, may never be, and cost money every year until the answer arrives. The method that handles this is the risk-adjusted net present value, rNPV, applied programme by programme and summed. The mechanics are a DCF with two extra ideas, probability and the launch curve, and both matter more than the discount rate.
Why a DCF is not enough
A conventional DCF forecasts revenue with a growth rate and discounts it at a cost of capital. For a company selling a product, that is a reasonable description of the uncertainty: the business exists, the question is how fast it grows. For a company whose lead asset is in Phase 2 it is not. The revenue either arrives at something like the forecast or it never arrives, and the probability of each branch is the single most important number in the analysis.
Raising the discount rate to compensate is the common shortcut and the wrong one. A 20% rate says the cash flows are risky in every year, forever; the real risk is a binary event in year three. Once the drug is approved the risk falls sharply, and a valuation that carried the trial risk in the rate has to change its rate on the day of the readout, which tells you the rate was never the right place for it.
The rNPV puts the probability where it belongs: on the cash flows that depend on the event, and only on those.
Probability of success, stage by stage
Every clinical stage has a published base rate of passing: roughly 60% from Phase 1 into Phase 2, around a third from Phase 2 into Phase 3, around 60% from Phase 3 to filing, and around 90% from filing to approval, with wide variation by therapeutic area. Oncology runs below those numbers; some rare-disease and validated-mechanism programmes run above them.
The probability that a programme reaches the market from where it stands is the product of the stages still ahead of it. A Phase 3 asset has two hurdles left and a probability of technical and regulatory success near 55%. A Phase 1 asset has four and a probability near 11%. That gap, not the discount rate, is why an early asset is worth a fraction of a late one with the same peak sales.
The base rates are the starting point, not the answer. An analyst adjusts them for the mechanism (validated or novel), the endpoint (objective or subjective), the size and design of the trial, and the quality of the data so far. The adjustment is where the analytical work is, and it should be written down beside the number.
- Probability to market = product of the remaining stage probabilities.
- Adjust base rates for area, mechanism, endpoint and data; state why.
- Carry stage risk in the probability, not in the discount rate. Never in both.
Which cash flows get weighted
The rule that separates a careful rNPV from a careless one: cash flows inside the current stage are not risk-adjusted, and everything after it is. The company is running its Phase 2 trial now and paying for it whether or not the drug works, so those costs carry a weight of one. The Phase 3 trial, the launch build, the revenue and the cost of goods happen only if Phase 2 succeeds, so they carry the probability.
The mistake this prevents is subtle and common: risk-adjusting the revenue but not the later trials. A model that does that has a failed Phase 2 paying for a Phase 3 that would never have been run, which makes early programmes look worse than they are and hides the real shape of the bet.
Corporate costs, the unallocated cost of being a company, are not risk-adjusted at all. They are paid whatever the pipeline does, which is why they sit outside the programmes in the sum of the parts.
The launch curve: patients times price
A drug’s revenue is not a growth rate. It is the number of patients treated times the net price per patient, and both are built from parts that can be checked against the epidemiology and the market.
Patients: the addressable population with the indication, the share diagnosed and actually treated, and the share of those the drug captures, which ramps from launch to a peak over three to five years as prescribers adopt it and access is negotiated. Price: the gross list price, less the gross-to-net discount to payers and intermediaries, which in the US commonly runs from 20% to 50% and grows over a product’s life.
Peak sales is what falls out when penetration reaches its peak, and it is worth rebuilding from these parts every time someone quotes it. Then exclusivity ends. Loss of exclusivity, when patents and regulatory protection expire, cuts revenue to a fraction within a year or two for a small molecule and more slowly for a biologic, and moving that year by two in either direction changes a programme’s value more than most other inputs.
| Line | Figure | What it comes from |
|---|---|---|
| Addressable patients | 120,000 | Epidemiology for the indication |
| Diagnosed and treated | 60% | Treatment rates, guidelines, access |
| Peak penetration | 30% | Competitive position, label, physician surveys |
| Patients at peak | 21,600 | The product of the three lines above |
| Net price per patient a year | $45,500 | $65,000 list less a 30% gross-to-net discount |
| Peak net revenue | $983m | Patients times net price |
From programme to company: the sum of the parts
Each programme’s risk-adjusted cash flows are discounted and summed to its rNPV. The company is the sum of those rNPVs, less the present value of the corporate costs that every programme has to pay for, plus the cash on the balance sheet and less any debt. Divide by the diluted share count, including the options and warrants a biotech always carries, and set it against the price.
Read the sum for two things. Which programme carries the value: if the lead asset is 90% of the enterprise value, the company is a single trial with a pipeline attached, and should be sized as one. And how much of the value is probability: the unrisked NPV of the programmes less their rNPV. For a company with early assets that gap is most of the value, which is why one readout can move the share price by more than any change to any other input.
One discount rate for the whole company is the cleaner convention, because the stage risk is already in the probability. Analysts who use a higher rate for earlier programmes are counting the same risk twice; if you do it, say so and show the sensitivity.
Where biotech models go wrong
Gross price used as net, which overstates revenue by the whole gross-to-net margin. Exclusivity loss ignored or placed late, which can double a programme. Revenue grown at a rate rather than built from patients and price, so nobody can check it against the epidemiology. Later-stage costs left unweighted, so failed trials pay for trials that never happen. Corporate costs left out of the sum, so the company is worth more than its parts. Basic shares instead of diluted.
And the one that matters most in an interview: a probability quoted as the next stage’s chance of success rather than the chance of reaching the market. A Phase 2 asset does not have a 35% chance of being a drug; it has a 35% chance of reaching Phase 3, and something under 20% of being approved.
Frequently asked questions
What is rNPV in biotech valuation?
The net present value of a drug programme’s cash flows after each year’s flow has been weighted by the probability that the programme reaches that point. Cash flows inside the current clinical stage carry a weight of one, because the money is being spent regardless; everything after it, later trials, the launch and the revenue, carries the probability of getting there. Programmes are valued one at a time and summed.
How do you calculate the probability of success for a drug?
Multiply the probabilities of passing each remaining stage. Base rates by phase are published (roughly 60% for Phase 1, a third for Phase 2, 60% for Phase 3 and 90% for approval, varying by therapeutic area), then adjusted for the mechanism, the endpoint, the trial design and the data so far. A Phase 2 asset therefore has a probability of reaching the market near 20%, not the 35% chance of passing its current trial.
What is a launch curve?
The path of a drug’s revenue from launch to peak and on to loss of exclusivity, built from patients times price: the addressable population, the share diagnosed and treated, a penetration that ramps to its peak over several years, and a net price after the gross-to-net discount. Peak sales falls out of those inputs; the year exclusivity ends decides how long they last.
Why not just use a higher discount rate for a risky biotech?
Because the risk is a binary event in a specific year, not a constant drag on every cash flow forever. A high rate misstates both the timing and the size of the risk, and it has to change on the day of the readout, which shows it was never the right place for it. The rNPV carries the trial risk in the probability and leaves the discount rate to do the job it does everywhere else.
Related guides
Build the model, programme by programme.
The biotech rNPV model in the Models library carries three programmes at three stages, a launch curve from patients and price, stage-gated probabilities and the sum of the parts, every figure a live formula you can inspect on the page.