Vose SoftwareModelRisk Optimizer example · Pharmaceuticals and biotech
R&D Portfolio Chosen by the Optimizer
Which projects to fund for the most value, with limits on budget risk and on the chance of no launch?
From the model’s saved run
The Optimizer’s plan against the starting plan
The two portfolios, and how each performs on the same simulated futures (US$m).
S-curve (cumulative probability) of Starting plan, Optimizer’s plan from the saved simulation.
| Decision | Stage | Starting plan | Optimizer’s plan | Change |
|---|---|---|---|---|
| A ONC-301 solid tumours | Phase III | fund | fund | – |
| B ONC-210 leukaemia | Phase II | fund | fund | – |
| C IMM-220 psoriasis | Phase II | fund | shelved | dropped |
| D IMM-110 Crohn's disease | Phase I | shelved | shelved | – |
| E NEU-120 Alzheimer's disease | Phase I | fund | fund | – |
| F NEU-010 Parkinson's disease | Preclinical | shelved | fund | added |
| G RARE-230 enzyme deficiency | Phase II | fund | fund | – |
| H ONC-020 lymphoma | Preclinical | shelved | shelved | – |
| Measure | Unit | Starting plan | Optimizer’s plan | Change | |
|---|---|---|---|---|---|
| Projects funded | projects | 5 | 5 | 0 | |
| Portfolio NPV, mean | US$m | 556 | 545 | −12 | |
| Portfolio NPV, 1 in 10 (10th percentile) | US$m | −437 | −399 | 38 | ✔ better |
| Launches, expected | products | 1.34 | 1.22 | −0.13 | |
| Chance of no launch | chance | 18.4% | 21.2% | 2.8% | |
| Chance of breaking the budget | chance | 35.2% | 19.7% | −15.4% | ✔ better |
| R&D cost, mean (present value) | US$m | 515 | 468 | −47 | ✔ better |
| Budget risk accepted, at most | chance | 25% | 25% | ||
| No-launch risk accepted, at most | chance | 25% | 25% |
Breaking the budget: spending above US$170 million in at least one of the next 5 years.
S-curve (cumulative probability) of Starting plan, Optimizer’s plan, Budget from the saved simulation.
The business question
A mid-sized pharma company has eight development projects, from preclinical work to a Phase III cancer drug. Each phase costs money and may fail, and three pairs share a mechanism, so their Phase II results tend to agree. PH-04 compared four portfolios: one project per mechanism kept most of the value but broke the US$170 million budget about one time in three.
Here the ModelRisk Optimizer searches all 256 portfolios.
- Who it is for
- R&D portfolio committees, heads of development and strategy, finance teams in pharma and biotech.
- Main outputs
- The Optimizer's portfolio against a starting plan (PH-04's one project per mechanism) · for both: NPV (mean and 1 in 10), expected launches, the chance of no launch, the chance of breaking the budget, spending in the busiest year and R&D cost · for each project: chance of launch, mean NPV and value per dollar · spending against the budget.
What the Optimizer decides
1 = fund the project, 0 = shelve it. The starting plan is PH-04's portfolio of one project per mechanism: A, B, C, E and G.
| Decision | Starting plan | Optimizer’s plan |
|---|---|---|
| A ONC-301 solid tumours | Yes | Yes |
| B ONC-210 leukaemia | Yes | Yes |
| C IMM-220 psoriasis | Yes | No |
| D IMM-110 Crohn's disease | No | No |
| E NEU-120 Alzheimer's disease | Yes | Yes |
| F NEU-010 Parkinson's disease | No | Yes |
| G RARE-230 enzyme deficiency | Yes | Yes |
| H ONC-020 lymphoma | No | No |
- Goal
- Maximise the mean NPV
- Limit
- Chance of breaking the budget at most 25% (the chance of overspending in a budget year)
- Limit
- Chance of no launch at most 25% (the chance that no project launches)
The Optimizer’s run: Vertex Detector, 256 plans tried, each on 10,000 samples, in 86.3 seconds.
How the model works
Six sheets after this one, read from left to right. Each one feeds the next.
Inputs sheet
The annual R&D budget and its horizon, the correlation between the Phase II results of projects that share a mechanism, the discount rate, and the two limits: the budget risk and the no-launch risk accepted.
Projects sheet
The duration and yearly cost of each phase. Each project's current stage, chance of success in each phase and value at launch (P10 and P90).
Decisions sheet
The eight cells the Optimizer changes — fund (1) or shelve (0) each project — beside the starting plan, with the goal (the highest mean NPV) and the limits (the chance of breaking the budget and the chance of no launch).
Pipeline sheet
Three bivariate Normal copulas give correlated uniforms for the Phase II results of each pair; VoseBernoulli decides every remaining phase. The last phase each project works on, the launch year and the value at launch, discounted to today, follow.
Spend sheet
The phase each project works on in each year gives its spending. Each plan adds up the projects it funds: its NPV, its launches, and whether it spends more than the budget in a year of the budget horizon.
Results sheet
Both plans on the same simulated outcomes: NPV, launches, budget risk, spending in the busiest budget year and R&D cost; each project's value per dollar; spending of the Optimizer’s plan against the budget; the shared mechanisms.
ModelRisk techniques
Find similar models
- Product
- Type
- Application
- Uses
The results in detail
The projects
Each project as if funded: chance of launch, value and R&D cost (US$m).
Line chart: 90th percentile, Median, Mean, 10th percentile, Budget.
| Project | Stage | Launch chance | Mean NPV | R&D cost PV, mean | Value per US$ of R&D |
|---|---|---|---|---|---|
| A ONC-301 solid tumours | Phase III | 55% | 429 | 159 | 3.71 |
| B ONC-210 leukaemia | Phase II | 19% | 40 | 92 | 1.43 |
| C IMM-220 psoriasis | Phase II | 18% | 18 | 91 | 1.19 |
| D IMM-110 Crohn's disease | Phase I | 12% | −3 | 68 | 0.95 |
| E NEU-120 Alzheimer's disease | Phase I | 8% | 25 | 62 | 1.41 |
| F NEU-010 Parkinson's disease | Preclinical | 5% | 6 | 44 | 1.14 |
| G RARE-230 enzyme deficiency | Phase II | 34% | 44 | 112 | 1.40 |
| H ONC-020 lymphoma | Preclinical | 8% | −2 | 48 | 0.96 |
Value per US$ of R&D: the expected present value of the launch ÷ the expected R&D cost. Below 1, the project destroys value.
Shared mechanisms
Projects that share a mechanism succeed — and fail — together.
| Pair | Both launch simulated | Both launch if independent | Ratio |
|---|---|---|---|
| B and H | 3.3% | 1.5% | 2.1 |
| C and D | 4.3% | 2.2% | 2.0 |
| E and F | 1.1% | 0.5% | 2.5 |
How to read these results
- The starting plan, PH-04's one project per mechanism (A, B, C, E and G), is worth about US$555 million on average but breaks the budget about one time in three: when two of its three Phase II projects succeed, their Phase III trials start while A's is still running.
- The Optimizer keeps A, B, E and G, shelves C, the psoriasis drug, and funds F, the early Parkinson's project, which spends little in the budget years. It gives up about US$12 million of mean NPV, about 2%, and cuts the chance of breaking the budget to about one in five.
- The budget limit decides: every portfolio worth more on average breaks the budget at least a third of the time. The chance of no launch rises a little, to about 21%, inside its limit. The portfolio is not one of the four PH-04 compared: it adds B to PH-04's within-budget portfolio.
- F shares E's amyloid mechanism, so their Phase II results tend to agree: in the third table, E and F launch together two to three times as often as if they were independent. When both succeed, their Phase III trials overlap in year 5, which adds a little to the budget risk. The Optimizer judges every portfolio with that correlation included.
- Try a one-in-ten budget risk on the Inputs sheet and run the Optimizer again: it can then keep only one of the three Phase II projects, and the chance of no launch becomes the limit that binds.
Run the Optimizer yourself. Open the workbook, open the Optimizer from the ModelRisk ribbon (the saved task opens with it) and press Start; then simulate to compare the plans.
Inputs and assumptions
Every assumption in the model, with its uncertainty and where it comes from. The figures are illustrative.
Budget and finance 4 inputs
| Assumption | Unit | Value |
|---|---|---|
| R&D budgetFor each year of the budget horizon below. | US$m a year | 170 |
| Budget horizonThe years the budget is set for. | years | 5 |
| Correlation of Phase II results within a mechanismOn the Normal scale: if the mechanism fails in one project, it probably fails in the other. | copula ρ | 0.80 |
| Discount rate | per year | 10.0% |
Limits for the Optimizer 2 inputs
What the portfolio committee will accept. The Optimizer must respect both.
| Assumption | Unit | Value |
|---|---|---|
| Budget risk acceptedThe highest chance of spending more than the budget in a year of the horizon. | chance | 25% |
| No-launch risk acceptedThe highest chance that none of the funded projects launches. | chance | 25% |
Units and conventions
- Money
- US$ million (US$m); costs discounted from the middle of each year, values from the launch year.
- Time
- Years from now; year 1 is the coming year.
- Value at launch
- The NPV, at launch, of the product's sales after all commercial costs.
- Breaking the budget
- A portfolio's spending exceeds the budget in at least one year of the budget horizon (five years as shipped).
- No launch
- None of the funded projects reaches the market in the ten years modelled.
Key assumptions and limitations
- Phases
- Whole years, back to back; a project stops spending when a phase fails.
- Correlation
- Only through the Phase II results of projects sharing a mechanism; all other results are independent.
- Unfunded projects
- Shelved: no cost and no value. Partnering them out could recover some value.
- Optimizer
- Each portfolio the Optimizer tries is judged on the same 10,000 simulated futures (common random numbers, seed 1) as the simulation on the Results sheet, so portfolios are compared fairly and the limits met in the run hold on the Results sheet too.
- Data
- The company, the projects and all figures are invented for this example.
How to use it
- Enter your budget, correlation, discount rate and the two limits on the Inputs sheet, and your projects on the Projects sheet.
- On the ModelRisk ribbon, open the Optimizer: the task is saved in this workbook. Press Start. When it finishes, apply the best solution: the funding cells on the Decisions sheet change.
- Run a simulation from the ModelRisk ribbon (10,000 samples with a fixed seed are set) to compare the new portfolio with the starting plan on the Results sheet.
- To watch the search from the beginning, copy the starting plan onto the Optimizer’s plan first.
ModelRisk functions used
- ModelRisk Optimizer
- Chooses the projects to fund: highest mean NPV, chance of breaking the budget and chance of no launch at most the limits on the Inputs sheet.
- VoseCopulaBiNormal
- Correlated uniforms for the Phase II results of two projects sharing a mechanism (array).
- VoseBernoulli
- Whether each phase succeeds; the U argument takes the copula's uniform for Phase II.
- VoseLognormalAlt
- Each project's value at launch from its P10 and P90.
- VoseOutput, VoseSimMean, VoseSimPercentile, VoseSimProbability
- Outputs and their statistics on the Results sheet.
Further reading
- Monte Carlo simulation explained (Vose guide)
- Paul, S. M. et al. (2010). How to improve R&D productivity: the pharmaceutical industry's grand challenge. Nature Reviews Drug Discovery 9, 203–214.
- PH-04 R&D portfolio which projects to fund: the same eight projects, with four portfolios compared side by side.
Download the workbook and run it yourself. It opens in Excel with ModelRisk. The free trial runs every example model.
Related example models
- ModelRisk★★★★★R&D Portfolio: Which Projects to FundEight projects, one budget: which to fund for value, for a steady pipeline, and within the money we have?Eight projects with phase-by-phase attrition and correlated Phase II results; four portfolios compared on value, launches and budget.
- ModelRisk★★★★★Drug Development Risk-Adjusted NPVWhat is a compound in Phase II worth — and should we develop it ourselves or license it out?Phases with uncertain cost, time and success, then launch and patent expiry: the textbook rNPV against the simulated NPV.
- ModelChoice★★★★★Biosimilar Launch Strategy DecisionLaunch a biosimilar at risk, wait for a patent settlement, license it in, or stay out of the market?Three decisions and 13 chance events. Best first decision: At-Risk Launch, expected value $130.0m.
An illustrative example: the figures are invented for teaching and do not describe a real company, project or asset. Last updated 9 October 2026.