| Vose Software

Industry: Pharmaceutical
Product: ModelRisk
Application: Patent Portfolio Analysis


Pricing the option value of patent extensions across six assets at risk of loss of exclusivity

A blockbuster cardiology asset earning $2.9 billion a year, four years out from loss of exclusivity, can lose 80–90% of its revenue inside 18 months when the first generic launches at risk. A successful patent term extension can push the loss-of-exclusivity wall out by an average of 18 months. An inter-partes review (IPR) loss can pull it forward by 12 to 36 months. The legal team has to commit a finite budget — typically $1M to $6M per asset, depending on the breadth of filings, divisionals, and geographic coverage — against six or more assets simultaneously, with no way to know in advance which dollars will buy revenue and which will fund a defensive position the courts never test.

A specialty pharma group with six revenue-generating assets — one approaching LOE in six months, two more inside three years — used ModelRisk to recast its IP strategy as a portfolio of real options rather than a list of legal motions. Each asset's "file-and-fight" investment was priced by the simulated NPV uplift it generated, net of the filing cost. The output was an ordered shopping list: which IP investments pay for themselves handsomely, which are merely marginal, and what the whole-portfolio NPV distribution looks like under each strategy.

The headline result was the whole-portfolio NPV distribution itself — and the way its shape, not just its mean, moved once the lawyers' work was priced in:

Portfolio 15-year NPV — file-and-fight IP strategy versus base case

The mean lifts by about $2.9B, which the deterministic worksheet would have predicted. The real story is in the lower tail: the file-and-fight strategy lifts the P10 of portfolio NPV from roughly $20.0B to $23.4B, because the defensive filings cut the magnitude of the worst IPR outcomes. The strategy is not just additive expected value — it is downside-protective, and that protection is worth more to a company with covenants and dividend commitments than the mean lift itself.

Why deterministic patent ROI calculations fail

The traditional patent ROI worksheet looks like this: "If the PTE is granted (60% probability), it adds 18 months of $2.9B revenue at 80% gross margin, discounted back, less $4.5M legal cost — expected value ~$1.6B." This calculation is right in the limit, but it answers the wrong question. The board does not need the expected value of the PTE in isolation; it needs the joint distribution of portfolio NPV under "file-and-fight everything" versus "do the minimum," because the extensions interact with generic erosion, IPR risk, and each other through the budget constraint.

The probabilistic build modeled each asset with four uncertainty drivers:

  • Patent term extension — modeled as a Bernoulli with asset-specific success probability (15%–70%, calibrated to the USPTO PTE grant rate by therapeutic class and the strength of the residual claims), conditional on a Triangular extension length (mode 0.6 to 2.4 years) if granted.
  • Inter-partes review loss — a Bernoulli (5%–40% by asset), conditional on a Uniform(1, 3) year truncation of remaining exclusivity if the challenge succeeds. The IPR rate is fitted to PTAB outcomes in the same therapeutic class over 2015–2024.
  • Generic erosion retained share — Beta-distributed steady-state share of revenue retained 24 months post-LOE, scaled to a [0, 30%] range. The α/β parameters were set per asset based on therapy form factor (oral solid vs device-coupled vs biologic). Beta is the defensible family — a Normal here would let the simulator draw negative or above-100% retention.
  • Underlying revenue decay — 2% per year through the in-exclusivity period, capturing class growth offset by patient switching to newer entrants.

Defensive filings (divisionals, formulation patents, geographic expansions) were modeled to dampen IPR severity by 30%, reflecting empirical evidence that broader claim families reduce the magnitude of an invalidation rather than its probability.

Per-asset option value — what to fund and what to drop

The portfolio-level uplift is the sum of asset-level option values. Asset A's is by far the largest at roughly $1.6B; the rest descend to Asset F's marginal $31M, which barely clears its own filing cost.

Per-asset option value of extending and defending the patent

Asset F — the small anti-infective with six months of exclusivity remaining and a 40% IPR loss probability — returns only about $31M of expected NPV uplift on its $0.8M of planned IP spend: a positive but marginal payback, because the residual exclusivity is short and the IPR risk high. Set against Asset C — where the option value was over $250M for $2.5M of cost — Asset F is the obvious dollar to redeploy, and the team reallocated its planned filing the week the chart was first shown toward two divisional applications on Asset C.

Asset A is where the legal budget lives or dies

The cardiology blockbuster is so much larger than the rest of the portfolio that the IPR coin-flip on it dominates the portfolio NPV variance.

Asset A revenue trajectory — what the IPR coin-flip costs

The red trajectories — the 18% of trials in which IPR succeeds — collapse two to three years earlier than the blue ones. The deterministic plan (the dotted line) lives between the two clouds; on the expected trajectory it looks reasonable, but the median outcome is at least one of the two extreme regimes, never the average of them. This is the classic pitfall of taking expectations inside a nonlinear payoff: the expected revenue is not the revenue of the expected scenario.

What actually moves portfolio NPV

A one-at-a-time tornado decomposes portfolio-NPV uncertainty by parameter.

Tornado: what swings 15-year portfolio NPV most

Asset D's remaining exclusivity — it has barely eighteen months left, so a one-year swing either way reshapes its whole NPV — is the single largest mover. Close behind it sits the Asset A erosion-retained share, fundamentally a question of how aggressively the first generics price and how loyal the prescriber base is. Because Asset A is so much larger than Asset D, that second mover is the one the IP team acted on: rather than another month spent on the PTE petition language, the team commissioned an authorized-generic strategy review, which informs how to defend retained share in the first 24 months of generic competition.

What the model changed

  • Asset F's marginal $0.8M filing redirected to the Asset C divisional family — the same dollar buys ~$255M of option value on Asset C versus only ~$31M left on Asset F.
  • Asset A authorized-generic readiness program launched — driven by the tornado finding that retained-share volatility on Asset A is the largest value-weighted source of portfolio-NPV exposure.
  • Portfolio "file-and-fight" spend (~$19M across the six assets) reallocated rather than expanded — concentrating the marginal dollar on the assets where it buys the most option value, with the whole-portfolio mean NPV lifting by roughly $2.9B over the base case.
  • IP committee reports portfolio NPV as a P10/P50/P90, not a single point. Quarterly variance is reported against the band; budget reallocations happen when an asset's option value crosses zero.

ModelRisk Functionality Used

  • Bernoulli × Triangular composite distributions for the PTE-grant-then-length lottery, and Bernoulli × Uniform for the IPR-then-truncation lottery, on each asset.
  • Beta-distributed generic retained share scaled per therapy form factor — replacing the previous worksheet's flat 12% point estimate that hid the cross-asset variance.
  • 15-year NPV propagation per trial with a 9% discount rate, evaluated against the live revenue forecast schedule in Excel.
  • Before/after histograms of portfolio NPV under base and file-and-fight strategies, with the P10 contrast as the headline number for the board.
  • Per-asset option-value bar — net expected NPV uplift minus filing cost — used directly as the budget-allocation shopping list.
  • One-at-a-time tornado on the seven highest-impact inputs across the six assets, which redirected research and authorized-generic readiness work.

Patent portfolio management is not about which patent to defend; it is about which patent dollars buy real option value, and what shape the residual exposure has after the lawyers have done their work. Monte Carlo simulation in ModelRisk makes that option value visible at the asset level and the portfolio level simultaneously, so the legal budget can be allocated by NPV uplift rather than by which case is loudest in the room.