Interactive model leaderboard
Compare every pricing, payment, and actuarial model side-by-side on a shared metric surface — backtest MAPE, 90% CI coverage, auditability, and equity — then filter by scenario and region to see which model dominates under the constraints you actually face. The winner is chosen by a weighted composite you control below; the "why this model wins" panel spells out the deltas.
Domain
Scenario
Region
accuracy weight50%
coverage weight25%
auditability weight15%
equity weight10%
Saved views & shareable links
Save the current domain, scenario, region, and composite weights as a named view, or copy a link that restores this exact configuration on any device.
No saved views yet.
Worked case — regenerate every model & re-rank by dominance
Edit the costs, QALYs, comparator anchor, and WTP threshold below. Every pharma model recomputes its recommended price, ΔC, ICER, and net monetary benefit NMB = λ·ΔE − ΔC live. The ranking sorts by NMB (payer value); models with strong dominance (ΔC < 0 AND ΔE > 0) are flagged.
Comparator cost ($)
Comparator QALYs
New therapy QALYs
WTP threshold λ ($/QALY)
COGS ($)
Cost-plus markup ×
ERP basket median ($)
Ramsey reference price ($)
Ramsey income ratio Y_c/Y_ref
Ramsey α (equity exponent)
Orphan launch premium ×
ΔE (QALYs)
1.20
Comparator cost
$30,000
WTP λ
$100,000/QALY
Dominant model
Cost-plus
| # | Model | Price | ΔC vs comp. | ICER ($/QALY) | NMB | Verdict |
|---|---|---|---|---|---|---|
| 1 🥇 | Cost-plus P = COGS × markup | $24,000 | $-6,000 | $-5,000 | +$126,000 | Strongly dominant |
| 2 🥈 | Ramsey–Boiteux (α-equity) P_c = P_ref (Y_c / Y_ref)^α | $47,390 | +$17,390 | $14,491 | +$102,610 | Cost-effective at λ |
| 3 🥉 | External reference pricing P_c = f({P_k : k ∈ basket}) | $60,000 | +$30,000 | $25,000 | +$90,000 | Cost-effective at λ |
| 4 | Hybrid Value–Equity (this app) ICER · BI-soft · Ramsey α · orphan · RWE, floored & capped | $69,000 | +$39,000 | $32,500 | +$81,000 | Cost-effective at λ |
| 5 | Indication-tiered VBP P = Σᵢ wᵢ (λ ΔQALYᵢ + P_i^comp) | $138,000 | +$108,000 | $90,000 | +$12,000 | Cost-effective at λ |
| 6 | ICER (pure value) P = λ · ΔQALY + P_comp | $150,000 | +$120,000 | $100,000 | +$0 | Cost-effective at λ |
ICER color bands: green < λ/2 (highly cost-effective), sky < λ, amber < 1.5λ, rose ≥ 1.5λ. NMB uses your λ directly, so raising the WTP slider makes marginal models flip to cost-effective in real time.
Leaderboard
| # | Model | Cases | MAPE ↓ | CI cov → 0.90 | Audit /10 | Equity /10 | Composite |
|---|---|---|---|---|---|---|---|
| 1 🥇 | Hybrid Value–Equity (this app) ICER · BI-soft · Ramsey α · orphan · RWE, floored & capped | 48 | 15% | 0.87 | 9 | 9 | 80.4 |
| 2 🥈 | Ramsey–Boiteux (α-equity) P_c = P_ref (Y_c / Y_ref)^α | 18 | 27% | 0.75 | 8 | 10 | 61.0 |
| 3 🥉 | Indication-tiered VBP P = Σᵢ wᵢ (λ ΔQALYᵢ + P_i^comp) | 27 | 26% | 0.75 | 8 | 6 | 57.6 |
| 4 | ICER (pure value) P = λ · ΔQALY + P_comp | 36 | 29% | 0.72 | 9 | 6 | 55.0 |
| 5 | External reference pricing P_c = f({P_k : k ∈ basket}) | 23 | 42% | 0.61 | 8 | 7 | 34.2 |
| 6 | Cost-plus P = COGS × markup | 20 | 77% | 0.42 | 10 | 5 | 20.0 |
Weight sensitivity — how stable is each rank?
Resamples the composite weights uniformly around the current settings (each weight perturbed by ±spread, clamped ≥ 0, then normalised) and re-runs the ranking 300 times. Low rank SD and a high share of top-1 finishes mean the model wins across a wide neighbourhood of weight choices, not just at your current slider positions.
Weight spread (±)25 pp
Samples300
| Model | Mean rank | Rank SD | Best | Worst | % top-1 | % top-3 | Rank range |
|---|---|---|---|---|---|---|---|
Hybrid Value–Equity (this app) ICER · BI-soft · Ramsey α · orphan · RWE, floored & capped | 1.00 | 0.00 | 1 | 1 | 100% | 100% | |
Ramsey–Boiteux (α-equity) P_c = P_ref (Y_c / Y_ref)^α | 2.37 | 0.60 | 2 | 4 | 0% | 94% | |
Indication-tiered VBP P = Σᵢ wᵢ (λ ΔQALYᵢ + P_i^comp) | 2.80 | 0.54 | 2 | 4 | 0% | 94% | |
ICER (pure value) P = λ · ΔQALY + P_comp | 3.83 | 0.48 | 2 | 4 | 0% | 13% | |
External reference pricing P_c = f({P_k : k ∈ basket}) | 5.01 | 0.08 | 5 | 6 | 0% | 0% | |
Cost-plus P = COGS × markup | 5.99 | 0.08 | 5 | 6 | 0% | 0% |
Bar shows [best…worst] rank across samples; the dark tick is the mean rank. Green SD ≈ locked in; rose SD ≈ ranking flips easily when weights move.
🥇 Why Hybrid Value–Equity (this app) wins across Pharma
Composite score
80.4
Backtest MAPE
15%
90% CI coverage
0.87
Filtered cases
48
Runner-up: Ramsey–Boiteux (α-equity) · score 61.0 · MAPE 27% · cov 0.75
- Lower backtest MAPE than Ramsey–Boiteux (α-equity) by 11.7 pp on the filtered 48 cases.
- Tighter 90% CI calibration: |cov−0.90| improves by 11.6 pp.
- Higher auditability (9/10 vs 8/10) — regulators can reproduce every step.
Cases that produced the ranking
| Scenario | Region | MAPE | CI cov | n |
|---|---|---|---|---|
| US·ICER | US | 14% | 0.88 | 12 |
| UK·NICE | UK | 13% | 0.89 | 10 |
| EU·JCA | EU | 15% | 0.87 | 10 |
| LMIC·Ramsey | LMIC | 16% | 0.85 | 8 |
| Orphan launch | Global | 18% | 0.83 | 8 |