By name: sensitivity-analysis description: Banking-grade sensitivity, Monte Carlo, scenario, and break-even analysis for project finance. Use when an engagement needs tornado diagrams, P80 NPV, DSCR distributions, or stress testing for bankability discussions.
Sensitivity Analysis
Five complementary methodologies for financial risk assessment. All produce CPS-branded PNG charts and markdown reports suitable for board presentations and lender packs.
When to use
- Bankability studies and lender packs (term sheet, debt sizing)
- BOT / PPP feasibility (tariff negotiation, risk allocation)
- Investment decisions with uncertain inputs (Greenfield, expansion CAPEX)
- Board presentations requiring quantified downside (stress testing)
Methodologies
| Method | Purpose | Primary output |
|---|---|---|
| Tornado | Identify top value drivers | Variable impact ranking on NPV / IRR / DSCR |
| Monte Carlo | Probabilistic outcomes | P10 / P50 / P80 / P90 distributions |
| Spider | Multi-variable elasticity | Sensitivity curves for each input |
| Scenario | Discrete case comparison | Base / Upside / Downside / Stress |
| Break-Even | Threshold identification | Headroom to covenant breach |
How to invoke
- Configure project parameters in
plugins/finance/scripts/config.py(or the engagement's local override). Required:total_capex,annual_throughput_mta,tariff_usd_per_tonneopex_fixed_annual,opex_variable_per_tonnedebt_ratio,interest_rate,loan_tenor_years,waccbase_npv,base_irr,base_min_dscr(from your DCF model)
- Run
python plugins/finance/scripts/run_all_analysis.py - Review outputs in
<engagement>/charts/sensitivity/and<engagement>/reports/sensitivity/
Slash commands
/cps-fin:sensitivity— full tornado + spider suite/cps-fin:montecarlo— Monte Carlo with configurable distributions/cps-fin:scenarios— discrete scenario comparison/cps-fin:breakeven— break-even and headroom analysis
Quality checklist
- Base case NPV/IRR/DSCR matches the underlying DCF model (within 1 %)
- Tornado includes at least 6 variables ranked by impact
- Monte Carlo runs >= 10,000 iterations for stable P80
- Scenarios are MECE (Base / Upside / Downside / Stress, no overlap)
- Break-even reports headroom % to each covenant
- Charts use CPS branding (colors from
assets/cps-branding.json) - Banking conventions cited (P80 standard, DSCR ≥ 1.20x covenant)
References
references/p80-banking-conventions.md— when to use P80 vs P50references/dscr-thresholds.md— covenant interpretationplugins/finance/scripts/config.py— parameter and distribution definitions