Value · Risk
Risk analysis
A stochastic engine runs every Monte Carlo draw through the full DCF — with expert-elicited flex tables and a genuine Bayesian price/grade calibration — reported in the P10/P50/P90 convention lenders and ICs expect.
See it
What it does
Full-DCF Monte Carlo
Each sample runs through the enterprise DCF (tax, loss carryforward, interrelated revenues) — not a reduced-form NPV.
Expert flex tables
Discrete priors sampled by inverse-CDF to preserve skew and honour min/max bounds.
Bayesian price calibration
A mean-reverting (Schwartz/OU) process with a dense-grid posterior produces per-year price cones.
Correlated commodities + grade
Composite multi-commodity price paths with copula correlation, plus an AR(1) grade cone across years.
Exceedance reporting
P10/P50/P70/P90, credible intervals, probability NPV positive, survival curves and tornado.
Covenant risk
DSCR-breach and liquidity-breach probabilities per period; fixed seed + calibration snapshot saved for audit.
How it works — the engineeringTechnical detail
NumPy prior-predictive Monte Carlo through the full DCF, inverse-CDF discrete sampling, an OU/Schwartz mean-reversion with a grid posterior, and Gaussian-copula correlation — reported in exceedance statistics. The result is a defensible NPV distribution, not a single point.
enterprise_finance/services/ · enterprise_risk_bayesian · price_calibration · calculation_engine_risk_v3