merton.portfolio.loss_distribution

LossDistribution container for portfolio Monte Carlo / analytic outputs.

Classes

LossDistribution

Histogram of simulated portfolio losses (in currency units or fraction).

Module Contents

class merton.portfolio.loss_distribution.LossDistribution

Histogram of simulated portfolio losses (in currency units or fraction).

Parameters:

• losses – 1-D array of simulated losses (one per Monte Carlo draw).

• weights – Optional importance-sampling weights; uniform by default.

• contributions – Optional (n_sims, n_firms) matrix of per-firm losses, used for marginal-contribution analysis.

losses: merton._typing.FloatArray

weights: merton._typing.FloatArray | None = None

contributions: merton._typing.FloatArray | None = None

mean() → float

std() → float

var(level: float = 0.99) → float

Quantile-based VaR (α-percentile of the loss distribution).

expected_shortfall(level: float = 0.99) → float

Average loss conditional on exceeding the α-VaR.

economic_capital(confidence: float = 0.999, *, capital_basis: str = 'el') → float

Capital sufficient to cover unexpected losses at confidence.

• capital_basis="el" (default): VaR(α) − E[L] (Basel-style unexpected-loss capital).

• capital_basis="zero": VaR(α); treat the full quantile as capital (used by some IFRS-9 implementations).

quantile(q: float | merton._typing.FloatArray) → float | merton._typing.FloatArray

Empirical quantile of the loss distribution.

histogram(bins: int = 100) → tuple[merton._typing.FloatArray, merton._typing.FloatArray]

firm_contributions(level: float = 0.99, method: str = 'es') → merton._typing.FloatArray

Per-firm expected-loss contribution at the given quantile.

method='es': mean per-firm loss conditional on portfolio losses exceeding the α-VaR. method='var': per-firm loss at the worst Monte Carlo draw that defines the VaR quantile. Requires contributions to be set.

to_pandas()

summary(level: float = 0.99) → str