How the calculator maps a selected builder route into a teaching estimate.

The doctrine: net first, then apply the clause

A builder basket is a contractual permission, not a cash account. Its CNI limb, seed, eligible credits, usage and gates must be read together. The calculator is a teaching estimate of one explicitly defined shared-builder pool plus separately entered independent baskets. It is neither a document parser nor a lender-loss valuation.

For a single-accounting-period CNI clause, aggregate covenant-defined income first. Apply the selected percentage to a positive aggregate, or 100% to an aggregate deficit. Applying that switch separately to each losing year is a different, incorrect formula for that clause.

Three floor conventions are available. They are alternative hypothetical contracts, not elections that a borrower can switch between:

• No CNI floor: a negative aggregate CNI limb can consume starter and equity credits inside the shared builder.
• Whole-limb floor: take the greater of zero and the finished cumulative limb. Raw earnings history is never reset; later income must still offset earlier cumulative losses.
• Per-period floor: ignore each negative input year before aggregation. This is a true positive-period-only formulation. Annual inputs do not reproduce a quarterly clause if profits and losses offset within a year.

For example, CNI of +150 and −75 gives a net-first 50% limb of 37.5, not the 0 produced by a year-by-year 50/100 calculation. A per-period zero floor instead produces 75. All three statements describe arithmetic, not document prevalence.

The shared and independent ledgers

The default starter belongs to the common builder pool, as do eligible builder equity credits. The independent-starter toggle moves it outside that pool: enter its remaining capacity, not its original headline limit. Builder losses and usage cannot consume genuinely independent baskets. Independent general, declined-proceeds and asset-sale amounts are also entered net of usage and without overlap.

Excluded contributions are not automatically ordinary builder credits; neither are equity cures, retained sale proceeds or declined debt prepayments. Enter an amount only in its documented route. Do not count the same proceeds in CNI, equity and an independent route, and do not deduct the same historical payment twice.

The source bars use a display-only attribution: remaining shared capacity is assigned to seed first, then equity, then positive CNI. This makes losses and usage visibly reduce the remaining sources. It does not prescribe a legal allocation election. The signed CNI ledger remains visible even where a source bar cannot be negative.

Inputs and conditions

Capacity is not permission. An actual payment needs its own no-default, pro-forma leverage/coverage, investment, designation, transfer, anti-leakage and other conditions. A single probability cannot determine the maximum legal dividend, and a failing large payment does not prove that every smaller payment fails. No cash or debt-funding model is supplied here.

The bps bridge: conversion, not valuation

Every monetary amount uses the same €m unit. PV01 is therefore €m per basis point; dividing V by it produces basis points. The assumed debt has annual spread payments through the selected hold, fixed principal and 100% survival. It is not the risky annuity of an observed credit curve.

The coupon-slice figures are merely an amount-to-face fraction multiplied by the quoted spread. They have no annuity and are not spread-equivalents. None of these three objects is additive to another or to a quoted margin. A unit conversion cannot turn discounted permission into a credit price: the cash actually distributed, lender exposure, recoveries and risk-neutral dynamics remain unspecified.

Stochastic CNI and terminal policies

All policies pay at the same terminal date and are discounted once. The simulation covers the shared builder, including any shared starter/equity, but not independent baskets. Its weights are illustrative scenario weights, not an established physical forecast or risk-neutral measure. Discounting at a risk-free input does not fix that evidence gap.

The 10,000 paths comprise 5,000 antithetic pairs, with seed 31051987 and a Box–Muller normal transform. The standard error is the sample standard deviation of pair means divided by √5000; it measures simulation noise only. The displayed histogram contains q-weighted discounted amounts, not simulated Bernoulli exercise outcomes. Percentiles use linear interpolation between order statistics.

The q = 1 result dominates the two other nonnegative policies within this same-date model. The threshold policy can be above or below the q-weighted result. These are policy sensitivities, not universal optimal-exercise bounds. No forward-looking stopping rule, interim payment, repeated draw, future reload or post-distribution CNI feedback is modeled.

Worked example: corrected base case

Closing EBITDA is €100m, hold five years, starter €40m inside the shared builder, general independent capacity €25m, and no equity credit or prior usage. CNI is 10, 25, −40, 30 and 40; builder share 50%, whole-limb floor on, q 70%, volatility 30%, debt €600m and discount rate 4.5%.

The seeded stochastic shared-builder estimate is €42.15444m, corresponding to 160.49916 bps via the annual survival-1 PV01 and 29.85940 bps via the unrelated coupon-slice diagnostic. Discounted floor uplift against no floor is €2.36615m: the floor can protect the shared seed on adverse paths even though its deterministic exit uplift is zero.

These revised figures deliberately differ from the old running-clamp example. The CNI aggregation is corrected, and the stochastic numerator now includes the whole shared pool rather than just an isolated positive CNI amount. A higher headline is therefore not evidence that this covenant is worth more.

A deficit and an independent basket

With CNI +36 and −84, shared starter 50, equity 22 and usage 4, net CNI is −48. No-floor shared capacity is 20; whole-limb-floor capacity is 68; per-period-floor capacity is 86. An independent general basket of 10 adds 10 to every case. Setting q to zero weights the common pool to zero but does not erase that independent 10. This illustrates separate ledgers, not a verified legal permission.

Reading the exhibits

• Ledger: raw cumulative CNI, signed selected CNI limb, spendable shared pool and independent net permissions.
• Stochastic layer: discounted terminal shared-builder amounts, with sampling error and selected-floor uplift against the same paths without a floor.
• Source stack: nonnegative residual components that reconcile exactly to total capacity.
• Builder path: signed CNI limbs, before seed, equity or usage. Negative unfloored limbs are not hidden as zero spendable capacity.
• Year snapshots: each measurement date uses the same entered seed, equity, independent amounts and usage. These are not a dated issuance, reload or withdrawal schedule.

History dependence versus early exercise

The cumulative formulation depends on total earnings since the contractual start, not just final-year CNI. But with terminal-only exercise and fixed credits/usage, permuting the annual inputs leaves its terminal payoff unchanged. The aggregate is a sufficient state variable; full history need not be retained to evaluate this payoff. A per-period floor also needs the sum of positive periods.

Real builder rights can be used at interim measurement dates. Then sequencing, legal gates, earlier usage, financing, reload timing and sponsor policy matter, and optimal exercise may need a richer dynamic model. This tool does not solve that problem and does not claim that Monte Carlo is the only valid method.

What the model leaves out

• Covenant-defined CNI exclusions, reporting lags, restricted-group perimeter changes, currency translation and quarterly resolution.
• Serial correlation, mean reversion, realistic income tails, leverage, default/survival and cash generation; normal shocks can produce arbitrarily negative income.
• Cash to fund a distribution, new borrowing, debt capacity, taxes, recoveries, lender loss allocation and the financing effect on future CNI.
• Repeated exercise, dated usage/reloads, reclassification, basket stacking constraints and dynamic pro-forma legal tests.
• Market calibration. Simulation standard error does not include model, input, measurement or legal uncertainty.

Primary-source checks

The net-first sign convention, equity credits, usage and separate exceptions can be checked against an SEC-filed indenture, Section 4.07. Another SEC-filed offering document illustrates cumulative-period drafting and excluded-contribution distinctions. These are drafting examples, not universal rules or issuer analyses. The floor alternatives and route assumptions above are expressly selected teaching contracts; inspect the actual executed agreement before use.

Where this sits in the LevFin Book

The contingent-rights framework separates contractual taxonomy, teaching estimates and the evidence needed for a calibrated risk-neutral price. This lab stops at selected permission arithmetic and terminal-capacity sensitivities. See the LevFin Book for the wider covenant map.

Common questions