Sizes the debt a lender will actually fund -- leverage in turns of the EBITDA credit will accept, with fixed-charge and interest coverage at close and through a downturn -- when you need to separate what can be raised from what can be serviced.
Scanned 9/19/2026
Install to Claude Code
npx -y skills add andreworia/claude-finance-skills --skill debt-package-sizing --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Debt Package Sizing?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/andreworia-debt-package-sizing)More formats (shields.io, HTML) on the badges page.
---
name: Debt Package Sizing
description: Sizes the debt a lender will actually fund -- leverage in turns of the EBITDA credit will accept, with fixed-charge and interest coverage at close and through a downturn -- when you need to separate what can be raised from what can be serviced.
---
# Debt Package Sizing
## When to use
Use this skill before the financing is fixed in a bid: when a bank has quoted turns off management's adjusted EBITDA, or when the equity cheque is being set. It answers two questions and keeps them apart -- what a credit committee approves today, and what the business services through the downturn its own sector has already had.
## What it does
Produces a sizing note: statutory EBITDA rebuilt addback by addback to the number a lender will credit, debt in turns of that number, tranches with rates and amortisation, interest and fixed-charge coverage at close and at the trough, and the gap between the fundable quantum and the serviceable one.
## Method
### Step 1 -- Rebuild EBITDA the way a lender will
List every addback with its amount and its evidence, then classify each as accepted (one-time, documented, already actioned), rejected (recurring, market-driven or forward-looking) or contested (route it to the QoE provider). Run-rate synergies, contracts won but not shipped, and input costs normalised to a historical average are claimed most and rejected most. State the haircut in dollars and per cent.
### Step 2 -- Size the quantum in turns of that number
Total debt is the leverage multiple times credit EBITDA, and every ratio afterwards sits on that base. Run the same multiple off management's EBITDA too: the difference is not a modelling artefact, it is equity the sponsor contributes or price the sponsor does not pay.
### Step 3 -- Structure and price the tranches
Senior term loan, second lien or mezzanine where the quantum needs it, and a revolver sized to the peak working capital swing, not the average. State each tranche's spread, amortisation and maturity; year-1 cash interest includes the undrawn commitment fee.
### Step 4 -- Test coverage at close
Interest coverage is credit EBITDA over cash interest. Fixed-charge coverage is credit EBITDA less capex, over cash interest plus mandatory amortisation plus cash taxes -- total capex, because the growth capex sits in the plan this debt funds. Report both ratios on credit EBITDA and on management's EBITDA: the distance between the two pairs is the negotiation.
### Step 5 -- Service it through a downturn, then state both numbers
Take the documented peak-to-trough EBITDA decline for this business or its closest comparable set; do not invent a haircut. Re-run leverage and coverage at the trough, with the downturn landing in year 1, before the cash sweep has bought back any of a turn. Solve for the entry leverage that keeps trough leverage inside the covenant: covenant level times one minus the decline. What can be raised is a market fact and moves quarterly; what can be serviced is a property of the business and does not. Report both, and say whether the gap closes with equity or with price.
## Inputs
- Statutory EBITDA and the addback schedule with supporting evidence
- Bank quotes or indicative terms: turns, tranches, spreads, amortisation
- Capex, working capital seasonality and cash tax forecast
- Peak-to-trough EBITDA history for the business or its comparable set
- Proposed covenant levels
## Output format
1. Credit EBITDA bridge, addback by addback, each one classified
2. Quantum: turns, total debt, gap versus management's EBITDA
3. Tranches: quantum, pricing, amortisation, year-1 cash interest
4. Coverage at close, on both EBITDA bases
5. Downturn test: trough EBITDA, leverage, coverage, covenant
6. Fundable versus serviceable quantum, and how the gap closes
Total length: 600-900 words, written to survive a credit committee.
## Example
**Sizing note (fictional -- Calder Packaging Group, rigid packaging converter):**
Management presents adjusted EBITDA of $58.0M against statutory EBITDA of $47.0M: $11.0M of addbacks. Accepted are $2.0M of sale-process fees, $1.5M of owner-compensation normalisation and $1.5M of duplicate plant costs already actioned; rejected are $1.0M of closure costs still only planned, $3.0M of contracts won but not shipped and $2.0M of input-cost inflation normalised to a historical average. Credit EBITDA is $52.0M -- a $6.0M haircut, 10.3%. At the 4.5x lenders are quoting, that funds $234.0M against the $261.0M the same multiple implies on management's number: $27.0M of equity created by three addbacks.
Structure: a $190.0M term loan at 9.00%, $44.0M of second lien at 12.50% and a $30.0M undrawn revolver at a 0.50% commitment fee -- $22.75M of year-1 cash interest. Interest coverage at close is 2.29x and fixed-charge coverage 1.37x, on $14.0M of capex, $1.9M of amortisation and $3.0M of cash taxes; on management's $58.0M that ratio reads 1.59x, which is what the seller's banker will present. The sector's last peak-to-trough decline was 25%: at $39.0M of EBITDA leverage is 6.0x against a 5.75x covenant, a breach of a quarter of a turn in year one of a downturn. Leverage that clears the covenant through the same decline is 5.75 times 0.75, or 4.31x -- call it 4.25x, $221.0M. The market funds $234.0M. The business services $221.0M. The $13.0M between them is equity, or it is price.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!