AtlanticM&A

Atlantic Deal Value Bridge: the model

BETA

How the Atlantic LBO model works

One engine runs the deal from funding to exit, year by year, and shows every line it computes. The Excel download is the same model as live formulas, so any figure on screen can be checked in a spreadsheet.

  1. 1

    Sources and uses

  2. 2

    Operating model

  3. 3

    Cash flow

  4. 4

    Debt schedule

  5. 5

    Exit

  6. 6

    Returns and the value bridge

1. Sources and uses

How the purchase is funded and where the money goes. Sources are the debt tranches, sponsor and LP equity, management rollover and other instruments; uses are the purchase price, fees and cash to the balance sheet. The model checks that the two balance before anything else runs.

2. Operating model

Revenue, cost of goods and SG&A for each year of the hold, from the scenario you choose. EBITDA is revenue less cost of goods less SG&A. Scenarios let you set a base, upside and downside path side by side, and actual results can replace the plan as they come in.

3. Cash flow

From EBITDA to cash: interest and tax actually paid, capital expenditure, working capital driven by receivable, payable and inventory days, acquisitions paid in cash, and costs that sit outside EBITDA. Cash above the minimum balance can be paid out during the hold.

4. Debt schedule

Each tranche on its own line: opening balance, cash interest, PIK interest added to principal, scheduled amortisation and closing balance. A revolver draws when cash would fall below the minimum, and an excess cash flow sweep repays debt where the terms call for it.

5. Exit

Equity value at exit is exit EBITDA times the exit multiple, less debt, plus cash. The exit multiple is yours to set, or read from valuation curves by company size. A scenario matrix shows returns across a range of exit multiples at once.

6. Returns and the value bridge

The distribution waterfall splits exit equity: return of capital, preferred return, hurdle, catch-up and promote tiers. IRR uses the actual closing and year-end dates. The value bridge then shows where the gain came from: EBITDA growth, multiple change, and debt paydown with other cash.

The Excel is the model

The download is not a report of the results. It is the model, written out as formulae, in four sheets: Contents, Modelling Assumptions, Model Output and Live Model.

On the same inputs, the Live Model's formulae recompute to the figures in the app. That is tested on every change to the engine, so the number you present in a meeting is the number the engine computed.

Tested on real deals

We rebuilt 33 public LBOs from their filings and checked the engine lands on what happened. Net debt at exit is within 1% on 31 of 33 deals; the sponsor's return is within 1% on 30 of 32.

Read the corpus study

How we check it

Six rules every one of the 33 rebuilt deals is held to. They run as automated tests on every change to the engine or the data, so a change that breaks one cannot ship. Figures as of 17 October 2026.

1

The 1% standard, at every year end

Debt and cash must each land within 1% of what the filings report at every year end of the hold, not just at exit, and the sponsor's return within 1% of the filed figure. Every feature of the engine a customer can use must be proven by at least one deal that meets it.

27 of 33 deals meet it in full; all 40 engine features proven

2

Each year of the hold, against the filings

Capital expenditure, interest and tax as cash actually paid, operating cash flow, distributions and working capital are each compared with the filed figures year by year. Debt and cash are checked separately, because a net figure can hide two errors that cancel.

268 of 296 filed year ends within 1%; the other 28 each carry a reason

3

Every figure traces to a filing

Each figure we enter is located in a stored copy of the filing that prints it, down to the table and the line. A figure no filing prints is named, with the filing that was checked and found silent.

1,174 of 1,247 figures located; the other 73 each say why

4

Every day of ownership

Each hold runs on its real closing date and the filer's own year ends, so a deal closed in October is charged for the weeks it was owned, not a whole year.

32 of 33 on their filed dates, 26 with a partial first or last period

5

Gaps stated, never tuned away

Where the engine and the filing still differ, the difference is recorded on the deal with its cause. If a later change moves the figure, the test fails until the reason is restated. No explanation can outlive its number.

42 gaps declared, each checked against the live figure

6

One engine, and the Excel is the same model

Every deal is calculated by the same engine with no per-deal exceptions. The Excel download's live formulas are recomputed and must equal the engine's figures, and a new engine output cannot ship until the workbook renders it or a reason is recorded.

Every engine output is in the workbook or waived with a reason

What the model does not do

  • It does not score a deal or recommend one. It shows the numbers your inputs produce, and every calculation behind them.
  • It does not forecast. The operating path is the scenario you set, or the actuals you load.
  • It does not hide a step. Every line in the workbook view is the line the engine used.