1. Data the filings never printed
26 of the 33 deals have at least one figure no filing discloses, each named with the filing that was checked and found silent.
Research
BETA33 leveraged buyouts, rebuilt from their filings
We rebuilt 33 public LBOs, closed between 2003 and 2016, from the buyer's and target's own SEC and Companies House filings. For each one we entered the deal's filed terms and filed operating results, ran our LBO engine, and checked it lands on what actually happened: the debt and cash at exit, and the sponsor's return.
Figures as of 17 October 2026, generated from the engine's tests. They change when a deal is corrected or a gap is closed.
2.44x
median sponsor return (MOIC), as filed, across 32 deals
20 of 33
deals where EBITDA growth was the largest source of value
16 of 33
exited at a higher EBITDA multiple than they entered at
3 of 32
returned less than the equity put in (Freescale, Generac and HD Supply)
Where the returns came from
The engine splits each deal's equity gain into three legs: growth in EBITDA, the change in the EBITDA multiple between entry and exit, and debt paid down from cash flow. Across the 33 deals:
EBITDA growth
The largest leg in 20 deals. The median deal earned +1.17x its entry equity from it.
Multiple change
The largest leg in 9 deals. For the median deal it was -0.06x: entry median 9.7x, exit median 9.9x.
Debt paydown and other cash
The largest leg in 4 deals. The median deal earned +0.07x its entry equity from it.
Operational vs financial value creation
On the 19 deals whose bridge closesThe Atlantic Deal Value Bridge splits each deal's gain into operational legs (sales, margin, cash generation, acquisitions, distributions from trading) and financial legs (multiple change, recaps, disposals, PIK, fees). On the 19 deals whose bridge closes to within 1% of entry equity, weighting every deal equally in multiples of its entry equity:
Operational 63%
Financial 37%
Across all 33 deals the split is 73% operational, 27% financial, with -1.3% of the gain not attributed, and operations were the larger part in 20 of 33. The two figures differ because the deals whose bridge is still open lean operational; the per-deal tab under The 33 deals shows each one.
Why 19, not 33: on the other 14 the entry side does not yet reconcile, each for a recorded reason (a refinancing at the close, loan notes, an entry struck after the close, fees inside the price). We publish the split on the deals where the arithmetic closes, and show the rest openly rather than leave them out.
Legs measured in multiples of entry equity. EBITDA growth is valued at the entry multiple; the multiple change is applied to exit EBITDA, so the cross term sits in the multiple leg. Debt paydown is the remainder of the equity gain, so it also carries fees and cash returned during the hold. Median years modelled: 4.
The 33 deals
The last two columns show how far the engine's result lands from the filed one. Every figure traces to a filing: each deal's inputs, its citations and the full model are in the case study library for signed-in users.
| Deal | Sponsor | Closed | Years modelled | Entry multiple | Exit multiple | Sponsor MOIC (filed) | Net debt at exit vs filed | MOIC vs filed | Tie-out |
|---|---|---|---|---|---|---|---|---|---|
| Debenhams | CVC Capital Partners / TPG / Merrill Lynch Global Private Equity | 2003 | 3 | 8.6x | 8.5x | 3.18x | -2.91% | +2.42% | Outside 1% |
| Nalco | Apollo Management / The Blackstone Group / GS Capital Partners | 2003 | 1 | 7.5x | 8.1x | 2.44x | 0.00% | +0.01% | Within 1% |
| Celanese | Blackstone | 2004 | 2 | 4.7x | 6.3x | 5.04x | -0.34% | +0.18% | Within 1% |
| SS&C | The Carlyle Group (with founder William C. Stone rollover) | 2005 | 4 | 14.0x | 10.8x | 1.73x | -0.63% | +0.71% | Within 1% |
| Freescale | Un consortium comprenant The Blackstone Group, The Carlyle Group, des fonds conseillés par Permira Advisers et TPG Capital | 2006 | 4 | 10.8x | 9.9x | 0.50x | -0.18% | +0.98% | Within 1% |
| Generac | CCMP Capital, avec Unitas Capital | 2006 | 3 | 10.4x | 9.6x | 0.69x | 0.00% | +0.03% | Within 1% |
| HCA | KKR / Bain Capital / Merrill Lynch Global PE | 2006 | 4 | 7.8x | 6.9x | 4.52x | -0.02% | -0.55% | Within 1% |
| Michaels | Bain Capital / Blackstone | 2006 | 7 | 17.6x | 8.1x | 2.11x | +0.20% | -0.31% | Within 1% |
| Nielsen | A consortium of six firms, including Blackstone, Carlyle, KKR and Thomas H. Lee | 2006 | 5 | 14.1x | 10.3x | 1.41x | +0.38% | -0.77% | Within 1% |
| Sensata | Bain Capital | 2006 | 4 | 11.7x | 18.4x | 2.65x | -0.76% | +0.98% | Within 1% |
| Allison Transmission | The Carlyle Group and Onex Corporation | 2007 | 6 | 9.7x | 9.5x | 3.12x | -0.29% | +0.70% | Within 1% |
| Aramark | GS Capital Partners, CCMP Capital, JPMorgan Partners, Thomas H. Lee and Warburg Pincus | 2007 | 7 | 8.0x | 8.2x | 2.35x | -0.03% | +0.89% | Within 1% |
| Dollar General | KKR / Goldman Sachs Capital Partners | 2007 | 3 | 10.8x | 8.0x | 2.48x | -0.09% | +0.09% | Within 1% |
| First Data | Kohlberg Kravis Roberts | 2007 | 7 | 11.5x | 12.1x | 1.12x | -0.09% | +0.21% | Within 1% |
| HD Supply | Bain Capital, The Carlyle Group and Clayton, Dubilier & Rice | 2007 | 5 | 8.5x | 12.6x | 0.90x | +0.14% | -0.52% | Within 1% |
| Hilton | Blackstone | 2007 | 6 | 15.0x | 14.4x | 2.34x | -1.58% | +2.78% | Outside 1% |
| Pinnacle Foods | The Blackstone Group | 2007 | 6 | 9.3x | 9.9x | 2.44x | -0.04% | -0.05% | Within 1% |
| TXU / EFH | KKR / TPG / Goldman Sachs Capital Partners | 2007 | 6 | 9.4x | 8.6x | 0.00x | +0.13% | 0.00x | Total loss, checked at zero |
| US Foods | Clayton, Dubilier & Rice and KKR | 2007 | 9 | 13.6x | 8.9x | 2.00x | +0.08% | -0.14% | Within 1% |
| Booz Allen Hamilton | The Carlyle Group | 2008 | 2 | 19.2x | 8.6x | 2.27x | -0.07% | +0.07% | Within 1% |
| Nord Anglia | Baring Private Equity Asia (later joined by CPPIB) | 2008 | 3 (filed window) | 15.4x | 15.2x | 3.42x | +0.07% | -0.02% | Within 1% |
| SeaWorld | Blackstone | 2009 | 3 | 6.6x | 9.7x | 2.78x | -0.24% | +0.21% | Within 1% |
| Summit Materials | Blackstone / Silverhawk | 2009 | 5 | 2.5x | 14.2x | 2.28x | +0.61% | -0.58% | Within 1% |
| Burger King | 3G Capital | 2010 | 2 | 9.2x | 11.3x | 3.32x | -0.98% | +0.54% | Within 1% |
| Dave & Buster's | Oak Hill Capital Partners | 2010 | 4 | 7.2x | 7.3x | 2.57x | -0.70% | +0.66% | Within 1% |
| IMS Health | TPG Capital, CPP Investment Board and Leonard Green & Partners | 2010 | 4 | 13.1x | 11.9x | 2.68x | -0.05% | +0.04% | Within 1% |
| Northern Tier | ACON Investments et TPG | 2010 | 2 | 2.6x | 2.3x | 6.04x | 0.00% | 0.00% | Within 1% |
| TransUnion | Advent International and Goldman Sachs | 2012 | 3 | 9.4x | 13.1x | 3.31x | -0.63% | +0.81% | Within 1% |
| Axalta | The Carlyle Group | 2013 | 2 | 7.4x | 9.3x | 3.31x | -0.44% | +0.48% | Within 1% |
| SiteOne | Clayton, Dubilier & Rice (Deere & Company a conservé 40 %) | 2013 | 2 | 5.6x | 11.3x | 3.77x | +0.27% | -0.14% | Within 1% |
| Driven Brands | Roark Capital | 2015 | 4 | 14.0x | 24.1x | 2.42x | +0.51% | -0.26% | Within 1% |
| Petco | CVC Capital / CPP Investments | 2016 | 5 | 10.4x | 13.4x | 2.12x | +0.63% | -0.89% | Within 1% |
| SolarWinds | Silver Lake and Thoma Bravo | 2016 | 2 | 17.1x | 16.2x | 1.59x | -0.41% | +0.49% | Within 1% |
See the working for any deal. In the app each case study opens as a full model: its filed inputs and their citations, every schedule year by year, and levers to change. Available on a free trial.
Start a free trialWhat "within 1%" means
- Net debt at exit: the engine's figure minus the filed figure, as a percentage of the capital deployed at entry. 31 of 33 deals land within 1%.
- MOIC: the engine's sponsor return against the filed one, measured for the holder the filing names where it names one. 30 of 32 land within 1%.
- Outside: Debenhams (net debt -2.91%, MOIC +2.42%) and Hilton (net debt -1.58%, MOIC +2.78%). We publish them rather than drop them.
- Nord Anglia was bought in 2008, but the filings cover only the last part of the hold, so it is modelled over that filed window.
- TXU / Energy Future Holdings went through bankruptcy, and the sponsors' equity was wiped out: a filed total loss. Its MOIC is checked against zero rather than as a percentage.
What this study is not
- It does not compare outcomes with what the sponsors expected when they bought. Their underwriting cases are not public, and we do not have them.
- It shows that the engine reproduces filed outcomes from filed inputs. Agreement on these measures is evidence about the engine's mechanics, not a forecast of any deal.
- A 34th case study, PetSmart and Chewy, is documented but not modelled, so it is not counted here.
Why the engine does not land exactly on the filing
A rebuilt deal can differ from its filing for reasons that have nothing to do with the arithmetic. We record each one against the deal rather than tune it away; 42 differences are declared across the corpus, each checked against the live figure.
2. Periods no filing covers
On 6 deals the weeks between closing and the first year end appear in no filing. Hilton printed no cash flow statement for 2008 or 2009, and that gap is most of its miss.
3. Measures that change during the hold
Adjusted EBITDA is restated between filings (Aramark), a 53-week year sits beside 52-week years (Driven Brands), a subsidiary leaves the group (TXU's Oncor), and short first periods are reconciled only as part of a combined year.
4. Debt as carried vs debt as owed
Balance sheets show debt net of discounts and issue costs; the engine tracks the principal owed. Debt in euros is translated at each year's rate (Nalco, Celanese).
5. Whose return is filed
A filing usually states the return of one holder, such as the sponsor's share class, not the whole company's. On 18 deals we measure on the holder the filing names.
6. Cash paid vs amounts charged
Tax and interest are charged as the cash actually paid, from the cash flow statement, not the expense in the income statement. Where a filing does not split a cash line, the difference stays visible.
7. Rounding in the filings
Most figures are printed in millions or thousands, so a rebuilt deal inherits up to half a unit of rounding on every line it reads.
Cite this study
AtlanticM&A (2026). 33 leveraged buyouts, rebuilt from their filings. atlanticma.com/research/lbo-corpus