HoldCo valuation review: the 12M Projected Valuation tab
Pad Thai Padel Group · 25 Sep 2026 · reviewed from the sheet export downloaded 00:12 today · USD m at THB 33, IDR 16,500 · every figure traces to a cell
What the tab does
It takes five sites, reads their stabilised-year cash flow before interest and tax from the club sheets, adds back maintenance capex, applies our effective share, deducts the full 2027 Thai head office at 50%, and multiplies by 7.
| Site (stabilised year used) | Revenue | EBITDA 100% | Margin | Our % | Our share | Fees the site pays |
|---|---|---|---|---|---|---|
| Thonglor (CF col G) | 2.06 | 1.03 | 50% | 50.0% | 0.51 | 0.11 |
| Rama IV (CF col G = calendar 2027) | 0.74 | 0.16 | 22% | 50.0% | 0.08 | 0.03 |
| Cloud 11 (CF col G) | 0.65 | 0.23 | 35% | 31.0% | 0.07 | 0.03 |
| Rawai (CF col F) | 0.18 | 0.07 | 41% | 12.5% | 0.01 | 0.01 |
| Bali Canggu (CF col G) | 0.96 | 0.62 | 65% | 32.0% | 0.20 | 0.09 |
| Five sites | 4.59 | 2.12 | 46% | 0.88 | 0.27 | |
| Less Thai HQ 2027 (TH OpCo!F28) at 50% | -0.68 | 50% | -0.34 | |||
| Adjusted EBITDA, our share · × 7 | 0.54 → EV 3.75 |
Cloud 11 and Rawai revenue and EBITDA above are at the OpCo's JV share (61.9% and 25%) because their club sheets scale every line by it; at 100% they are 1.05 and 0.72 of revenue. Cells: '12M Projected Valuation' G6:G10, H6:H10, I11, G13, I14, H21. Margins here are EBITDA over revenue; the tab's own check table (rows 28-34) shows 60% for Thonglor and 53% overall because its formula adds the land rent back twice (E29 subtracts CF row 8 from CF row 54, which never contained it). The tab's headline EBITDA is right; its check table is wrong by 0.34.
Your four questions
What is missing
Four things, in order of size: the two wholly-owned clubs opening in 2027 (Sathorn and Bangna, together worth more than Thonglor's share); the operator fees that HoldCo receives from every club, partners' shares included; the platform, which the group model has at $0.74m of profit in 2028 at our 95%; and the correct share of Cloud 11 and Rawai, whose club sheets already carry the JV share so the tab halves them twice. Against that, the tab also leaves out HoldCo's own costs ($0.33m) and books EBITDA before VAT and property tax, which the group model puts above EBITDA. Net, the same 7× lands at $7.8m on the tab's own rules once fixed, $9.8m with Bangna, and $3.4m only if Thonglor's THB 1,800 an hour does not hold. The platform and HoldCo's costs belong in a HoldCo view rather than a five-site tab; on that view the workbook gives $6.5m for 2027 and $15.9m for 2028.
Does deducting the Thai HQ at group level make sense
Deducting it is right in principle; the perimeter is wrong. The head office is a Golden Rulo cost that serves 8 Thai clubs in 2027 and 11 from 2028, and the group model already nets it once across all of them before our 50%. The tab lifts the whole line and sets it against four Thai clubs plus Bali, which carries no Thai HQ at all. On your point that it should be attributed proportionately: yes, if the perimeter stays at five sites, then by revenue those four clubs carry 71% of the 2027 line and 48% of 2028, so the charge at our 50% is $0.24m or $0.17m, not $0.34m. Two of three refuters preferred the other fix: widen the perimeter to the clubs the HQ actually serves (Sathorn now, Bangna next) and keep the full charge, because pro-rating leaves half the cost against clubs the tab never values. Either is defensible; five sites with the full HQ is not. On the size, "cannot be that high" does not hold against benchmarks: THB 1.8m a month is 10.7% of Thai network revenue in 2027 falling to 6.2% in 2029, inside the 6 to 11% that Basic-Fit, The Gym Group and Life Time run, and $82k per club. It only looks enormous measured against five sites. And the head office is the operator that earns the fees: Thai fees alone cover 44% of it in 2027 and 65% in 2029, all fees cover it entirely by 2029.
Where are the fees
Deducted inside every club sheet (rows 38 and 39: 2% of revenue plus 7 to 10% of pre-fee profit) and booked as HoldCo income in the group model (TH OpCo row 62, ID OpCo row 61, HoldCo rows 12 to 14: $0.35m in 2027, $0.54m in 2028, $0.73m in 2029). The valuation tab never adds them back. For the five sites in the tab that is $0.30m a year at 100%, of which $0.26m reaches HoldCo. Deducting the head office while dropping the fees that fund it counts the same cost twice; it is the single largest gap in the tab.
Are the numbers in line with our projections and with the market
Site by site the tab reconciles to the club sheets, so the gap to our own projections is structural, not a site-level error. Two site assumptions are above market: Thonglor at THB 1,800 an hour and 65% occupancy is 2.25× the yield Rama IV actually realises and above every Bangkok list price found (ours is 1,400), and Bali's 65% margin is flattered by a prepaid lease and no local overhead. Rama IV itself is sliding: 82% occupancy in 2025 to 57% in July 2026, revenue down 38% year on year in April to July, and the club sheet now runs off a single December of costs because the actuals rebuild broke its references. The 7× is at or above what listed fitness operators trade on; no padel operator has sold with a disclosed multiple. The valuation is low because of the deductions, not the multiple.
Bridge from the tab to a consistent number
Each step below survived a three-way refutation pass (formula check, valuation method, materiality). Steps that failed it are listed afterwards.
| USD m, our share, stabilised year | Step | EBITDA | EV at 7× |
|---|---|---|---|
| Tab as built (I14, H21) | 0.54 | 3.75 | |
| Cloud 11 and Rawai at 50% instead of share × 50% (their sheets already carry the JV share) | +0.07 | 0.61 | 4.25 |
| Add the operator fees these five clubs pay that reach HoldCo | +0.26 | 0.87 | 6.09 |
| Deduct VAT and property tax (club-sheet row 80), as the group model's EBITDA does | -0.16 | 0.71 | 4.99 |
| Five sites, tab's rules, full HQ kept | 0.71 | 5.0 | |
| Add Sathorn: 100% owned, first full year 2027, the same rule that admits Thonglor and Rawai (share 0.32, fees 0.10, its taxes -0.03) | +0.40 | 1.11 | 7.76 |
| Base: all 2027 owned clubs, full HQ kept | 1.11 | 7.8 | |
| Add Bangna: 100% owned, contracted, opening July 2027, first full year 2028 | +0.17 | 1.28 | 8.95 |
| Pro-rate the HQ to the six Thai sites' share of 2028 Thai revenue (77%) | +0.07 | 1.35 | 9.44 |
| Rama IV on trailing twelve months with corporate events net of their costs | +0.05 | 1.39 | 9.76 |
| High | 1.39 | 9.8 | |
| Low, from the five-site line: Thonglor at 1.5× Rama IV's realised yield (about THB 1,100 to 1,200 blended) and Cloud 11 at 50% | -0.22 | 0.49 | 3.44 |
Fees 0.26 = Thonglor 0.11, Rama IV 0.03, Cloud 11 0.05 and Rawai 0.02 grossed to 100%, Bali 0.09 × 60%. Taxes from club-sheet row 80 at the corrected shares. Sathorn and Bangna use the tab's own method (cash flow before interest and tax plus maintenance capex, first full year, 50%). The platform (95% of $0.74m profit in 2028) and HoldCo's own costs ($0.33m) are deliberately outside this bridge; they belong in the HoldCo view below. The deck's $9.0m ask is a sum of parts that also prices the platform, the brand and fee stream and the strategic partnerships; this tab only prices site stakes.
What the refutation pass rejected
- Pro-rating the HQ while keeping five sites (2 of 3 against): it moves half the cost onto clubs the tab never values. Widen the perimeter instead, or pro-rate only once you have.
- Adding the platform and HoldCo costs into a five-site site valuation (2 of 3 against): different asset, different basis; shown separately below.
- Calling the HQ-in-TH-OpCo, fees-to-HoldCo split an error (3 of 3 against): it is how the Read Me defines the structure. Kept as a diligence question.
- Treating the broken Rama IV references as a large valuation error (2 of 3 against the size): the reference is broken, but one March of revenue and one December of costs roughly offset; a trailing-twelve-month basis gives 0.18 against the tab's 0.16.
- Concluding that Rama IV's 2026 slide proves its stabilised EBITDA must be below 0.16 (3 of 3 against the conclusion, none against the data).
- Treating the 7% fee overrides and the stabilised-year labels as valuation defects (2 of 3 against each): real inconsistencies, small effect, listed under defects and open questions.
Head office: how big it is and where it should sit
| USD m | 2027 | 2028 | 2029 |
|---|---|---|---|
| Thai head office (TH OpCo!28; THB 1.806m a month, +3.5% a year) | 0.68 | 0.70 | 0.73 |
| Thai clubs open (TH OpCo!6) | 8 | 11 | 11 |
| Thai network revenue at 100% (TH OpCo!8) | 6.33 | 9.54 | 11.83 |
| HQ as % of Thai network revenue | 10.7% | 7.4% | 6.2% |
| Group network revenue at 100% (Summary!8) | 7.26 | 12.47 | 19.20 |
| HQ as % of group network revenue | 9.4% | 5.6% | 3.8% |
| Operator fees reaching HoldCo (TH OpCo!62 + ID OpCo!61 + HoldCo!14) | 0.35 | 0.54 | 0.73 |
| HQ net of the fees it earns | 0.33 | 0.16 | 0.00 |
| HoldCo's own costs (HoldCo!24), not in the tab | 0.33 | 0.34 | 0.35 |
The four Thai sites in the tab carry 57% of 2027 Thai network revenue (3.63 of 6.33) but the tab charges them 100% of the head office. Allocated by revenue the charge is 0.39 at 100% (0.20 at our 50%), not 0.68 (0.34). By club count it is 4 of 8 in 2027 and 4 of 11 in 2028. And the head office is the operator that earns the fees: by 2029 the fee income covers it entirely, so a valuation that deducts the HQ and drops the fees counts the same cost twice.
One design point to have an answer for, not an error: the head office cost sits in TH OpCo, where the Thai partner bears half of it, while the operator fees flow 100% to HoldCo. The Read Me states it that way on purpose. A JV partner doing diligence will ask why they fund half the operator and receive none of its fees; the clean structures are a management-services charge from HoldCo, or the fees staying in Thailand.
The HoldCo view: the model's own answer by year
| USD m, HoldCo proportionate | 2027 | 2028 | 2029 |
|---|---|---|---|
| 50% × Thailand OpCo EBITDA (already after the full HQ) | 0.37 | 0.83 | 0.98 |
| 32% × Indonesia OpCo EBITDA | 0.18 | 0.38 | 0.42 |
| Malaysia, Singapore, Brunei (entity sheets already at our club shares, HoldCo owns each entity) | -0.02 | 0.12 | 0.53 |
| Operator fees to HoldCo, 100% | 0.35 | 0.54 | 0.73 |
| Platform, 95% share of profit (HoldCo!11) | 0.37 | 0.74 | 0.96 |
| HoldCo's own costs (HoldCo!24) | -0.33 | -0.34 | -0.35 |
| Proportionate EBITDA | 0.92 | 2.27 | 3.26 |
| × 7 | 6.5 | 15.9 | 22.8 |
| HoldCo net income (HoldCo!35), for reference | 0.47 | 1.32 | 1.85 |
The platform line is profit after its own costs, so it behaves like EBITDA here; a buyer would price it on a different basis than clubs. 2027 still carries $0.37m of pre-opening costs at 100% inside the Thai line. The deck's $9.0m sits between the 2027 and 2028 figures. Never quote an EV/EBITDA multiple to investors; this table exists to show what the tab leaves out, not to set a price.
What Rama IV actually does (THB, 'Rama IV Actuals')
| THB m | Calendar 2025 | Trailing 12M to Jul 2026 | 2026 Jan–Jul | 2026 run-rate |
|---|---|---|---|---|
| Revenue (row 20) | 26.55 | 24.02 | 13.11 | 22.5 |
| of which courts (row 15) | 10.56 | 9.16 | 4.90 | 8.4 |
| of which corporate events (row 16) | 4.48 | 2.69 | 1.22 | 2.1 |
| Expenses (row 67) | 17.12 | 16.97 | 10.22 | 17.5 |
| Profit (row 6) | 9.42 (35%) | 7.04 (29%) | 2.89 (22%) | 4.95 (22%) |
| In USD k | 805 rev / 285 profit | 728 / 213 | 397 / 88 | 681 / 150 |
Monthly revenue fell from THB 2.73m in January 2026 to 1.14m in July, and July was a loss. The tab's Rama IV line (0.74 revenue, 0.16 EBITDA) is the 23-month average times 12, which is close to the 2026 run-rate; it is not the 2025 year. On three courts Rama IV earns about USD 93–107k of court income per court a year, an effective THB 545–629 per court-hour.
| Court assumptions vs Rama IV | Courts | THB/hour | Occupancy | Effective THB per court-hour | Court revenue per court, USD k | Total revenue per padel court, USD k |
|---|---|---|---|---|---|---|
| Rama IV actual 2025 / TTM | 3 | ~1,100 | 57% / 50% implied | 629 / 545 | 107 / 93 | 268 / 243 |
| Thonglor (Thonglor Inputs C5, C7, C9) | 6 | 1,800 | 65% | 1,175 | 199 | 343 |
| Cloud 11 (3 padel + 2 pickleball), at 100% | 3 | 1,100 | 63% | 694 | 110 | 350 |
| Rawai, at 100% | 4 | 1,100 | 50% | 547 | 93 | 180 |
| Bali Canggu (IDR at 500 per THB) | 5 | 950 | 65% | 618 | 111 | 193 |
Cloud 11, Rawai and Bali sit at or below what Rama IV has already achieved per court-hour. Thonglor assumes 1.9× Rama IV's realised yield per court-hour (a THB 1,800 blended rate at 65%), which is the single assumption an investor will test hardest.
Standard performance expectations
Public multi-site operators run head office at 6 to 11% of revenue; hotel-style management contracts charge 2 to 4% of revenue plus 10 to 20% of cash flow, and the owner's EBITDA is measured after those fees with no further operator overhead. No padel operator has changed hands with a disclosed EBITDA multiple.
| Metric | Benchmark | This model | Verdict |
|---|---|---|---|
| Head office as % of revenue | Basic-Fit FY2025 6.3% ex marketing, 11.1% with it; The Gym Group FY2025 11.3% (£27.7m on £244.9m), guided below 11%; Life Time 8.2% | 10.7% of Thai network revenue in 2027, 7.4% in 2028, 6.2% in 2029 | Normal size from 2028; the tab's problem is the base it is charged to, not the amount |
| Head office per club | The Gym Group £107k; Basic-Fit €54k | $85k (8 clubs, 2027), $64k (11 clubs, 2028) | In line |
| Operator fee terms | Hotel management agreements: base 2 to 4% of revenue, incentive 10 to 20% of cash flow (HVS) | 2% of revenue plus 7 to 10% of pre-fee profit | Standard; owner EBITDA is after fees, and the operator's HQ is what the fees pay for |
| EV/EBITDA, fitness and leisure operators | Basic-Fit 7.1×, Xponential 6.7×, Smart Fit 8.5×, Planet Fitness 13.4× (franchisor); David Lloyd reported sale range implies 7.8 to 10×; private gyms 3 to 6× | 7× on projected year-2 EBITDA | At or above market for five sites, two unopened; the low number is not the multiple |
| Padel deal comps | Game4Padel £27m (Feb 2025), Padel Haus $33m post (Jul 2025): venture rounds, no EBITDA disclosed | n/a | No EV/EBITDA comp exists; say so rather than imply one |
| Club EBITDA margin | 28 to 35% facility level in mature markets; 15 to 40% at 6 to 12 courts; under 20% at 3 to 4 courts; Basic-Fit club EBITDA less rent 36% | Thonglor 50%, Bali 65%, Rawai 41%, Cloud 11 35%, Rama IV 22% (actual 35% in 2025, 22% in 2026) | Thonglor and Bali above the top of the range; the others in range |
| Court occupancy | Year 1 40 to 45%; break-even 55 to 65%; 60 to 70% by year 3 | Thonglor 65%, Cloud 11 63%, Bali 65%, Rawai 50%; Rama IV actual 82% (2025), 68% (2026 to July), 57% (July 2026) | Defensible for Bangkok, with the 2026 slide as the warning |
| Bangkok court price per hour | No Drama 900 to 1,200; Padel Tennis 800 to 1,200; Pad Thai Padel list 1,400; market commentary 1,200 to 1,500 | Thonglor 1,800 blended; Rama IV realised 801 (2025) and 711 (2026) per occupied hour | Thonglor is above every list price found and 2.25× our own realised yield |
| Rent as % of revenue | Basic-Fit 21%, The Gym Group 17% | Thonglor 10.6%, Cloud 11 17%, Rawai 5%, Bali 0% (ten-year lease prepaid, sits outside EBITDA) | Bali's margin is flattered by the prepaid lease and by carrying no Indonesian overhead |
Sources: Basic-Fit FY2025 results release (11 Mar 2026); The Gym Group 2025 full-year results RNS (I read this one myself); Life Time Q4 2025 release; HVS "A new approach to hotel management fees"; valueinvesting.io and multiples.vc public comps (Sep 2026); tennis365 and Crunchbase for Game4Padel and Padel Haus; racquetsports.institute, sheets.market, dojobusiness.com and padelcreations.com for padel club economics; nodramapadel.com, playerbox.co and Time Out Bangkok for court prices. A search hit claiming "padel operator 9 to 13× EV/EBITDA" turned out to be an aerospace SPAC filing and was discarded.
Defects in the tab and in the edited sheet
- HighH8, H9: JV share applied twice. The Cloud 11 and Rawai club sheets already scale every line by their share (
'Cloud 11 CF'!$C$20.619,'Rawai CF'!$C$20.25), so multiplying again by share × 50% leaves us with 19% and 3% instead of 31% and 12.5%. Fix: H8 = H9 = Inputs!C19. Worth +0.07 EBITDA. - HighRow 13: full Thai HQ against a subset, no fee line. Charge HQ pro rata to the sites in the perimeter (or include every Thai club it serves) and add a row "(+) operator fees paid by these clubs to HoldCo" at 100%.
- High'Rama IV CF' rows 32 to 37, 40, 62, 63, 80: stale columns after the actuals rebuild. Column R was "Average" in the 16 Sep file and is now December 2025; column I is March 2025. Rows were re-pointed, columns were not, so Rama IV runs on one December of costs and one March of court revenue, each × 12. Repoint to the Average column (AB) or a trailing-12-month column. Corporate events (row 16, THB 2.7m trailing) never enter the cash flow while December's event expenses do.
- WatchRows 28 to 34: the check table does not check. E29:E33 subtract land (CF row 8) from an opex total (CF row 54) that never contained it, so G34 (2.44) misses I34 (2.10) by exactly the land rent and the margins read 60% and 52%. True margins: Thonglor 50%, Cloud 11 35%.
- WatchC6:C10 stabilised-year labels. Rama IV column G is year 2 (calendar 2027), not "Yr 3". Per Sites, Thonglor, Cloud 11 and Bali have 2027 as first full year, so column G is 2028, and Rawai alone uses column F. The "Nov 2027 / Oct 2027 / Jul 2027" dates come from CF start dates edited in the sheet that nothing else in the model reads.
- WatchIncentive fee 7% typed over the group input.
'Thonglor Inputs'!C53,'Cloud 11 Inputs'!C60,'Rawai Inputs'!C55and'Bangna Inputs'!C65were 10% via the named input on 16 Sep; Rama IV, Bali, Sathorn, Rama III and the JV template still use 10%. One fee term for the network, please. - WatchCloud 11 share 61.9% (
'Cloud 11 Inputs'!C61, was 50% on 16 Sep) while the Sites note still says 50%. - WatchEBITDA definition. The club sheets put VAT and property tax below the cash flow the tab uses (row 80), while TH OpCo books them above EBITDA (row 27). The tab's site EBITDA is about 0.26 at 100% (0.12 to 0.16 our share) higher than the group model's own definition of the same clubs.
- Note'TH HQ Costs'!C25 (year-3 roles, THB 425k a month) is not wired into TH OpCo row 28. D20 is before HQ while D21 is after; G14 feeds nothing; the 7× is typed in F20, F21 and again in C24:K24. Rama IV maintenance capex (F40) is one month's revenue, not twelve.
- Note'Thonglor Inputs'!C9 occupancy is typed as 0.653 over a build-up (K17) that still computes 0.609 at 75% peak and 50% off-peak.
What changed in the sheet since the 16 Sep investor copy
Besides the new tab: Thonglor court price THB 1,500 to 1,800, occupancy link replaced by 0.653, F&B covers 80 to 110 and F&B staff 4 to 6, wellness removed, sports-centre staff 14 to 8, padel build cost THB 15m to 47m and wellness 8m to 2m; Rawai F&B covers 25 to 50 and ticket 300 to 400; Cloud 11 share 0.5 to 0.619; club cash-flow start dates; the 7% fee; and the Rama IV actuals rebuilt to 23 months, with every overlapping month's expenses restated. Thonglor's year-2 cash flow rose 41% from those edits (0.72 to 1.02) and now carries most of the Thai value in the tab.
Only you can answer
- Who built the tab and for whom: the 7× reads like the Sansiri note's fallback reference, but the "PTP proportionate" basis is a HoldCo view.
- Where the head office belongs. Today its cost sits in Golden Rulo (Thai partner pays half) while the fees it earns go 100% to HoldCo. Either the team moves up with the fees, or the fees stay in Thailand.
- Cloud 11: 50% or 61.9%.
- Incentive fee: 7% or 10%, one number.
- Thonglor at THB 1,800 an hour and 65%: intended price point, or a scenario someone typed in.
- Rama IV corporate events: recurring revenue or not, and whether July 2026 is a complete month.
Method: the sheet's export (45 tabs) was diffed cell by cell against the 16 Sep investor copy; every figure above was recomputed in Python from the cached values; five independent reviewers produced 55 findings, deduplicated to 36; the 23 material ones each faced three refuters (formula, method, materiality) and 14 survived. No file was modified.