← Writing

Is Beachcomber worth buying? I built the comp set to find out.

A trading comparables set for New Mauritius Hotels, built from audited filings. Along the way: Rs 1.7bn of enterprise value that did not exist, Rs 9.3bn of state money hiding inside equity, and two companies on an identical P/E that are nothing like each other.

2 September 2026 · Equity research · 14 min read · Sources: audited annual reports, SEM Factbook 2026, CBRIS

The short version

New Mauritius Hotels trades at Rs 13.95. On this evidence I would not buy it at that price.

It is priced above every one of its listed peers on enterprise multiples, and it carries the heaviest debt load of the nine companies I looked at. You are paying a premium for the balance sheet least able to absorb a bad season.

5.63x

EV/EBITDA vs peer median 4.47x

3.05x

net debt / EBITDA — highest in the set

Rs 11.79

what the peer median implies

−15%

below the traded price

The interesting part was not the answer. It was that the spreadsheet produced two wrong numbers first, and neither was an arithmetic mistake.

Everything below comes from these filings. They are public — open them alongside and check me.

New Mauritius Hotels FY2025 Sun FY2025 Lux Island Resorts FY2025 Constance FY2024 Beau Vallon FY2024 Morning Light FY2025 Riveo FY2025

The question was simple. New Mauritius Hotels — Beachcomber, the largest hotel group in the country — trades at Rs 13.95 a share. Is that cheap?

There is only one honest way to answer it: find every company that does the same thing, work out what the market pays for each rupee they earn, and see whether the target sits above or below them. That is a trading comparables set. Nine companies, every figure taken from audited accounts and page-cited. It took about a week.

The answer, when it came, was less interesting than what I had to fix to get there. Two of the numbers the spreadsheet produced were wrong, and neither was an arithmetic mistake. A third pair of numbers was identical and meant completely different things.

This post is about that: how much of a published financial number is a choice rather than a fact, and how you find out before you put money behind it.

What I found on the way to that answer

Two listed companies own the same hotels, and one came out Rs 1.7bn bigger. Hotelest holds 51% of Constance and does nothing else. One wrong input caused it; correct that input and the two land Rs 0.41m apart. See it
Sun and Lux sit on the identical P/E of 5.90 and are not remotely alike. On enterprise value it is 6.08x against 3.77x. The P/E hides the debt; the enterprise multiple shows it. See it
Rs 9.3bn of Bank of Mauritius Covid money sits inside shareholders’ equity, not debt. Sun prints gearing of 16.8% and 39.6% on the same page of the same report. Both are correct. See it
Two of the nine companies never publish a depreciation line at all, so no EBITDA can be derived for them. The peer statistics here rest on four companies, and every one of them says so on its face. See it
Beachcomber is priced above the 75th percentile of its listed peers. That is the arithmetic behind the answer above, built step by step. See it

Step oneChoosing who counts as a peer

Before a single number, you have to defend the list. The Stock Exchange of Mauritius files companies under a sector label, and that label is not reliable. I tested each candidate against its registered nature of business in the companies registry instead — the same source I use in Project Nightfall to trace who owns what in Mauritius.

Three got thrown out:

Stevenhills — filed by the exchange under Leisure and Hotels. The registry describes it as a bookmaker taking fixed-odds bets on football played outside Mauritius. Not a hotel.
Attitude Property — a landlord whose tenants happen to be hotels. Real estate.
Beachcomber Hospitality Investments — preference shares only. There is no ordinary share price, so no multiple can exist.

The habit. A peer set is defended in words before it is defended in numbers. If you cannot write one sentence saying why a company belongs, it does not.

Step twoWhat you are actually paying for

Buy every share in a company and you do not own it outright. You inherit its debt, you inherit the outside shareholders sitting inside its subsidiaries, and you get to keep the cash already in the till. Enterprise value is the price of the whole business once all of that is settled.

Sun Limited, FY June 2025. Market capitalisation is only 60% of what the enterprise costs. Source: Sun Limited Annual Report 2025, statements of financial position p.82 and note 16.

How to read thisRead it left to right, like a staircase. You start with what the shares cost, then each step adds another claim on the business. Each bar begins where the last one ended — that is why they float. The red bar is subtracted, because cash the company already holds reduces what you effectively pay. The black bar at the end is the total.

Note the fourth bar. Almost Rs 2bn of Sun’s enterprise value is a convertible bond that its own balance sheet files under equity, not debt. Hold that thought — it becomes the biggest finding in the whole exercise.

The rule that stops you fooling yourself

Once you have two different measures of size, you can pair them with the wrong profit figure and produce a number that means nothing.

Enterprise value belongs to everybody with a claim, so it pairs with profit measured before any of them are paid. Market capitalisation belongs to shareholders alone, so it pairs with what is left after interest and tax.

How to read thisTwo boxes, two different pots of money. The blue box is everyone who has a claim on the business; the black box is only the shareholders. Under each box are the profit measures that belong to that same group. Always take one number from a box and one from the list beneath it — never mix across.

Cross them — divide enterprise value by net income — and you charge the shareholders for the lenders’ money as well. The more indebted company then looks dearer for no operating reason at all.

Step threeThe same number can mean two different companies

This is the part nobody warns you about. A multiple is a ratio, and a ratio throws away everything except its two inputs. Two companies can land on exactly the same number and be nothing alike.

Sun and Lux are the cleanest example in the set. Both are Mauritian resort operators. Both report to June 2025. Both were priced on the same day. And both trade on a price-earnings ratio of 5.90x — identical to two decimal places.

Same P/E, same sector, same year end, same pricing date. Everything below the first row diverges. Sources: Sun AR2025; Lux Island Resorts AR2025; SEM Factbook 2026 pp.24–25.

How to read thisTwo companies side by side, same five measures. The green row at the top is the one where they are identical. Every row underneath it is where they differ. The point of the chart is how far apart two businesses can be while showing a shareholder the same headline number.

On EV/EBITDA one costs 6.08x and the other 3.77x. A 61% difference in what the enterprise costs per rupee of operating cash — between two companies a shareholder is paying the same earnings multiple for.

The reason is not performance. It is who funded the business.

Gross funding split. Sun’s enterprise is 55% shareholder money; Constance’s is 68% borrowed. The right-hand column is net debt divided by EBITDA.

How to read thisEvery bar is the same length, and each represents all of the money funding that company — 100%. The colours show where that money came from: blue is shareholders, black is banks, pale blue is the state bond. A long black section means a business built mostly on borrowing. The figure on the right is how many years of profit it would take to repay the debt.

Read across those bars and the peer set stops looking like one sector:

Sun and Lux are equity-funded businesses. Debt is 18% and 27% of funding; net debt is under three-quarters of one year’s EBITDA. When you buy the equity you are buying most of the enterprise.
Constance is a debt-funded business. 68% of its funding is borrowed. Its equity is a thin slice on top of Rs 5.4bn of borrowings — which is exactly why it trades at 0.31x book while Sun trades at 2.02x.
Beau Vallon and NMH sit in between, at 53% and 57% debt.

What this does to a multiple. A cheap-looking EV/EBITDA on a heavily geared company is not the same bargain as the identical number on an ungeared one. In the first case most of the enterprise you are buying already belongs to a bank. The equity is a smaller, riskier slice of the same hotels.

Which is why leverage is the number I would check first

Net debt ÷ EBITDA. Lux 0.63x, Sun 0.72x, Beau Vallon 2.39x, Constance 2.78x, New Mauritius Hotels 3.05x. Computed from each company’s audited borrowings and cash notes.

How to read thisOne bar per company: how many years of EBITDA it would take to repay their debt, net of cash. Shorter is safer. The pink zone past 3x is the rough level at which lenders start attaching conditions to what a company may do.

Sun could repay every rupee it owes, net of cash, out of roughly eight months of EBITDA. New Mauritius Hotels would need three years. Same industry, same island, same guests — entirely different exposure to a bad season.

Hold this against the valuation. The target is the most levered company in its own peer set. Any premium it trades at has to be justified despite that, not in ignorance of it.

The first lieRs 1.7 billion that was not there

Two of the nine are the same business. Hotelest Limited exists to hold 51% of Constance Hotels Services. Its accounts say so in one line:

The only activity of Hotelest Limited is to hold 51% of the share capital of Constance Hotels Services Limited.

Hotelest Limited, abridged audited financial statements, year ended 31 December 2024, note 1

Because Hotelest controls Constance, IFRS 10 makes it report Constance whole — every rupee of revenue, every hotel, every liability, at 100% rather than 51%. So both companies’ accounts describe the same hotels, to the rupee.

How consolidation actually works — open if this is new

Hotelest has no hotels, no staff and no revenue of its own. Every operating line in its accounts arrives through consolidation. Source: Hotelest Limited abridged FS FY2024; Constance Hotels Services abridged FS FY2024.

How to read thisTwo versions of the same company. On the left, what Hotelest’s accounts would say if it simply held shares. On the right, what they actually say. The profit at the bottom is the same in both. Everything above it changes.

The trigger is control, not the size of the stake. More than half the votes usually gives it, which is why 51% is the number people quote — but a stake can clear 50% without control, and control can exist below it. Below control you get the one-line treatment instead: Constance’s own filing consolidates its subsidiaries in full while its overseas associates appear as a single line, Share of results of associates 206,087.

The 49% Hotelest does not own comes back once, at the very bottom, and it ties exactly:

Hotelest owners’ profit 0.51 × 314,379 Constance owners = 160,333 less Hotelest’s own running costs = (3,536) --------- 156,797 ✓ as printed Non-controlling interests 0.49 × 314,379 Constance owners = 154,046 plus Constance’s own minorities = 22,166 --------- 176,212 ✓ as printed

Same hotels. Same debt. So the two enterprise values should match. They did not.

Constance and Hotelest, 31 December 2024. The middle bar is what a mechanical calculation produces. The right-hand bar is what happens when one input is corrected.

How to read thisThree bars, but only two companies. The black bar is Constance. The middle red bar and the right green bar are the same company, Hotelest, calculated two ways. The green dashed line is where both should land.

The whole gap is one line: the minority interest, the 49% of Constance that Hotelest does not own. Both companies add it to enterprise value. They just price it differently — and one of the two prices is sitting on a screen.

Hotelest holds 51% of Constance and nothing else. Sources: Hotelest Limited abridged financial statements FY2024 (minority interest Rs 2,458.59m); Constance closing price at 31 December 2025, SEM Factbook 2025, DEM table p.35.

How to read thisThe top half is who owns what. The bottom half prices that 49% twice on the same day — once the way the balance sheet does, once the way the market does. The red bar is 3.2 times the green one.

Constance trades at 0.31 times book, so carrying its minority at book triples it. Here is the whole thing reconciled — the overstatement, and what happens when you remove it:

THE CAUSE · Rs million Hotelest’s minority interest, at book 2,458.59 49% of Constance’s actual market cap 765.65 --------- overstated by 1,692.94 = 3.2x too big THE EFFECT Hotelest enterprise value, as first built 9,172.98 less that overstatement (1,692.94) --------- Hotelest, corrected 7,480.04 Constance, built independently from its own filing 7,480.45 --------- still apart by 0.41 = 0.01%

Two numbers, 0.41 apart — do not confuse them. The bracket on the chart above measures 1,692.53: the gap you can actually see between Constance and Hotelest-at-book. The calculation here gives 1,692.94: the overstatement that caused it. The 0.41 between them is simply how close the two companies land once the input is fixed — and it is the reason to trust the result. When two routes to the same business agree to 0.01%, the method is sound and the earlier number was wrong.

Why a minority belongs in enterprise value at all

EV/EBITDA is a fraction, and consolidation has already decided the bottom of it: EBITDA holds every hotel. So the top has to buy every hotel too. Count only what the parent’s own shareholders own and you are pricing half the hotels against all of the profit.

Hotelest Limited, FY Dec-24. EBITDA Rs 1,791.2m. Source: Hotelest Limited abridged financial statements FY2024; enterprise value built in the model.

How to read thisThe blue squares on top are the share of the hotels the enterprise value pays for. The black squares underneath are the share the profit comes from. On the left they do not match. That mismatch costs a full multiple turn.

The rule that falls out of it. A comp set holding both a listed parent and its listed subsidiary has to mark the minority to market — or drop one of the two rows.

The second lieRs 9.3 billion hiding inside equity

During Covid the Mauritius Investment Corporation — owned by the Bank of Mauritius — bought convertible bonds from almost every hotel group in the country. Five years on the money is still there, and almost all of it is classified as equity.

Convertible bonds and preference shares as a share of enterprise value. The right-hand column shows EV/EBITDA computed with those claims included, then excluded. Sources: each company’s FY2024 or FY2025 audited accounts.

How to read thisOne bar per company, scaled to the size of that business. The blue tip is the state bond. On the right, the same company’s multiple with the bond counted, then ignored.

Between 13% and 17% of every enterprise value in the sector is this one instrument. Count it as a claim and the multiples are one set of numbers; ignore it and they are another, half a turn to a full turn lower.

And nobody is hiding it. Three companies tell you, three different ways, that this is a decision rather than a fact.

One: it says so outright

For the classification and measurement of the funds received from the Mauritius Investment Corporation Ltd, a policy choice is available… The Directors have opted to treat the convertible bonds as equity…

Constance Hotels Services Limited, Annual Report 2024, note 2(t)

Two: it prints both answers

Sun’s narrative reports gearing of 16.8%. The table beside it, on the same page, reports 39.6%. Both are correct. They differ on one question: is a Bank of Mauritius convertible debt?

Sun Limited Annual Report 2025, chief finance officer’s report. Two numbers for the same balance sheet, printed side by side.

How to read thisTwo bars, one company, one page. Same balance sheet, same date. The only difference is whether the Bank of Mauritius bond counts as debt.

Three: it builds around it

New Mauritius Hotels is moving its Moroccan resort into a separate company and selling 51% of it — “to share risk, fund the extension project and repatriate capital to Mauritius” — then operating the hotel under a management agreement with Fairmont. Sell control and the resort’s debt leaves your balance sheet with it. On Zanzibar the report is blunter still: the acquisition will be funded “in line with our gearing target”. New Mauritius Hotels Limited, Annual Report 2025, chief executive’s interview, p.8.

The Morocco structure, drawn

The structure as described by NMH. Source: New Mauritius Hotels Limited, Annual Report 2025, chief executive’s interview, p.8.

How to read thisLeft is what NMH has today: it owns the resort outright, so the resort’s debt is its debt. Right is what it becomes. The building never moves. Only the ownership around it does.

In plain terms, NMH stops owning the hotel and starts running it for a fee. It takes cash out, keeps 49%, and the debt that built the resort leaves with the 51% it sold — because what you no longer control, you no longer consolidate, and what you do not consolidate never reaches your gearing ratio. The report gives the structure and the gearing target; the deconsolidation is the standard consequence of losing control.

None of this is deception. Every figure is disclosed, audited and correct. But three of the largest hotel groups in the country are telling you, in their own words, that the balance sheet is a set of decisions. Read the accounting policy note before you trust a ratio.

Step fourWhat the filings simply do not tell you

Mauritius has tiered disclosure. Companies on the Official Market publish full annual reports. Companies on the DEM can satisfy the rules with a single page. Two of my nine publish accounts with no depreciation line at all — which means EBITDA cannot be built from them.

companies screened

usable in the EV/EBITDA median

93%

of sector market cap those four cover

that disclose a room count

The temptation is to estimate the missing depreciation from peers and manufacture an EBITDA. That would put an invented number into the denominator of the headline multiple. I left the cells blank and published the observation count under every statistic instead.

A blank cell is a finding. An invented one is a liability. Four observations covering 93% of the sector’s value is a defensible peer statistic, as long as you say “n = 4” out loud.

Step fiveWhat the reports do tell you, if you read them

Owning a hotel and running one are different businesses

Lux Island Resorts keeps 26.0% of what its guests spend. Sun keeps 31.8%. Nearly six points apart, which reads like Sun simply runs better hotels.

It is not that. The two companies are not doing the same job.

Sun owns its resorts and runs them. The people at the front desk are Sun’s own staff, and their wages sit inside Sun’s costs. Lux owns its hotels but does not run them. It pays another company to do that — and that company belongs to the same parent group.

The management of the different hotels is entrusted to its sister company, The Lux Collective Ltd, under a long term management contract.

Lux Island Resorts Ltd, Integrated Annual Report 2025, note 1

Lux pays Rs 504.18m — 4.8% of revenue — to a company inside the same parent group for the job Sun does in-house. Source: Lux Island Resorts Integrated Annual Report 2025, note 1 (the management contract) and note 26 Other Operating Expenses (the Rs 504,180 thousand).

How to read thisThe top half shows why the fee exists: Sun is one company doing both jobs, while Lux and its operator are two companies under one parent, with money moving between them. The bottom half is the consequence. The shaded block on the middle bar is that fee — put back, to show what Lux would keep if it ran its own hotels the way Sun does.

So Lux pays Rs 504m for a job Sun does in-house. That fee is an expense, so it comes out of Lux’s profit before the margin is calculated. Sun has the same job to do, but pays its own staff, and those wages are already inside its costs too — the difference is only whose income statement the money passes through.

Put the fee back and the comparison becomes fair:

Lux EBITDA 2,741.79 + management fee paid out 504.18 AR2025 note 26 --------- = what Lux would keep in-house 3,245.97 ÷ revenue 10,555.61 --------- = 30.8% against Sun at 31.8%

But does the fee not just cover the wages? No — and this is worth checking before you trust the add-back. Lux carries its own payroll: note 25 shows Rs 2,593m of employee benefit expenses, 24.6% of revenue, an ordinary hotel wage bill. The Rs 504m management fee sits on top of that. Lux employs the people; it pays its sister for the brand and the running of the place. So if Lux operated its own hotels it would keep the fee and still pay the same wages — which is exactly what the add-back assumes.

The six-point gap is really one point. The other 4.8 points never measured hotel quality at all — they measured a decision about which company in the group signs the payroll. And because both companies sit under the same parent, that fee is set between related parties rather than by a market.

Every hotel in the country is halfway through a refurbishment

Riveo closed Four Seasons Anahita for seven months and booked Rs 128m of closure costs. Morning Light lost Rs 53m in 2024 “following a slow start post-renovation”, then made Rs 52m in 2025 on the same building. Sun, Lux, Constance and NMH are all mid-programme.

Consequence. A trailing-twelve-month EBITDA multiple for a hotel is partly a bet on where that hotel sits in its refurbishment cycle. Morning Light is the clean proof: same asset, same management, loss one year and profit the next, with the renovation as the only variable.

A Mauritian hotel comp set is partly a Maldives comp set

Company	Outside Mauritius
Lux Island Resorts	Réunion and Maldives — 30.6% of EBITDA
Constance Hotels Services	Maldives 35.8% of revenue; Seychelles and Madagascar associates are 40% of pre-tax profit
New Mauritius Hotels	Morocco and Europe — 19.7% of revenue

The verdictSo — is it worth buying?

Every number, in one table

This is the model itself — the four Tier 1 peers, then Beachcomber held out as the target. Rs million, prices at 31 December 2025.

First, what each business costs. Every row is the bridge from Step two run again, and every row adds across to the rupee, so you can check it rather than take my word for it.

Scrolls sideways on a phone.

Company	Price Rs	Shares m	Market cap	+ Debt	+ Minority	+ Other claims	− Cash	= Enterprise value
Sun	43.10	174.364	7,515.09	2,510.20	1,561.81	1,991.74	1,016.11	12,562.73
Lux Island Resorts	52.00	137.116	7,130.03	3,244.27	—	1,460.28	1,505.34	10,329.24
Constance	14.25	109.653	1,562.56	5,369.32	(38.13)	961.72	375.01	7,480.45
Beau Vallon	4.00	175.645	702.58	1,175.96	—	348.25	123.20	2,103.59
New Mauritius Hotels	13.95	548.982	7,658.30	16,215.42	151.71	4,559.95	1,645.48	26,939.89

Constance’s minority is negative because its own subsidiaries carry accumulated losses. Lux and Beau Vallon have no minorities at all. Beachcomber’s Rs 16.2bn of debt is larger than the entire enterprise value of three of its four peers.

Then, what each business earns.

Company	Revenue	EBITDA	EBIT	Net income	NAV
Sun	6,502.24	2,067.07	1,730.08	1,273.84	3,722.47
Lux Island Resorts	10,555.61	2,741.79	1,929.63	1,208.28	8,380.26
Constance	6,149.09	1,794.26	1,083.32	314.38	5,095.34
Beau Vallon	1,147.41	440.27	327.70	221.35	1,519.94
New Mauritius Hotels	16,890.38	4,781.56	3,756.57	1,737.57	11,958.07

Divide the first table by the second and you get the comp set. The three shaded rows are the peer group summarised across those four companies; Beachcomber sits underneath for comparison.

Company	EV/Rev	EV/EBITDA	EV/EBIT	P/E	P/NAV
Sun	1.93	6.08	7.26	5.90	2.02
Lux Island Resorts	0.98	3.77	5.35	5.90	0.85
Constance	1.22	4.17	6.91	4.97	0.31
Beau Vallon	1.83	4.78	6.42	3.17	0.46
75th percentile	1.86	5.10	6.99	5.90	1.14
Median	1.53	4.47	6.66	5.44	0.66
25th percentile	1.16	4.07	6.15	4.52	0.42
New Mauritius Hotels	1.60	5.63	7.17	4.41	0.64

Look at the last two columns against the first three. Beachcomber is above the peer median on every enterprise multiple — 1.60 against 1.53, 5.63 against 4.47, 7.17 against 6.66. But on P/E it is 4.41 against a median of 5.44, and on P/NAV 0.64 against 0.66. On the equity measures it looks cheap.

Both readings are arithmetically correct, and the contradiction is the whole argument for using enterprise multiples. P/E is measured after interest, so a heavily borrowed company shows a small market cap sitting on top of the same profits and screens as a bargain. Enterprise value puts the debt back in. Beachcomber carries 3.05x net debt to EBITDA, the most in the set — which is exactly why it is the cheapest here on P/E and the dearest on EV/EBITDA.

Sources: Sun Limited FY June 2025; Lux Island Resorts FY June 2025; Constance Hotels Services FY December 2024; Beau Vallon Hospitality FY December 2024; New Mauritius Hotels FY June 2025 — all audited or abridged audited accounts, linked in full at the foot of this page. Share prices at 31 December 2025 from the SEM Factbook 2025, Official Market pp.24–25 and DEM p.35.

Tier 1 operators only: Sun, Lux, Constance, Beau Vallon. On four observations, quartiles are indicative rather than robust — which is exactly why the output is a range.

How to read thisOne row per valuation measure. The pale blue box covers the middle of the peer group — roughly half the companies fall inside it. The dark blue line is the middle company. The red dot is Beachcomber. A dot to the right of the box means the market is paying more for Beachcomber than for its peers.

New Mauritius Hotels trades above the 75th percentile of its listed peers on the primary multiple, and above the median on all three. Applying the peer medians back to its own figures gives a spread of implied values:

Implied equity value per share by method, against the traded price of Rs 13.95 at 31 December 2025. The band runs from the EV/EBITDA method at the low end to the P/E method at the high end.

How to read thisEach valuation method gives a different answer for what a share is worth, and the pale band covers that whole spread — Rs 3.84 at worst, Rs 17.20 at best. The solid blue line is the middle answer. The red dashed line is what the share actually costs today. Blue to the left of red means the methods say it is expensive.

Rs 13.95

traded, 31 Dec 2025

Rs 11.79

median method implies

−15%

the gap

5.63x

vs peer median 4.47x

The answer, in one paragraph

On this evidence, no — not at Rs 13.95. The market is already paying a premium to the listed peer group on every enterprise multiple, and the company carrying that premium is also the most indebted business in the set at 3.05x net debt to EBITDA. You are paying above the peer range for the balance sheet least able to absorb a bad year.

The case for the other side

A premium is not automatically wrong, and three arguments are available to defend this one:

Scale and brand. NMH is roughly twice the revenue of the next-largest peer, and Beachcomber is the strongest hotel brand originating in Mauritius.
The debt is buying something. Morocco and Zanzibar are funded expansion, not distress borrowing, and the Marrakech structure is designed to sell 51% of an SPV and repatriate capital.
The peer set is four companies. A median on n = 4 is indicative. Two of the four have quirks of their own — Constance is a debt-funded holding structure, Beau Vallon faces 56% dilution from its convertible.

What a comp set actually settles. Not whether to buy. It sizes the premium precisely — 5.63x against 4.47x, Rs 13.95 against Rs 11.79 — so the argument stops being about whether the stock is expensive and starts being about whether scale, brand and an expansion pipeline are worth 26% more than the peer group. That is a question about the business. The spreadsheet just makes sure you are asking it about the right number.

The working file

Football field chart of Tier 1 peer EV multiple ranges built in Excel — The original football-field chart from the workbook. Every cell behind it is a formula: change one company’s tier flag and the medians, percentiles and charts all follow.

FinallyWhat I got wrong

I kept a log of every mistake. Fifteen entries. Every single one was a which-line error, never an arithmetic one — the maths was fine, the wrong number went into it.

What I did	What was actually true
Divided profit by share count and got an EPS of 0.00881	The profit is in thousands; the share count is a literal count. Only one side was scaled.
Read the balance sheet down to “total liabilities” and stopped	The equity block above it holds the convertible bond, the minority answer and the NAV.
Backed out a share count from group profit	Earnings per share uses the owners’ profit. My count was 7% too high.
Took the cash figure off the cash flow statement	It was net of overdrafts that were already inside the debt figure — counted twice.
Read diluted EPS as something that had happened	It is a hypothetical. No shares had been issued.
Expected a 51% parent to report 51% of its subsidiary	Control means you consolidate at 100% and hand the rest back on one line at the bottom.

The last one is worth dwelling on, because it is the same lesson as the whole post. Diluted earnings per share is not a record of an event. It is the accounts telling you what would be true under a different assumption — printed right next to a number that describes what is.

Financial statements are full of that. The job is not to read the numbers. It is to find the sentence that says which numbers you are reading.

If you want the rest of what I have built, it is on the work page, and other write-ups go up in Writing.

If you take five things away

Say the units out loud before dividing. Statements are in thousands; share counts are not.
The equity block has three lines, not one. Take the owners’ subtotal, never the total.
Gross debt pairs with gross cash. Mixing them double-counts the overdraft.
Book value is not market value. A minority interest carried at book on a company trading at 0.31x book will wreck an enterprise value.
Read the accounting policy note. That is where the company tells you which choice it made.

AppendixEvery document this was built from

Nine companies, eleven filings. Where a figure in this post carries a page number, it points at one of these. Everything is publicly available.

Audited annual reports

Document	What came from it
Sun Limited Annual Report 2025	Balance sheet p.82 · profit or loss p.83 · convertible bond note 16 · owners’ continuing profit note 34, p.127 · shares and treasury note 15, p.62 · the two gearing figures, chief finance officer’s report p.26 · scheme of arrangement and discontinued operations
Lux Island Resorts Integrated Annual Report 2025	Statements p.63–64 · management contract with The Lux Collective, note 1 p.66 · subsidiaries note 8(a) p.87 · issued capital note 14 p.90 · convertible bond note 16 p.91 · borrowings and covenant breach note 17 p.92 · management fees note 26 p.98 · earnings per share note 31 p.99 · cash and net debt note 32 p.100 · segments note 33 p.101 · related parties note 36 p.102
Constance Hotels Services Annual Report 2024	“A policy choice is available” — convertible bond accounting policy, note 2(t) p.91 · borrowings split note 17 p.110 · cash net of overdrafts note 26(b) p.118
New Mauritius Hotels Annual Report 2025	Morocco special purpose vehicle and gearing target, chief executive’s interview p.8 · profit or loss p.61 · balance sheet p.62 · segments note 12 p.72 · earnings per share note 26 p.79 · cash note 39 p.89 · MIC bonds note 41 p.90 · preference shares note 43 p.91 · borrowings note 44 p.91–92 · Royal Palm Marrakech held for sale note 50 p.97

Abridged and condensed filings

Published under DEM Rule 18 or SEM listing rules. These are the whole public disclosure for four of the nine companies — which is why two of them have no EBITDA in this analysis.

Document	Filed	Auditor
Sun Limited — audited abridged financial statements FY2025	16 Sep 2025	—
Constance Hotels Services — abridged audited FS, year ended 31 Dec 2024	28 Mar 2025	Ernst & Young
Hotelest Limited — abridged audited FS, year ended 31 Dec 2024	28 Mar 2025	Ernst & Young
Beau Vallon Hospitality — audited abridged FS, year ended 31 Dec 2024	24 Mar 2025	RSM (Mauritius)
Tropical Paradise Co Ltd — audited condensed FS, year ended 30 Jun 2025	18 Sep 2025	—
Morning Light Co. Ltd — abridged audited FS, year ended 31 Dec 2025	18 Mar 2026	—
Riveo Limited — audited abridged FS, seven months to 30 Jun 2025	2025	—

Market and registry data

SEM Factbook 2026 — closing prices at 31 December 2025, earnings per share, dividend yield and net asset value per share. Official Market pp.24–25; Development & Enterprise Market pp.35–36. One common pricing date for all nine rows.
Mauritius companies registry (CBRIS) — registered nature of business, used to screen the peer set, and share counts where a filing did not disclose one. Extracts: New Mauritius Hotels C1439, Lux Island Resorts C6590, Constance Hotels Services C4335, Beau Vallon Hospitality C4703, Tropical Paradise C13796, Morning Light C13540, Riveo C205351, Hotelest C14126.

Accounting standards

The IFRS Foundation’s own text is behind a paywall. The Australian AASB standards reproduce it word for word and are free, so quotations are taken from those.

IFRS 10 Consolidated Financial Statements — Appendix A defines non-controlling interest; paragraph 22 places it inside equity; paragraph B94 is why it can go negative. AASB 10, full text.
IAS 1 Presentation of Financial Statements — paragraph 7 defines profit or loss, other comprehensive income and total comprehensive income; paragraph 81B requires the split between owners and minorities on the face of the statement. AASB 101, full text.
IAS 32 — why a convertible bond can sit in equity: the test is whether anyone can force the issuer to hand over cash.
IAS 33 — earnings per share, and why diluted EPS is a hypothetical.
Companies Act 2001 — section 215 requires audited statements to be filed with the Registrar; section 14 gives any member of the public the right to a copy. Full text, MauritiusLII.

Method

Enterprise value = market capitalisation + total debt + minority interest + other claims (preference shares and convertibles) − cash. Debt is borrowings only; lease liabilities are excluded throughout. EBITDA is taken after impairment and expected-credit-loss charges, with associates excluded. Net income is the figure attributable to owners of the parent, continuing operations. Financial years differ — Sun, Lux, Tropical Paradise and Riveo report to June; Constance, Beau Vallon, Morning Light and Hotelest report to December — and are not calendarised.

Where a filing did not disclose a figure, the cell was left empty and the observation count published beneath the statistic. No figure on the sheet is estimated, interpolated or taken from a secondary source.

Disclosure

This is a student project written up for my own record. It is not investment advice and not a recommendation. I hold no position in any company mentioned, and I have no relationship with any of them.

Every figure was checked against the source document at the page cited. If you find an error, I would genuinely like to know — the working file is more useful to me correct than flattering.

Also on this site

The BAI collapse in Mauritius, explained — all seven chapters of the nTan forensic report on the 2015 BAI collapse, technique by technique, each one matched to the statute it would be tested under.