Hepsiburada's thesis in one sentence is that its controlling shareholder absorbs an early-stage margin collapse, the lira problem dents the story differently than headline numbers suggest, and a regulated fintech stack converts growth into earnings before minority holders lose patience. Compression in the underlying margin story, touched off by the installments race that drove collection fees higher, arrived at the same time as the balance sheet's negative-equity moment, and the sponsor rather than the market answered the call for fresh money. Two top-ups inside a year, the second fully paid in September, leave the dispute about solvency closed and the dispute about minority treatment open. The discount would be exactly the sponsor thesis if the parent's willingness sat in question, and the funding record shows it does not.
The most important recent development is the June launch of Hepsitaksit, an installment-lending product that facilitated point-four percent of gross merchandise value in its first month. The mechanism is direct: the payment behaviour that lets a hyperinflationary consumer keep buying sits at the center of the loss account through card-collection fees, and a licensed lending product pulls that flow onto the platform's own ledger. The June quarter still showed fintech-linked income in retreat as volume scaled, so the engine has not yet covered its own cost, and rate policy at a tight-thirty-seven-percent policy rate keeps funding expensive until the parent's bank purchase adds a deposit-backed leg. Sponsorship answered the structural question during the year, since any drawdown in the balance sheet gets refilled from Almaty.
The key tension is that the growth account now shows the signature of credit-cycling rather than commerce compounding, with financial-expenses deterioration widening net loss even as order counts rise. In a market where instalment culture runs the checkout, cheap lending economics arrive only after a first loss cycle is absorbed. Meanwhile active-customer growth runs in the low single digits while per-account frequency does the mechanical work, a profile that rewards the installed base and gets judged harshly when breadth stalls. The cost of defending the checkout through bank instalments is not a rounding error, it is the single largest swing factor between the restated EBITDA line and the reported loss, since collection fees on divided payments exceed the entire platform's operating profit several times over. Any reading of this company that treats the fee line as background noise reads the income statement backwards, because the fee line is where the competition actually gets paid.
The catalyst calendar is crowded. A fully paid-in capital increase of nine-point-three billion lira reached the trade registry this month after a unanimous shareholder vote, the parent's purchase of Rabobank's Turkish bank sits at closing with a banking licence attached, and the third-quarter print in early November stands as the first clean window into whether advertising, shipping, and collection-fee drag recede from the cost lines. A regulatory deepening of the Temu restriction, which has already stripped cross-border parcels from the cheapest competitor, decides the competitive backdrop at the same time.
Hepsiburada began in 2000 as an electronics retailer in Istanbul and turned itself into a three-layer platform: first-party retail that owns inventory and margin risk, a third-party marketplace where merchants pay commissions on their own stock, and a services layer spanning Hepsijet logistics, Hepsipay payments, Hepsiexpress grocery, and an advertising network sold to on-platform sellers. The June quarter split revenue at sixty-five parts first-party, thirteen parts marketplace commissions, sixteen parts delivery services, and six parts other, which means the marketplace and services businesses carry a far larger share of platform profit than the revenue line admits. The 2021 direct listing on Nasdaq made D-Market Elektronik Hizmetler ve Ticaret the first Turkish technology company on a United States exchange, a scarce access currency that survives even as the operating centre of gravity stays in Istanbul.
Ownership changed the company's risk category in January 2025, when Kaspi.kz of Kazakhstan closed a purchase of sixty-five percent from the Dogan family for roughly one-point-one billion in sponsor currency and later extended the holding by another ten points to seventy-six percent. A Nasdaq-listed Kazakh super-app operator took the Turkish lira problem onto its own balance sheet, converting HEPS from a standalone hunter of capital into a sponsored subsidiary whose parent treats the ninety-million-person market as its second home market. The purchase bundle included everything that makes a local operator investable and a local minority irrelevant at once: board control, strategy authority, and the practical ability to fund any shortfall whenever the board decides. What the sponsor bought, in its own framing, was nine extra years of addressable-market runway for a template it had already saturated at home, and what the minorities retained was a claim whose recovery value now passes through the sponsor's own programme rather than through the local capital market. Mikheil Lomtadze, Kaspi's cofounder, chairs the local board, and the operating levers chosen since reflect parent doctrine: own the traffic, own the rails, defend the share, and fund the gap from above.
The competitive frame is a heavy duopoly with Trendyol, whose estimated share of Turkish online retail sits several multiples wider than Hepsiburada's, still backed by Alibaba as a strategic holder after the local ride-hailing sale freed fresh capital. Temu's cross-border model died by decree rather than by competition: the duty-free allowance for foreign parcels first fell to a thirty-euro ceiling, shipping charges entered the same cap, and a January decree abolished the exemption outright, effective February. The Chinese platform suspended cross-border sales and operated only local listings, which redirected demand toward domestic marketplaces and handed Hepsiburada a protected audience segment precisely as its own costs peaked. The protection works through price floors rather than through market-share gifts, since the regulated segment skews toward low-ticket goods where Hepsiburada's own economics have always been thinnest. Market estimates put the redirected volume in the billions of lira per year, and the direction of flow, from foreign parcel networks to domestic marketplaces, benefits the two incumbents roughly in proportion to their merchant breadth rather than equally.
The baseline year sets the size of the repair job. Full-year 2025 closed with gross merchandise value of two hundred fifty-seven and a half billion lira, revenue of eighty-four and a half billion, EBITDA of one-point-one billion at four-tenths of a percent of value, and a net loss of five-point-seven billion. Free cash flow printed near nine billion lira in that year because merchant payables finance the merchandise cycle, a float that props liquidity even as the operating core bleeds, and the two shareholder top-ups that followed exist precisely because float cannot fund permanent losses. The listing history matters for readers because it defines the minorities' exit option. Turkish-domiciled ADSs trade on Nasdaq with thin volume, wide spreads, and a habit of repricing on macro headlines from Ankara rather than company news, so the free float prices Turkey risk and microcap-liquidity risk at once. Two shareholder top-ups in a year, each offering shares only to holders of the domestic book with the depositary's pre-emptive rights disapplied, quieted solvency debate and raised a fresh one about whether the sponsored structure drifts toward a squeeze-out at private-market terms. The better answer sits in behaviour rather than in promise, and the record so far shows the parent paying above the traded price for its own refills, the single most reassuring fact a minority holder has.
The moat of record is the match between producer and shipper, with Hepsijet running own-van delivery plus the Hepsiexpress dark-store network across the eighty-one provinces with same-day and next-day windows. Delivery service revenue from off-platform contracts grew five percent in the June quarter while merchant-facing shipping costs grew twenty-two percent, proof that the engine runs faster than the tariff it charges so merchants stay. The subsidy wedge is deliberate market-making: carrying merchant economics today buys the delivery network's density, and density is the asset a challenger cannot buy on a weekend. Advertising, grouped inside other revenue, fell sixteen percent in the quarter even as platform events set traffic records, which shows monetisation layers being starved while the logistics layer gets fed.
Hepsipay operates under a Central Bank payment licence with a wallet linked to checkout and a settlement rail for merchants, and it ran headlong into the instalment culture of Turkish consumer credit, where card issuers already dominate the checkout with divided payments. The June launch of Hepsitaksit attacks that gap from the platform side, letting buyers split purchases outside the card network while merchants settle faster, and it reached a modest point-four percent of value in its opening month. The strategic point is not the early scale, it is the plumbing: lending income replaces collection-fee expense in the same behavioural slot, so each instalment shifted to the native rail turns a cost line into a revenue line. A deposit-funded bank inside the group, pending the closing of the Rabobank purchase, gives that rail a cost of funds no marketplace wrapper can match.
The first-party retail layer looks low-margin and asset-heavy, and it is, but it also anchors delivery density, buys assortment breadth that merchants resist stocking, and generates the pricing data that feeds the advertising engine. Hepsiexpress extends the same engine into grocery with dark stores in the major metros, trading basket margin for order frequency in a category where weekly habit drives route density. The technology layer is real rather than ornamental, spanning search, recommendation, and a merchant-facing toolkit, though it carries a comparatively lean cost base at half a percent of value, thin for a platform that claims technology as its lead product. The cost base signals one of two things, either genuine modernisation of the parent's engineering culture into the subsidiary, or underinvestment that shows up later as slower feature velocity, and the November print plus hiring data settle which of the two holds.
The compounder across all layers is licence accumulation, which in Turkey takes years: payments under the Central Bank, logistics under separate capital, advertising inside the operating company, and, with the bank closing, banking supervision on top. Kaspi built this exact stack in Kazakhstan and posted its own organic results there in the same period, with payments volume growing a third and marketplace value growing at a faster clip, evidence the template migrates when management holds the instruments steady. The open question is conversion speed, because Turkish habits favour bank instalments and wallet share here starts from a small base rather than the dominant position the parent enjoys at home. The parent's own trajectory shows a route from licences to dominance that took roughly a decade in Kazakhstan, and the Turkish base carries a comparable population with a deeper banking infrastructure to integrate rather than to build. Speed of conversion, not existence of the licence stack, is the variable that separates the bull construction from the base in the valuation section, which is why the bank closing and its first deposit-funded products deserve a standing place on the catalyst list.
The June quarter closed with gross merchandise value up three percent year over year in restated real terms, orders up thirteen, and revenue up three percent at twenty-two-point-eight billion lira. Management attributed the softness to demand moderation under inflation pressure and an extended holiday, yet order frequency rose to seven-point-four orders per active customer on a rolling-year basis, and marketplace share of value held just below sixty-nine percent. The physical engine keeps gaining throughput while the value lines stall, which is textbook high-inflation economics: tickets shrink because wallets shrink, and the platform compensates with trips.
The cost curve is where the quarter lost its margin. Advertising grew forty percent, shipping and packaging twenty-two percent, payroll twelve percent, all against flat real revenue, while impairment charges collapsed ninety-five percent on better credit collections. Underneath the operating lines, financial expenses and fees swelled to three-point-two billion lira against financial income of nine hundred twenty-five million, the widened gap explained by fees on card collection as the merchant checkout subsidised instalments to defend competitiveness. The collection-fee channel deserves the emphasis it gets here because it sits outside every ratio investors usually screen: it is neither advertising nor shipping, it is the toll the banking system charges for the instalments a platform offers in a competitive market, and it scales with checkout behaviour rather than with revenue. A platform that lets its merchants win on payment terms is, in accounting terms, buying growth at a rate set by the banks, and the second-quarter line shows exactly what that purchase cost. EBITDA fell to two hundred thirty-nine million from nine hundred seventy-seven million a year earlier, and the net loss doubled to one-point-nine billion, with monetary gains of one-point-one billion inside the loss as the index arithmetic of hyperinflation accounting cushioned what the fee lines then gave back.
The before-picture explains how the after-picture got so far out of line. Full-year 2025 ran EBITDA at four-tenths of a percent of gross merchandise value, down from well over a percent the year before, while free cash flow nearly doubled to nine billion lira on the strength of the payables float. In the first half of 2026 the same divergence widened: EBITDA of six hundred eighty-nine million restated against unadjusted EBITDA above two billion, and free cash flow of three hundred fifty-four million, down about five-sixths year over year. The reason the restated line compresses faster than the business itself shrinks is mechanical: restatement multiplies historical cost bases and old income by conversion factors, so revenue earned early in the year loses nearly a third of its purchasing power on re-expression, while fee income earned late in the year barely moves. The unadjusted stack matters for exactly this reason, because it strips the index arithmetic out and shows an operating business that still makes money in nominal lira even as restatement compresses everything above it. Readers who anchor only on the restated EBITDA line end up pricing twice the contraction the cash records show.
The balance sheet carries the print of that pressure. Cash and equivalents ended June at seven-point-zero billion lira against thirteen-point-three billion at year-end, financial investments added two billion, bank borrowings took back less than half a billion, leaving net cash near eight-point-six billion, roughly one-fifth of market value in sponsor currency. Total equity turned negative at minus five hundred eighty-two million after an accumulated deficit near twenty-four and a half billion, a state that hyperinflation restatement deepens because indexation lifts cost bases faster than retained income rebuilds them. Current liabilities of thirty-two billion sat above current assets of twenty-four and a half billion, financed by the trade payables float, which makes working-capital discipline, not just the parent's cheques, part of the solvency story.
The forward case rests on margin restoration with an explicit governance commitment, since the parent posted its own standards for the subsidiary's eventual return on equity in its plans and has already paid twice to keep funding on track. Restated EBITDA needs to move from six-tenths of a percent of value toward the general-merchant norm of one and a half, because the market prices platforms on fee margins, not on merchandise turnover, and the unadjusted run-rate above shows the raw material exists inside the same business. Whether the promise reaches minority holders or arrives as parent-only uplift depends on structure, since the sponsor consolidates the local subsidiary in full and currently books results one hundred percent through its own accounts, an arrangement that turns any reorganisation of the depositary into the decisive minority event.
Bringing the parent playbooks into Turkish infrastructure is the sharpest test of pressure since the Kazakh bank purchase closes. The Rabobank deal gives Hepsitaksit a funding leg inside a supervised bank rather than as a marketplace feature, which changes both the cost of funding and the legal ceiling on the lending book, and the parent has flagged conversion of the wallet base as the bridge to it. The conversion risk is real, because super-app behaviour in a Kazakh consumer base does not translate automatically to a Turkish base with deep instalment habits and low single-digit wallet penetration. Execution here is measurable within quarters, since loan-book take-up, wallet top-ups, and instalment migration each print monthly in operating disclosures.
A third execution risk sits inside the cost curve. Shipping cost growth outpacing delivery revenue growth by several multiples in the June quarter signals that subsidy depth has become a competitive weapon aimed at merchants, and a margin ledger that fights price wars through cost subsidies tends to open new fronts rather than close old ones. Combined with advertising spend climbing forty percent, the posture reads as share defence buying frequency, and the third-quarter print in November stands as the first checkpoint on whether those ratios recede or harden into structure.
The calendar wraps the year in a tight loop. The new shares from the September top-up enter registration in the autumn, the bank purchase receives its closing mechanics, the third-quarter report lands in early November, and any fresh capital event from the parent would reset the minority arithmetic without warning. In a year in which the sponsor absorbed the funding gap twice, execution risk is no longer about solvency, it is about sequence: margin restoration, bank integration, and minority-structure decisions form a chain whose links arrive in that order over the next four quarters, and each link changes the value of the one after it, since a funded margin bend gives the bank integration room to breathe, and an integrated bank makes any eventual minority structure richer than the current one. The sequence, not any single print, is the forward story.
The first risk is competitive fatigue against a leader with deeper pockets and a larger share base. Trendyol extends out over a wider distribution base with Alibaba's strategic backing behind it, and subsidies by merchants into so-called zero-fee checkout can run inflationary for the whole industry rather than for one competitor. If the parent's patience carries a clock as well as a wallet, another quarter of widening advertising and shipping ratios could force a scale-down in share defence, which in turn lowers order frequency and starves the per-account engine that currently does all of the platform's growth work.
Active-customer growth sat at two-point-five percent in the June quarter against order frequency above seven orders per active customer, meaning the engine of expansion is a deeper-installing cohort rather than a larger one. If wallet growth stalls inside the top cohort, order frequency saturates, and growth arrives only from margin restoration instead of volume, which is a different investment than the market holds today. The mechanism behind that stall is straightforward: Turkey's high-income cohort already shops the platform weekly, the next cohort needs cheaper credit or faster delivery to join, and both levers cost money precisely when the parent asks the platform to earn its own keep. A base that stops widening turns fixed-cost arithmetic from friend to enemy, since advertising written to recruit shoppers stops compounding and starts reading as pure expense, the exact inversion of the argument used to justify the spend.
A third risk is the currency channel itself, working through real-income effects rather than simple top-line translation. IAS 29 restatement means profit appears and disappears with index arithmetic while free cash flow responds to the lira liquidity behind the merchandise float, so the reported loss can widen even as nominal fee income grows. A stalemate episode in which policy rates stay near thirty-seven percent while real wages stagnate shrinks basket size yet again and pushes the platform deeper into subsidy, and basket size is already printing in retreat.
A fourth risk is the regulatory pendulum. The February decree shifted demand to local players and generated inflation pressure in the same month, and the Turkish state responds to that pressure either by loosening import controls or by turning competition enforcement inward on local champions. An inward turn against the marketplaces follows the same playbook the authority used on the foreign entrant, and Hepsiburada holds the largest direct exposure to that target among local platforms after the leader. The mechanics of that scenario are unglamorous but real: price-comparison studies, seller-fee inquiries, and data-localisation requirements each carry a cost that lands disproportionately on the second platform, because the leader absorbs fixed compliance from a bigger revenue pool. Investors holding other emerging-market depositories already recognise the pattern, where a single regulatory decision repriced an entire sector overnight, and the lesson travels intact to Istanbul.
The valuation framework has to start from a sponsor thesis from which no price target follows, because the useful question is when minority holders get paid, not at what price the last trade cleared. Three variables carry the exercise: the equity value recognised by the parent in any consolidation, the margin path that makes restated EBITDA sustainable above one percent of value, and the eventual shape of the depositary line in a reorganised structure. Held together, the fair structure values the platform as a fee engine plus a loan book rather than as an inventory retailer, and the market's current quote prices it as the latter.
The marketplace layer earns a take rate near ten percent of gross merchandise value before delivery, with a net service margin inside that near one-third of take, which at the annualised run rate produces a service-fee annuity worth a mid-single-digit multiple of revenue under any reasonable risk discount. The annuity framing matters because fee pools re-price with inflation mechanically in the short run, so value lives in the take-rate trajectory and in merchant retention, not in the nominal pool. A one-point change in merchant take, worth billions of lira at scale, moves the annuity more than a year of volume growth does, which is why the subsidy wedge described in the moat section and the margin bend described in the risk section are the same variable read from two sides. Apply that to the two-hundred-fifty to two-seventy billion lira restated service-fee pool and the net cash of eight-point-six billion lira, near a fifth of market value in sponsor currency, covers a meaningful part of the current quotation with the parent credit untouched. The bear case strips that back: a five-times multiple on a two percent restated margin against three hundred sixty-six billion lira of annualised value and a fifty percent haircut on cash lands near one-point-three per receipt. The base case pairs a three-times multiple on a stabilised margin with full credit for net cash and reproduces the current price near two-point-seven, while the bull case lends the parent's own profitability targets to the discount rate, values the payments stack on its own commission, and reaches roughly three-point-nine before any further capital event.
Two private-market anchors sharpen those bands beyond what public comparables can do in a hyperinflationary market. The sponsor paid one hundred thirty lira and a half per underlying share in both top-ups inside the past year, the first while the receipt sat below three in sponsor currency, the second with the receipt nearer two-point-six, meaning the controlling holder paid above market on its own refill both times. Per Kaspi's own annual accounts the local platform cost about one-point-two billion sponsor currency to reach seventy-six percent, setting an implicit whole-company anchor above the current quotation before any integration value is added, and minority holders trade well inside that anchor today.
Counterargument, stated plainly: hyperinflationary accounting strips comparability away from every multiple in the construction, and restated numbers carry as much machinery as meaning. The sceptic holds that negative equity, a free-cash-flow print near three hundred fifty-four million lira against a nine-billion start-of-year float, and an accelerating loss all point to a lower multiple rather than a higher one, and that reading has weight. The defence is that EBITDA on an unadjusted basis stood above two-point-one billion lira in the first half while the restated line printed six hundred eighty-nine million, making part of the visible collapse index arithmetic rather than commercial fact, though the honest answer is that neither reading shows a profitable platform and the bull case leans on the parent's own targets as much as on audited numbers.
The record shows a platform whose growth engine still compounds and whose margin engine is early in its turn, paired with a sponsor that has already demonstrated a willingness to fund the gap on more than one occasion inside eighteen months. The marketplace retains its scale, the logistics engine holds the second position by breadth of coverage, and the payments stack carries the licences a challenger cannot buy quickly at any price. What Kaspi's ownership removes is solvency risk in the plain sense, since the negative-equity position was rewritten by shareholder money twice within a single year, and what it leaves unresolved is the terms on which minority holders eventually share in the repair.
That judgment follows the structure the parent has built. The integration of the bank licence is the highest-conviction doorway to converting payments volume into recognised earnings rather than advertised ones, and the sponsor pushes there with the closing of the bank deal, the September top-up, and the delivery arm gaining its own chief executive. Each of those moves maps onto the thesis variables in turn: the top-up pins the solvency variable, the bank closing pins the fintech-conversion variable, and a reorganised depositary would pin the structure variable that decides whether minorities ever get paid for any of it. In a multi-scenario frame the base case sits near the current price, the bear case loses roughly half, and the bull case gains roughly half, which makes this a bet on the slope of margin restoration rather than on the level of the quote.
The behavioural pattern the market is learning is that the discount over the coming two quarters rides on regulatory and consumer-cycle news rather than on company execution, and the gaps come from policy rather than product. A marketability discount holds as long as the parent keeps supporting the base, and the pattern of two top-ups inside twelve months is the strongest single evidence on that point. In a market where the sell side overweights execution and underweights structure, the downside deserves more respect than the momentum reading gives it, and the sponsor's presence caps exactly the tail that killed comparable hyperinflation-era depositories.
The risk-reward stays asymmetric in the direction the sponsor has already chosen, with the payments stack as the lever that turns volume into recognised profit. The November print, the bank closing, and any further capital event stand as the discovery window, and the ask from minority holders is patience while the parent's programme does the funding. What the thesis offers in return is a claim on the second platform of a ninety-million-person market at a private-market anchoring price, with the funding risk parked on a Nasdaq-listed parent's balance sheet rather than floating with the lira itself. The load-bearing judgment is that the discount is a fair price for uncertainty about structure, not a signal that the assets are worthless, and the two top-ups are the strongest available evidence that the parent values this platform above the price at which minorities can buy it. Patience, not prediction, is the position's defining feature, and every catalyst in the next two quarters either confirms or breaks that patience.