Online Marketplaces Dashboard.

The three names sit inside the same TSE bucket and share almost nothing else. Mercari (4385.T), ZOZO (3092.T) and Rakuten (4755.T) are three orthogonal economies joined by an accident of classification — a C2C resale marketplace with a fintech graft, a fashion concession, and a super-app conglomerate. Their betas run from 0.25 to 0.84, their volatilities from 28% to 46%, their correlations to TOPIX from 0.23 to 0.44. There is no homogeneous benchmark that can legitimately aggregate them, and no consolidated multiple that reads any of them correctly.

The decade settled one thing for this bucket, and it is unflattering. Against a TOPIX up 220.7%, ZOZO returned +92.2%, Mercari –9.8% since IPO and Rakuten –31.7% — between 128 and 252 points of relative destruction, with no name beating the market on three, five or ten years. Leadership rotated completely across six regimes: each name led one and sank in another. What moved the returns was never realised quality. It was the multiple, and the multiple followed the direction of the trajectory. The market prices the second derivative — the visible, accounted inflection of a line — and punishes deceleration regardless of the return on capital. ZOZO is the exhibit: the highest-quality name in the bucket, a 39% return on capital, and the worst performer of the last regime on a marginal core deceleration.

What follows below sits in three layers. The economic engine and the cross-operator inputs describe what runs underneath all three. The archetype map and the names section sort them. The mispriced reads and the structural watchlist track what each consensus is getting wrong and what would force a reframing. The common thread is that every one of the three carries a documented long bias and none is actionable at spot — each is waiting for a catalyst to be accounted rather than anticipated.

The single fracture line that governs the bucket is asset-light take versus financial balance sheet — it decides who converts operating profit into cash, who has it absorbed by a credit book, and who has it drowned inside a conglomerate. The monetisation hierarchy sorts the three cleanly. ZOZO takes a 27.3% commission on fashion inventory it never owns, held for a decade and converted to 93% gross margin and roughly a fifth of revenue in cash, with no stock risk and no credit risk — a rent paid for curation and twenty years of purchase data, not for mere intermediation, and one a C2C marketplace at ~10% cannot touch. Mercari takes ~10% on a self-reinforcing domestic C2C network. Rakuten takes no single rate at all — its value is a regulated financial spread, the FinTech segment earning ¥199.9bn of operating profit in FY2025, more than twice the marketplace and 2.25× Internet Services.

The masked-core theme is what makes consolidated margin an unreliable instrument across the whole bucket. Mercari's domestic marketplace earns 35.0% on almost no capital, but the consolidated line reads 14.6% and reported free cash flow is negative — the drag is the balance-sheet-heavy Merpay credit book, not the platform, whose own conversion is strong. ZOZO's 93% gross margin and 39% return on capital are read as ex-growth because the core grew only +5%. Rakuten reports ¥11.9tn of net "cash" that is in fact deposits sitting inside Rakuten Bank, and a P/E that reads on a negative number — the FinTech's real profitability is drowned by a Mobile segment that lost ¥161.8bn, most of it now the depreciation of a network already built and paid for. In each case the aggregate hides the economics, and only a sum-of-the-parts recovers them.

Cash conversion completes the picture and inverts the intuition. ZOZO converts cleanly — free cash flow at 94% of NOPAT on a negative working-capital float — and is the only name in the bucket whose consolidated cash bridge is both legible and positive. Mercari's bridge is broken at the consolidated line by the credit book: each incremental yen of Merpay lending consumes balance sheet funded by debt, and receivables have stretched to 441 days, so reported free cash flow is negative while the retreated platform proxy sits near +¥43bn, about 6.6% of the market cap. Rakuten's consolidated cash flow swings by trillions on the movement of banking deposits and cannot be read at all. The rule that ties this together: where the consolidated statement masks the quality (Mercari) or the artefact (Rakuten), the market misprices; where it is legible (ZOZO), the price reflects the quality minus a structural overhang.

The first cross-operator input is the rate regime, and it dominates every structural factor. The ZIRP inflation of 2020–2021 and its destruction across 2021–2023 explain most of the decade's performance, ahead of any fundamental — Mercari's P/E ran from 162x to 17x and ZOZO's price-to-book from 28x to 9x, Mercari tripling its book while the price fell. The market learned, at its own expense, that unprofitable long-duration growth is a rate bet, not a value. The consequence for modelling is non-negotiable: the current decade-floor and decade-ceiling multiples are regime artefacts, terminal multiples must be normalised on the ex-peak median, and Rakuten — the only name near the top of its own corridor at 3.15× book against a 2018 trough of 1.28× — carries the maximum drawdown risk regardless of the value of its parts.

The second input is the untested fintech credit-risk factor, and it is the hidden systemic link across two-thirds of the bucket. Mercari's Merpay and Rakuten's Card-and-Bank both carry a favourable-cycle cost of risk — around 70bps, recovery near 99.4% — on books financed by debt and never crossed through a Japanese consumer-credit recession. A turn in that cycle would strike both simultaneously: these are not decorrelated positions, they share a macro factor. On Mercari the stress runs to roughly –20% of group operating profit at 300bps; on Rakuten it compresses the 20.5% FinTech margin that anchors the whole SOTP. ZOZO is the exception — its critical cost is industrial and steerable (logistics fulfilment, a slowly rising cost of acquisition), not a latent credit risk.

The third input runs through parent-child governance and the dividend together. Two of the three sit under a control overhang — ZOZO under LY Corporation at ~50%, Rakuten Bank listed beneath Rakuten — which TSE reform makes as much a potential catalyst as a permanent discount. And the dividend is the only multiple-independent source of return the sector offers, concentrated entirely on one name: ZOZO pays a 4.3% shareholder yield on a 72% payout with its first material buyback, which rises to 5.8% at its bear price and turns the downside from a permanent loss into a timing disappointment. Mercari and Rakuten both pay zero — Mercari reinvesting into the core and the credit book, Rakuten diluting and suspending — which is exactly why their bears run deeper. There is no aggregate tailwind here to lean on; the return has to come from something name-specific and accounted.