Variant Perception

Figures converted from GBP at historical FX rates — see data/company.json.fx_rates. Ratios, margins, and multiples are unitless and unchanged.

Where We Disagree With the Market

The market is debating whether $575m of FY25 above-1%-yield interest income is a "rate-cycle gift" reversing or a "deposit franchise annuity" compounding — and both sides are arguing about the wrong thing. Customer balances are growing 33% YoY; rates are falling perhaps 100-150bps over 24 months. At those tempos, the level of interest income stays roughly flat in dollar terms through FY27 even as rate-driven yield declines, because balance compounding offsets it. That re-frames the central FY26 print debate: the question is not "will the windfall reverse" but "can balance growth keep pace with rate cuts," and the answer the data has already given (H1 FY26: balances +33%, PBT -13%) is that it is close — closer than either the 32× bull multiple or the 18× bear anchor implies. Two further disagreements compound this: the Nasdaq listing is being priced as a governance test when the US institutional baseline is structurally more permissive of founder dual-class than UK ESG screens; and Wise Platform — with five named bank/fintech contracts already live — is being valued at zero by consensus models because management has not yet disclosed it as a separate revenue line.

Variant Perception Scorecard

The 72 score reflects three narrow disagreements that are observable inside ten days (Nasdaq debut), thirty-five days (FY26 print), and six months (OCC charter, H1 FY27 print) — short enough that a PM can underwrite. Consensus clarity is mid-range because the bull/bear tape is genuinely split (sell-side averages $16.26 on $10.17-$20.85 range; one model goes to $7.06). Evidence strength is high because each variant claim ties to a specific upstream data point — balance compounding rate, founder economics, named platform contracts — rather than to a generic "market is wrong" framing.

Consensus Map

The consensus map matters because two of the seven issues are partially miscoded by the market. Issue #1 (margin reversal) embeds a static-balance assumption that is empirically false at current growth rates. Issue #3 (governance discount) is being applied with a generic template that does not survive the founder-economics specifics. Issue #5 (Platform) is being counted at zero by methodology, not by argument. Those three are where the variant edge sits.

The Disagreement Ledger

Disagreement #1 — Float compounding offsets rate compression

A consensus analyst would say the $575m of FY25 above-1%-yield interest income is a windfall that mean-reverts as policy rates normalise — and they would point to H1 FY26 PBT -13% YoY as confirmation. Our evidence reframes the same data: that decline came with cross-border volume +24% AND customer holdings +33%, which means the operating engine accelerated while the rate engine dragged. At forwards-implied UK base rate near 3% by end-FY27 (vs ~4.5% spot) and assuming balance growth decelerates to 25%, our arithmetic puts total FY27 interest income at $720-840m versus a roughly $390-450m bear case — a $260m+ PBT difference that is the entire delta between the bear's $9.07 target and a base case in the $13.24-$14.57 range. The market would have to concede that the FY24-25 reported earnings level was not a pure cycle peak; it was a cycle peak layered on top of a deposit franchise that is itself compounding fast enough to substitute. The cleanest disconfirming signal is customer-holdings growth slipping below 20% YoY at the FY26 print, or any Bank of England action that takes UK base rate below 2% materially faster than forwards imply.

Disagreement #2 — Nasdaq inverts the governance discount

A consensus analyst would say the dual-class extension to 2035 is a governance overhang and cite Hinrikus, ISS / Glass Lewis, and PIRC as evidence. Our evidence reframes the marginal-seller question: it is the UK pre-listing screen — the FTSE 4Good / UK governance code apparatus — that has been pricing the discount, not US growth-fund mandates that have lived comfortably with Meta, Snowflake, Palantir, and the Roblox cohort at premium multiples to non-dual-class peers. Käärmann's $4.6bn economic stake against $269k cash compensation is the structural negative of the textbook principal-agent problem; the FinanceFeeds note explicitly observes "US markets have historically been more accommodating of such structures." The market would have to concede that the discount narrows on listing day rather than widens, which is the exact opposite of the Seeking Alpha framing. The cleanest disconfirming signal is Nasdaq ADV failing to widen above $33m within thirty sessions, OR a US ESG fund publicly screening out (Calvert, Domini) — neither has happened pre-listing.

Disagreement #3 — Wise Platform is a free option in consensus

A consensus analyst would say Wise Platform is "the supposed second leg" with no quantifiable revenue. Our evidence reframes the methodology: the value is unbooked because management has not disclosed it, not because it does not exist. Standard Chartered, Morgan Stanley, Nubank, Itaú, Wealthsimple and Raiffeisen are all live or contracted — five named institutional partners against an addressable legacy-bank revenue pool the AInvest piece sizes at $225bn. dLocal, the closest B2B FX-rails comparable, trades at 7.3x EV/Sales for a thinner-margin business. Even modest contribution (5% of FY27 group income at a B2B multiple) adds $1.3-2.6bn of equity value; a 10% contribution and a 9x multiple gets to $4bn — 23% of the current market cap. The market would have to concede that the optionality is real and capitalisable on disclosure rather than aspirational. The cleanest disconfirming signal is FY26 segment reporting (4 Jun 2026) leaving Platform inside "Other revenue," or any of the five named partners reducing scope publicly.

Evidence That Changes the Odds

The strongest piece of evidence is row 2: H1 FY26 prints volume +24% and holdings +33% against PBT -13%, in real time, against a rate-cut backdrop. That single data triplet contains both the bear signal (the rate engine dragging) and the variant signal (the operating and float engines absorbing roughly half the hit). The FY26 full-year print is the next clean read, and rows 3-4 frame the listing-day governance test that runs in parallel.

How This Gets Resolved

The resolution path is unusually concentrated: rows 1, 2, 3, 6, and 7 all land between 8 May and end-June 2026 — a fifty-day window that contains the listing, the FY26 print, the index reviews, and the AGM. By 1 July 2026 a PM should have most of the data needed to keep, retire, or invert all three variant views. Rows 4, 5, and 8 stretch through H2 2026 into early 2027 and are the slower-to-resolve overhangs.

What Would Make Us Wrong

The bear's static-balance assumption is wrong on the math, but it is not wrong on the direction. If UK base rate cuts faster than forwards imply — say the BoE goes to 2.0% by mid-2027 to combat a domestic recession — and SME confidence weakens enough to slow customer-holdings growth from 33% to 15%, our "balance compounding offsets" arithmetic compresses to a point where the above-1%-yield line drops to roughly $325m by FY27, not the $450-520m we anchor on. That scenario does not require either the bull or the bear thesis to be correct on the franchise; it just requires a UK macro pull-down to compound with the rate move. The cleanest single tell is customer-holdings growth slipping below 20% YoY at the FY26 print — at which point our first variant view should be retired immediately rather than litigated.

The Nasdaq governance-discount inversion is the most fragile of the three. The Meta / Snowflake comp set is real, but Wise has a specific feature those names do not — three live regulator settlements, a CEO with two personal regulator fines, and an interim CCO. A US ESG fund or activist could correctly argue that those facts justify a discount irrespective of the founder economics, and the marginal-seller flip we expect on listing day might not materialise. If first-30-day Nasdaq ADV stays below $26m and a major US payments analyst initiates with an Adyen-discount framing rather than an Adyen-comp framing, the listing was a technical event, not a re-rating event, and our second view should be retired alongside the bull's $19.86 target.

The Wise Platform optionality view depends on management actually disclosing the line. They have not in two years, and there is no contractual reason they must. If FY26 reporting and H1 FY27 reporting both leave Platform inside "Other," the value continues to sit on Wise's balance sheet without ever being capitalised in the equity — and a perfectly correct DCF that gives Platform a small explicit value does not have to mean the market is wrong; it means the market is rationally refusing to underwrite an undisclosed line. The cleanest tell is the segment table on 4 June 2026: if Platform is not separately broken out, mark it back to optional and demand a disclosure event before re-engaging.

The single thing that connects all three risks: a falling-rate, slowing-volume, opaque-Platform print at FY26 results would simultaneously falsify the float-compounding view, the listing-narrows-discount view, and the Platform-option view. That is the asymmetric scenario worth pre-mortem-ing.

The first thing to watch is the customer-holdings line at the FY26 print on 4 June 2026 — if it lands above $37bn, the central variant disagreement is intact; below $33bn, retire it.