Sparkassen's €0.99 Crypto Order: A Closed Ledger With No Exit

Evidence suggests the most important number in Germany's Sparkassen crypto launch is not €0.99 per order. It is the number nobody printed. I spent four days reconstructing the fee architecture from the Deka price sheet and the four-entity custody stack, and the same gap appeared in every document: the spread is disclosed nowhere. That omission is not a formatting error. It is the product. Context first, because the headline has been recycled since 2021. DSGV floated the idea, reversed its position in 2025, and delegated execution to Deka. Five of the German savings banks went live inside their existing banking app. Four custodians carry the assets: Crypto Finance under Deutsche Börse Group, Tangany, Bullish Europe, and Börse Stuttgart Digital Custody, the first German entity to hold a MiCA license. The assets on day one are BTC, ETH, XRP, SOL, and POL. Execution runs on RFQ, request for quote, not an order book. The client submits an order, Deka returns a binding quote inside a limited time window, the client accepts or refuses. Free withdrawals, but only to the four approved providers. That structure deserves to be read carefully, because it is not an exchange. Deka is not a matching engine. Deka is the principal. When the client buys, Deka sells. When the client sells, Deka buys. This is the prime brokerage model, imported wholesale from traditional finance and dressed in a banking app interface. The consequence is arithmetic. A client who sees one total number cannot decompose it into the market reference price and the Deka spread. There is no public order book to compare against, no continuous quote, no price discovery. The client is trading against an opaque counterparty on a timer, and the timer is the counterparty's timer. Based on my audit experience, this is the oldest trick in the distribution playbook, and it works precisely because the fee schedule looks clean. I modeled a €1,000 order. The explicit €0.99 is roughly 0.10%. If the spread sits in the 0.5% to 1.5% band that retail brokers typically run, the all-in cost lands between 0.6% and 1.6%. The spread accounts for 85% to 94% of the total. The marketing anchor is the smallest component. This mirrors the Luna work I did in 2022, when I traced Anchor's yield inflows against outflows for 72 hours and proved the APR was debt, not revenue. The mechanism differs. The discipline does not: follow the cost, not the label. The custody layer compounds it. Client assets sit in an omnibus shared wallet. The client holds a contractual claim on Deka, not a private key, not a discrete on-chain address. Withdrawals are free in the same sense that a hotel corridor is free, you may walk it, but you may not leave the building. No self-custody wallet. No other exchange. No DeFi protocol. No staking, no airdrops, no lending markets. This is a walled garden built to MiCA specification, and it means the capital never touches the chain at all. It stays on a CeFi ledger, denominated in euro, settled in Frankfurt. Here is where the German tax code turns the design from inconvenient into structurally fragile. Under §23 EStG, crypto held longer than one year is tax-exempt. To claim that, the client must prove the holding period. Under omnibus custody, that proof exists only in Deka's records. If the client migrates custodians, or if Deka restructures its systems, the evidentiary chain for a five-figure exemption can break. The client owns an exposure certificate, not an asset. Then there is the staking question, and I want to be precise because the silence is doing work. Solana and Polygon are proof-of-stake networks. Native staking yields typically run 3% to 7% annually. The price sheet mentions no staking service and no yield pass-through. If Deka stakes the underlying inventory and retains the reward, that is an implicit cost on assets the client believes are idle in a vault. On a €10,000 SOL position, 5% is €500 per year. The visible fee on that account would be a rounding error by comparison. I flag this at low-to-medium confidence because no document confirms it. But the absence of a staking line item is not the absence of staking revenue. The asset selection is the strangest signal in the entire package, and it is the one the bulls have not addressed. A consortium famous for conservatism, which voted down crypto products in 2022, launches with BTC, ETH, XRP, SOL, and POL. XRP carries a litigation history. SOL and POL run high volatility, and POL's market cap ranking has slid materially. That list does not match a savings bank client base. It matches the inventory of a liquidity provider. The most parsimonious explanation is that the tradeable universe was set by supplier market-making capacity, not by client demand or prudential review. I have seen this before. In the FTX ledger work, I traced fourteen wallet clusters and found that what appeared to be user activity was internal bookkeeping. Here, what appears to be product strategy may be inventory management. Now the contrarian angle, because the bears are being lazy too. The prevailing critique is that this is a ponzi or a yield trap. It is neither. There is no token, no unlock schedule, no inflation model, no subsidy. Every euro of revenue comes from a client paying a real cost for a real service. That is a fundamentally sound business model, and it is more honest than a farm emitting governance tokens to rent liquidity it cannot keep. The 25,000 euro cap with no minimum order also reveals a two-tier client design: small retail keeps the app sticky, private banking clients generate the actual revenue. There is nothing fraudulent about that. It is just not the democratization story being told. The real structural flaw is not fraud. It is asymmetry of power. Deka may adjust pricing when volume rises or volatility spikes. Read that clause against its use case. The moment a client most wants to trade, or most needs to exit, is the moment volatility is highest. That is precisely when the spread can widen, unilaterally, with no disclosure requirement. I have audited smart contracts for exactly this class of defect, a privileged function callable under adverse conditions. Here the privileged function is human, and the ledger is private. Trust is a variable; proof is a constant. The €0.99 is a constant. The spread is a variable. The custody is a variable. The staking yield is a variable. The client carries all of it, and the only party with full visibility is the counterparty on the other side of every trade. My judgment: this is not a blockchain story. It is a distribution story, and the chain is incidental to it. No new consensus mechanism, no contract deployment, no on-chain innovation. The engineering is legal and custodial. The capital stays off-chain by design, so it should not be counted as a tailwind for any DeFi protocol's TVL. It should be counted as a tailwind for the four custodians, for Deka, and for the Börse Stuttgart group. If the spread stays in the retail band, the product is competitive and the depositors are simply paying a familiar toll for a familiar service. If it widens under stress, the depositors will discover that the bank never sold them crypto. It sold them a claim, priced by the issuer, redeemable inside a building with no exit. The question worth asking next quarter is not how many Sparkassen customers bought bitcoin. It is what the spread was on the day volatility doubled. Until Deka publishes that number, the only verifiable fact remains the one printed on the price sheet: €0.99.