Extended is moving its perpetual futures settlement from Starknet to Arc on October 22. Sub-accounts holding more than $1 in USDT or wBTC must convert or withdraw. If they don't, the platform force-closes them. The announcement contradicts itself on the deadline.
That is the entire public record. Six data points, one of which is a redirect to read the original. No audit disclosure. No contract addresses. No conversion pricing mechanism. No explanation of why a threshold this small triggers an irreversible liquidation. I have spent three months inside order-book matching logic before, and I can tell you what this configuration means: the migration script cannot handle a non-zero residual position, so the platform chose a one-dollar cutoff instead of engineering a proper state transfer.
This is not a technical upgrade. It is a settlement-layer replacement, and the direction runs from mature to new. Starknet has been in mainnet production for roughly two years, settling to Ethereum L1 through STARK validity proofs. Arc is Circle's EVM-compatible L1, which as of 2025 sits in testnet or early mainnet [confidence: medium]. Extended is migrating from a chain with cryptographic finality guarantees into a chain with BFT consensus and a relatively permissioned validator set. The security assumption degrades from mathematics to a known list of entities.
Nothing in the announcement claims faster throughput, lower cost, or better execution. The only visible driver is collateral: USDT and wBTC are being excluded, and settlement moves onto a USDC-denominated chain. Commercial and regulatory logic is doing all the work here. The technical narrative is absent because there is no technical narrative to sell.
The second-order risk is what nobody is pricing. Cairo contracts and Solidity contracts do not port cleanly. Extended's liquidation engine, risk parameters, oracle price feeds, and funding-rate modules must be rebuilt or adapted on Arc. The announcement discloses zero information about audits, open-sourcing, or testnet validation. Based on my audit experience, a cross-VM migration of a derivatives clearing system with no published audit is not a rollout. It is an unhedged bet placed with user capital.
The one-dollar threshold is the most diagnostic detail in the entire announcement. A platform that force-closes positions at one dollar is not expressing a risk-management view. It is expressing an engineering constraint. State migration requires zeroed sub-account balances, and rather than build logic for residual dust, the team elected to liquidate it. When engineering convenience outranks user cost at the margin, that ordering tends to persist through every subsequent operational decision.
The contradictory deadlines in the same announcement compound this. When a document cannot agree with itself on when an irreversible action occurs, the release process has a defect. In a migration where the downside is forced liquidation, that defect is amplified, not absorbed.
Consider what the collateral change actually does. Pre-migration, Extended holds multi-asset collateral: USDT, wBTC, and presumably USDC. Post-migration, the inferred state is USDC only. Collateral concentration moves from distributed to single-point. More importantly, the platform's settlement risk, its collateral issuer risk, and its gas-denomination risk all collapse into one entity: Circle. Circle holds freeze authority over USDC. Arc and USDC sit inside the same corporate perimeter. The risk isolation that multi-asset collateral provided is gone.
This is the part the bulls are not wrong about, and I will give them the credit. Institutional capital does not settle on chains it cannot audit against a regulatory framework. USDC is MiCA-compliant. USDT has been delisted across multiple EU venues. If Extended's actual objective is onboarding institutional flow, then migrating to Arc is not a retreat from decentralization. It is a deliberate trade of a ZK validity-proof narrative for a regulated settlement rail with a fiat on-ramp. That is a coherent business decision.
But coherence is not the same as execution quality. A platform that wants to be read as compliance-ready must meet compliance-grade disclosure standards. It must publish the conversion pricing mechanism. It must publish the new contract addresses. It must publish the audit. It must publish the custody arrangement for funds in transit. None of that exists. An announcement that targets institutional credibility while withholding every disclosure an institution would require is not a compliance strategy. It is compliance theater.
The conversion itself deserves scrutiny. When a platform executes an asset conversion at its own internal quote, the spread between the reference price and the executed price is an undisclosed cost borne by the user. There is no mechanism in the public record for a user to verify whether the USDT-to-USDC or wBTC-to-USDC rate they received was fair. That is a value transfer operating under information asymmetry, and it is happening during the window when the user has the least leverage to refuse.
The user base makes this worse. The existence of a sub-account system with granularity down to one dollar indicates Extended serves API traders, quant desks, and market makers. These are the users most sensitive to migration friction. They run automated strategies, maintain multi-account structures, and hold specific collateral baskets by design. Their migration costs are not emotional. They are operational, and they are measurable. When a platform imposes forced liquidation on this cohort, the attrition is not a sentiment dip. It is a permanent reallocation of order-book depth to competitors.
Starknet should read this as a negative signal about application-layer retention. A platform that chose to build and operate there, and then chose to leave, carries more evidentiary weight than a platform that never deployed. It suggests the chain has structural shortcomings in stablecoin settlement, regulatory friendliness, and institutional client acquisition. Paradex remains, and the ZK-as-a-service positioning may absorb the rest, but the direction of travel is legible.
For Arc, the calculus is different. Derivatives generate the densest fee and liquidation demand of any on-chain sector. A real perpetuals venue migrating in is worth more as validation than any TVL incentive program, because it produces organic gas consumption denominated in the chain's own settlement asset. Every settlement converts into USDC demand, which is Circle extending from issuer to settlement-layer rent collector. If the migration completes cleanly, expect copycats. If it stalls or produces user losses, expect the opposite.
The migration window itself is the most fragile period. Market makers must redeploy on Arc, re-authorize allowances, rebuild inventory, and re-establish hedging infrastructure. Order-book depth thins. Slippage widens. Execution costs rise for everyone who stays. A portion of that flow will not return, because the cost of returning exceeds the cost of trying a competitor.
Watch three things. First, whether Extended publishes the conversion pricing mechanism before October 22, or whether users are expected to accept an undisclosed rate. Second, whether the migration produces a liquidation spike that the platform earns fees on, because a venue profiting from forced closures it engineered is an incentive misalignment worth naming. Third, whether trading volume returns to pre-migration levels within thirty days, or whether the platform responds with maker rebates that compress the entire sector's fee structure.
Every forced exit leaves a footprint. Trace the liquidation events on both chains after the cutoff. The addresses that never come back are the honest measure of what this migration actually cost, and no announcement will report them.