Follow the deposit. Inspect the decision.

Try it in a few minutes

1. Open the app, connect an account and choose Get test funds. Check that the wallet uses the Setsuna demo network, chain 31337.
2. Choose USDC · setsUSDC, deposit 100 test USDC, and inspect your shares and the vault’s actual holdings. MON has a separate vault.
3. Read Why this allocation. Compare recorded rates, the contract’s proposed move and any reason it must wait. A target is not a completed allocation.
4. Open a transaction under Vault activity and public controls. Then redeem your available shares and check the returned underlying asset.

A rebalance has a ten-minute cooldown and a 10% gross-movement limit. You may see a wait or a no-movement decision. Use the recorded receipts below to inspect an earlier successful allocation; there is no need to force a new move during a short demonstration.

What has been verified

On 9 October 2026, the public browser acceptance run completed 25 signed transactions across funding, both Earn withdrawals, Spot and Perps. Balances and receipts survived a cloud restart. The browser used an injected test wallet with actual signatures; physical-device and real wallet-extension checks remain pending.

• MON allocation receipt: 1 MON supplied to Neverland.
• USDC allocation receipt: 14.264141 USDC supplied to Aave and 1.760859 USDC to Morpho.
• Rule-enforcement evidence (JSON): new rebalance calls for both vaults were rejected with CooldownActive in read-only simulations at the recorded blocks. No rejected transaction was broadcast.
• Public acceptance record (JSON): deployment addresses, checks and all 25 signed transaction hashes.

These are functional checks with synthetic assets, not customer deposits, adoption or measured investment performance. Projected annual income is not realized yield. Withdrawals depend on liquidity, and the prototype has not had an independent security review.

Environment

The fork starts from Monad mainnet block 111560409 and does not ingest later mainnet transactions. Spot and Earn use that fork’s protocol state. Perps uses a fixed historical BTC mark of $82,964.20, with a disclosed fork-only oracle configuration change. All trades remain on the fork; mainnet balances are untouched.

Read the allocation rules · Networks and contracts · Current public deployment manifest

Earlier Perpl market research

The following mainnet observations informed the optional margin-protection feature, which is separate from Earn. They come from 706,441 historical events and do not establish demand for Setsuna’s vaults.

What this shows

About one active account in nine added margin to a live position by hand during the observed week. Liquidations had a median of 41 AUSD at stake, and 12 accounts were liquidated more than once. These observations describe the earlier protection use case.

What it does not show

Whether an account is a person or a bot, whether this week was typical, or whether anyone will pay for protection. Positions that were topped up by hand were liquidated more often than others, because traders top up positions already in trouble. That is selection, not evidence that topping up fails or works.

Method

Read-only eth_getLogs against the Perpl exchange at 0x34b6…2a6f, blocks 106,771,593 → 108,773,683, in 100-block windows with every window recovered after retries. Position lifecycles are rebuilt per account and market from events ordered by block, transaction and log index, de-duplicated, with anything unresolvable counted as ambiguous. Account type was read at a stated block: all 64 manual top-up accounts are ordinary wallets, 14 of them with EIP-7702 delegation.

The scripts, raw output and summary live in the repository under scripts/research/ and docs/research/, and reproduce the figures above.