Check a strategy

scenario-bad-tampered-rules

Recorded sandbox run on one node, read twice. Not two independent providers, not a mainnet strategy.

Result

Not matching

• The strategy's status

  Matching

  The strategy is closed and settled.

• Rules match the commitment

  Not matching

  The revealed rules do not match the commitment.

• Action 1: Swap, ok

  Unverifiable

  Can't check: the rules were not revealed.

• Action 2: Open, ok

  Unverifiable

  Can't check: the rules were not revealed.

• Action 3: ClosePosition, ok

  Unverifiable

  Can't check: the rules were not revealed.

• The decision log chain matches the heads on chain

  Matching

  Each action's chain head follows from the one before and its decision.

• No close missed at the sampled blocks

  Unverifiable

  Can't check: the rules were not revealed, or do not match the commitment.

• No change to the account's keys or code at the sampled blocks

  Unverifiable

  No change to the account's keys or code was seen between the first and the last sampled block, but the creation block was not established from chain data, so anything before the first sample is not seen.

• The code hash equals the pinned hash

  Matching

  The deployed code hash equals the pinned hash. These are the sandbox constants from the recording (pins.json), not independently pinned.

• The strategy uses the pinned counterparties

  Matching

  The DCL contract, quote token and pool equal the pinned constants. These are the sandbox constants from the recording (pins.json), not independently pinned.

Not checked in this version

• Valuation: stop loss and take profit use the pool's spot point at each sampled block (Q15). The page cannot check that this price was fair; a price moved across the whole confirmation span would pass.
• That the deployed code is a reproducible build of the audited commit (only a pinned hash is compared, when one is published).
• That the account was created in the single creating transaction, and that principal equals the sum of the funding events.
• That the split of fees followed the stated percentage.
• Refusals by public limits are shown as context. They are not yet recomputed from the record, so they are never counted as an excuse.
• Only the sampled blocks are compared: a key added and removed between two samples, or a close condition that comes and goes between them, is not seen.
• The decision log chain, unless the agent records chain heads (a run without them has no chain to check).
• That the creation block and the agent's history are as recorded; the recorded runs supply a stand-in creation block.